Economic Comparison Between Conventional and Disposables-Based Technology for the Production of Biopharmaceuticals J. L. Novais, N. J. Titchener-Hooker, M. Hoare The Advanced Centre for Biochemical Engineering, Department of Biochemical Engineering, University College London, Torrington Place, London WC1E 7JE, United Kingdom; telephone: +44 20 7679 3796; fax: +44 20 7383 2348; e-mail: nigelth@ucl.ac.uk Received 27 December 2000; accepted 30 April 2001 Abstract: Time to market, cost effectiveness, and flexibility are key issues in today’s biopharmaceutical market. Bioprocessing plants based on fully disposable, presterilized, and prevalidated components appear as an attractive alternative to conventional stainless steel plants, potentially allowing for shorter implementation times, smaller initial investments, and increased flexibility. To evaluate the economic case of such an alternative it was necessary to develop an appropriate costing model which allows an economic comparison between conventional and disposables-based engineering to be made. The production of an antibody fragment from an E. coli fermentation was used to provide a case study for both routes. The conventional bioprocessing option was costed through available models, which were then modified to account for the intrinsic differences observed in a disposables-based option. The outcome of the analysis indicates that the capital investment required for a disposables-based option is substantially reduced at less than 60% of that for a conventional option. The disposables-based running costs were evaluated as being 70% higher than those of the conventional equivalent. Despite this higher value, the net present value (NPV) of the disposables-based plant is positive and within 25% of that for the conventional plant. Sensitivity analysis performed on key variables indicated the robustness of the economic analysis presented. In particular a 9-month reduction in time to market arising from the adoption of a disposables-based approach, results in a NPV which is identical to that of the conventional option. Finally, the effect of any possible loss in yield resulting from the use of disposables was also examined. This had only a limited impact on the NPV: for example, a 50% lower yield in the disposable chromatography step results in a 10% reduction of the disposable NPV. The results provide the necessary framework for the economic comparison of disposables and conventional bioprocessing technologies. © 2001 John Wiley & Sons, Inc. Biotechnol Bioeng 75: 143–153, 2001. Keywords: disposable equipment; bioprocessing; process economics; time to market Correspondence to: N. J. Titchener-Hooker Contract grant sponsors: Biotechnology and Biological Sciences Research Council; Millipore; Kvaerner Process; Lonza Biologics; Hyclone; Biotage; Portuguese Ministry of Science and Technology © 2001 John Wiley & Sons, Inc. INTRODUCTION To be successful in today’s biopharmaceutical market it is no longer enough to have a good technology portfolio, a strong intellectual property position, and access to capital (Gamerman and Mackler, 1994). Flexibility, cost effectiveness, and time to market are becoming the key issues with which biopharmaceutical companies have to be concerned (Basu et al., 1998; Burnett et al., 1991; Ernst et al., 1997; Gamerman and Mackler, 1994; Hamers, 1993). First, it is essential that biopharmaceutical development and manufacturing facilities have increased flexibility. This is dictated both by constant priority changes arising from the generation of safety and clinical data in the development phase (Basu et al., 1998) and by the need to allow for future expansions. As a consequence of priority changes, the capability of multiproduct processing is gaining an increasing interest although there are still complicated regulatory issues associated with potential crosscontamination (Hamers, 1993). The need to allow for future expansion requires careful decisions during the design of any facility. The problem is that when such decisions are made early in the development process they are difficult to change later due to regulatory constraints (Basu et al., 1998; Ernst et al, 1997). On the other hand, the delay of the decision to build brings construction onto the critical path (Nicholson, 1998). Early decision-making would be advantageous but will be associated with higher risk because there is less confidence in the likelihood of success of the product (Burnett et al., 1991). This is particularly critical for biopharmaceutical products due to their high failure rates (Struck, 1994). Second, the industry must operate under growing government- and market-enforced price controls and a need for cost-benefit justification (Gamerman and Mackler, 1994) which brings a demand for better cost-effectiveness. Additionally, there is less confidence on the part of the investors and a lack of available capital which forces companies to control their capital needs. Finally, it is essential to get into the market as quickly as possible due to increasing competitiveness (Burnett et al., 1991) and to maximize revenue during patent life. To cut healthcare budgets generic drugs are being favored and newly released drugs no longer command large premiums (Nicholson, 1998). Companies must therefore move more quickly from discovery to patent, and then to trials and efficient production. According to Basu et al. (1998) delays in entering the market translate into millions of dollars of lost revenue. In this article we explore the use of biopharmaceutical plants based on disposable components as a means to overcome some of these pressures. In such a re-engineering stainless steel vessels would be replaced by presterilized disposable plastic containers requiring virtually no maintenance and incurring minimal cleaning costs. The same concept applies to the connections which can be replaced by disposable plastic pipes. This option therefore has the advantage of switching capital costs to consumables costs as required. It also allows for the better management of uncertainty in future process volumes. The disposable concept can be extended throughout the production process. Separation processes such as cell harvesting and protein clarification and concentration can be achieved by tangential flow filtration with disposable membranes. The final purification steps can be accomplished in disposable prepacked chromatography columns or by batch adsorption in disposable plastic bags. Disposable technology also makes use of noninvasive pumps and valves, such as peristaltic pumps and pinch valves. All the instrumentation is disposable or noninvasive and heat transfer can be achieved by disposable heat-exchangers (Pearl and Christy, 2000). The competitiveness of such plants is increased due to minimization of the resultant cost of down-time for cleaning, sterilization, and corresponding validation procedures as well as labor and operating costs associated with these operations. The use of disposable equipment also allows for quick changeover between products which is invaluable in the clinical phases of development where often multiple products are evaluated simultaneously. A key factor determining the speed to market of disposables-based plants is associated with the decision of when to build the manufacturing facility. The simpler construction of disposables-based plants implies that shorter implementation times can be realized which allows for more detailed process optimization before moving onto construction. Alternatively, these shorter construction times may allow for earlier entry to market and at a lower risk due to the smaller investment involved. To determine the importance and interest of such a technological approach it is necessary to evaluate disposablesbased processes from an economic point of view and to compare it with traditional bioprocessing methods. This is a difficult task due to a lack of adequate costing models for biopharmaceutical facilities. The few models currently available have been originally developed for conventional process engineering facilities (Peters and Timmerhaus, 1991; Sinnott, 1991). Some published case studies for bioprocesses do exist and can be used as a basis for costing (Datar et al., 1993; Ernst et al., 1997). In addition, it is 144 important to consider process development issues such as scale-up of the disposable option or transfer and scale-up to a conventional option. Though this is not the subject of this article it is recognized that in both cases it will be necessary to carry out process studies using the disposable option in such a way that product equivalence will be achieved. In the present work therefore the main concern was to establish a comparison between the two approaches rather than to obtain absolute costing figures. The comparison was made on the basis of the net present value (NPV), which requires the calculation of the capital investment and of the annual running costs. This article proposes a methodology to cost disposables-based plants followed by an application of this methodology to a particular case study. METHODOLOGY The initial methodology developed for the economic evaluation and comparison of the conventional and disposable approaches to bioprocessing is intended to be as generic as possible. This methodology was then applied to an E. coli process, which is presented in the Results section. For ease of comparison between process costs based on conventional and disposable equipment, it is initially assumed that the two operate at the same yield throughout, e.g., same biomass and product concentration in the fermentor or same flux rate in a microfilter. Comparable yields are not likely to be achieved for a high cell density fermentation but are probably achievable for mammalian cell culture; flux rates are likely to be independent of the operating mode. The costing is based on that for conventional equipment where the majority of information is available. Conversion factors are then used to assign the cost based on disposable equipment. Any difference in yield between conventional and disposable options requires an understanding of the total process. Factors affecting the yield in a fermentor will have a different overall effect on the costing compared with factors affecting the yield in the recovery and chromatographic separation stages. For example, in the case of an intracellular product a reduction in fermentor cell density will not affect the nutrient costs but will affect the fermentor size and the load on the critical cell harvest step. The remainder of the process should not be affected. The objective of the following costing analysis is to allow a comparison between conventional and disposable options as indicated in Figure 1A and 1B respectively. The following sections will determine the routes towards capital and operating costs and an overall analysis based on the Net Present Value (NPV). A full description of the proposed framework is given in Figure 1 based on the costing analysis given in the following sections. Many different scenarios may be possible for the comparison of the costs. In this article it is assumed that: ● A green-field site is used. BIOTECHNOLOGY AND BIOENGINEERING, VOL. 75, NO. 2, OCTOBER 20, 2001 used (Peters and Timmerhaus, 1991). The specific value for such a factor applicable to bioprocessing plants is not easily available from the literature but can be obtained from the sum of the individual factors which constitute the fixed capital investment. The fixed capital investment for a conventional bioprocessing plant (FCIconv) is therefore given by: S( D 10 FCIconv = LconvEconv = c fi Econv (1) i=1 Figure 1. Diagram of the methodology followed for the calculation of the net present value (NPV) of (A) conventional, and (B) disposables-based routes of bioprocessing. (A) The costing route is determined by a design analysis of the bioprocess leading to a sizing of the equipment and specification of process materials’ requirements. This leads to an evaluation of the equipment and hence, fixed-capital investment and operating costs, which along with the schedule of project development yield the NPV for the project. (B) The costing route for the disposable option uses the conventional equivalent initially assuming operation at the same process yields as the conventional option. The fixed capital investment is determined by a series of factors relating the conventional and disposable options [Eq. (2)]. The operating costs are determined as for the conventional option with appropriate factors that reflect the increased costs due to disposable items. The effect of a difference of yield between the conventional and disposable options is accomodated by a resizing of the conventional equivalent followed by a re-evaluation of the fixed capital investment. For operating costs the effect of a change of yield requires that the conversion factors associated with the running costs are re-evaluated to take into account the change in materials’ costs. ● The plant is used throughout the year as a single product facility. ● In the disposable option all equipment parts that are in direct contact with the process streams will be disposable and, where necessary, presterilized (e.g., with g-radiation). ● In the disposable option all media and buffers are purchased preformulated and sterile ready for use. The time to build, commission, and validate each affect time to market and are therefore crucial factors in the comparison between the conventional and disposable options. For the NPV calculations an identical project schedule for the two options was considered appropriate (Fig. 1). The additional competitive advantage of being quicker to market that may be brought by a disposables-based approach will be discussed later in the article. Capital Investment In an approach initially proposed by Lang (1948) for chemical engineering plants, the fixed capital investment (FCI) can be calculated by multiplying the equipment cost by a factor, which depends on the type of process plant being where Lconv is a “Lang” factor for conventional bioprocessing plants and Econv is the cost of the process and utilities equipment. The factors f1 to f10 relate to Econv to give the cost of process and utilities equipment (f1, f1 4 1), pipework and installation ( f2), process control ( f3), instrumentation ( f4), electrical power ( f5), building ( f6), detail engineering ( f7), construction and site management ( f8), commissioning ( f9), and validation ( f10). A contingency factor, c, may also be included. The fixed capital investment for a bioprocessing plant based on disposable equipment (FCIdisp) may be estimated from the cost of the installed equipment and utilities for a conventional plant as follows: S( D 10 FCIdisp = LdispEconv = c8 fifi8 Econv (2) i=1 where Ldisp is a “Lang” factor for disposable bioprocessing plants, f 18 to f 810 are factors which translate the cost of the individual elements which constitute the capital investment of the conventional plant into the cost of elements for the disposable option. c8 is the relevant contingency factor. The factors f 81 to f 10 8 for the conversion from conventional to disposable can be estimated from the following assumptions: ● Equipment and utilities ( f 18 ): In a plant using disposable process equipment the capital investment costs for process equipment are decreased. Also, the disposable option would need reduced or even no clean-in-place and steam-in-place capabilities. As a consequence, the cost of utilities equipment is substantially reduced as only features such as cooling water, chilled water, process air, and vacuum will be required if all of the process can be turned over to disposable operation. ● Pipework and installation ( f 82): The capital costs associated with pipework are decreased in a disposable option because the cost of the disposable tubing becomes an operating cost. Connections to utilities which do not come into contact with the product stream would normally be considered as nondisposable. Equipment installation costs are also decreased due to the reduction in fixed equipment. ● Process control ( f 38 ):Process control costs are likely to remain unchanged although in the disposable case there may well be a move toward more manual operation in the interest of speed to market. Conversely, the need for NOVAIS ET AL.: ECONOMIC EVALUATION OF DISPOSABLES-BASED BIOPROCESSING 145 ● ● ● ● ● ● ● more noninvasive monitoring may lead to greater costs in computing for data interpretation for control purposes. Instrumentation ( f 48 ): Instrumentation capital costs are reduced because some of the instruments may be disposable (for example, thermocouples) and therefore appear as a running cost. Alternatively, instrumentation may be redesigned to be noninvasive (e.g., UV-detectors) and hence, lead to no change in capital cost. Other instrumentation such as gas mass spectrometers are not in contact with the process material and also do not lead to a change in capital cost, as they will be needed for both modes of operation. Where disposable alternatives to high-cost invasive instrumentation (e.g., pH meters) are not available then either separate validation for turn around (e.g., cleaning and recalibration) must be put in place or recourse is needed to data interpretation from actual available measurements (e.g., cell density by optical window, exit gas analysis, etc). Again, in such a case it is assumed the capital cost is not affected. Electrical power ( f 85): Assuming power consumption and capital costs are related, it is likely that electrical power capital costs are probably independent of whether conventional or disposable equipment is used. Alternative methods of mixing for a disposable process are likely to have similar power requirements to those for a conventional process. Conversely, a reduction in size of facility could lead to a significant decrease in air conditioning costs because this will be related to the volume of the facility. Building ( f 86): The variation of the cost of the building can be estimated by establishing the layout of a typical coventional plant and predicting the changes of the floor area of each section based on the particular needs of a disposable plant. If a system with wheel-in/wheel-out equipment is considered, it can be assumed that processing areas can be reduced in size but that larger, low-cost storage areas will be needed. Also, utilities’ areas will be smaller and media preparation and equipment wash areas might disappear. The effect of changes in the function of the areas and consideration of their differential cost should lead to a reduction in building costs when using a process based on disposable equipment. Detail engineering ( f 78 ): The costs associated with detail engineering are expected to be reduced for the disposables option due to the less-refined construction needed. Construction and site management ( f 88 ): Construction and site management costs should be decreased due to the smaller building area required for the disposable option. Commissioning ( f 89): The commissioning costs of the disposables-based plant are considered to remain unaltered when compared to those of a conventional plant. Validation ( f 810): The validation of a disposables-based vs. a conventional process will differ due to process qualification (PQ): —Reduced or no cost of validation for the cleaning, sterilization, and turnaround of process equipment when using disposable equipment. This argument is 146 already used in the qualification of use of disposable containers. The challenge and hence costs of cleaning validation for more complex equipment such as membranes also bears on this factor. —The cost of validation of linkages between equipment (sterile welding vs. conventional sealed pipe joints) is likely to remain the same. With estimates for the conventional plant equipment costs (Econv) and for the factors fi and f 8i it is possible to calculate Lconv, Ldisp, FCIconv, and FCIdisp. We use estimates of factors f 81 to f 810 to demonstrate how the translation from a conventional to a disposable option may be evaluated. Running Costs A model based on a bacterial fermentation process was derived from the breakdown of the running costs observed by Datar et al. (1993) for their particular case study. The breakdown was reduced down to five categories (labor, materials, utilities, depreciation, and other costs) and adapted so as to exclude general expenses such as R&D and sales expenses from the overall running costs for simplification purposes. This results in: 5 RCconv = RCconv (x i (3) i=1 where RCconv is the running cost of the conventional plant, x1 to x5 are the fractions of the running cost which give the cost of its individual components: labor (x1), materials (x2), utilities (x3), depreciation (x4), and other costs (x5). Other costs include patents and royalties, waste treatment, and indirect manufacturing expenses. The costs of the disposable option can be predicted from considerations on how each category of costs varies when compared to the equivalent conventional option. It is expected that materials (raw materials including disposable equipment) will increase significantly and that utilities costs and depreciation costs will be reduced, the latter due to the lower capital investment involved. Hence, Eq. (3) becomes: 5 RCdisp = RCconv (xy i i (4) i=1 where RCdisp is the running cost of the conventional plant and y1 to y5 are factors which convert the individual conventional running cost fractions into disposables-based ones. The factors y1 to y5 for the conversion from conventional to disposable can be estimated from the following assumptions: ● Labor (y1): Costs associated with cleaning and sterilization will be decreased but there will be increased costs due to the need for assembling/disassembling of components, as well as the operation of sterile welding systems. BIOTECHNOLOGY AND BIOENGINEERING, VOL. 75, NO. 2, OCTOBER 20, 2001 ● ● ● ● Labor costs associated with in-house media and buffer preparation will be decreased. Materials (y2): Costs associated with raw materials will be increased, as these will be bought as preformulated media and buffers. These cost items are higher so as to include the expense of the containers and the operating costs incurred by the supplier for the preparation and sterilization of the media and buffers. The cost of disposable items (e.g., membranes, vessels, chromatographic media, pipework, etc) will become a major factor. Utilities (y3): Costs associated with steam and cleaning requirements should be reduced. Depreciation (y4): This cost should be reduced as it is only associated with the process plant capital investment, which is reduced in a disposables’ option. Other (patents, royalties, waste treatment, etc) (y5): This cost is possibly unaffected—with, for example, the high effluent treatment costs for cleaning agents associated with the conventional option being offset by the increased costs for solid-waste treatment of the disposable option. Net Present Value The net present value (or net present worth) gives an indication of the profitability of a project. The generic equation for its calculation is (adapted from Peters and Timmerhaus, 1991): m+t NPV = ( n=0 CFn − FCIn ~1 + r!n Sn − RCn − FCIn m+t = ( ~1 + r!n n=0 m−1 ( n=0 −RCn − FCIn ~1 + r! n m+t + ( n=m Sn − RCn − FCIn ~1 + r!n Effect of Process Performance Although the disposables-based process is intended to be identical in performance and characteristics to its conventional counterpart, it may well be that some unit operations have to be altered to be operated in a fully disposable fashion. Such changes may affect negatively the yield of these particular steps and consequently either more time has to be spent in process development or the fermentation volumes may have to be increased so that the final product mass is the same as in the conventional alternative. This will also impact the design specifications of subsequent operations. For example, the membrane area needed for cell harvesting has to be increased to cope with a higher fermentation volume. The ultimate effect of such differences is a lowered NPV for the disposable plant, which has to be quantified and compared to that of the conventional plant. This is done according to the procedure detailed in Figure 1B. As discussed earlier, the evaluation of the impact of a change of yield on the overall process is a complex one with each stage having a different impact on the overall process according to product location and subsequent recovery and purification stages. Hence, the impact of the performance of different stages has to be assessed specifically for that stage. RESULTS (5) Case Study where NPV is the net present value of the project, r is the discount rate (or annual interest rate of return), CFn is the net cash flow in year n, FCIn is the fixed capital investment in year n, t is the life of the project (in years), m is the year of entry to market, Sn is the value of sales in year n and RCn is the value of the running costs in year n. For n < m, i.e., before entry to market, Sn will be zero and Eq. (5) becomes: NPV = linear least squares and interpolation can be used to deal with fractions of years gained in time to market. (6) with the first part of the equation dealing with the period before manufacture commences and the second part with the period thereafter. Effect of Time to Market Considering that opting for a disposables-based process may allow for earlier entry to market it is interesting to evaluate the impact this may have on the NPV. This is achieved using Eq. (6) but considering that the start of manufacture and sales occurs earlier than year m. (This does assume that the life-span of the project (t) is unchanged, which will result in an underestimate of the benefits.) Non- The case study presents a comparison between a conventional bioprocessing plant and its disposable equivalent and is based upon the production of an antibody fragment using a recombinant E. coli at a 300-L working-volume fermentation scale of operation with the antibody expressed in the periplasmic space. Given that bioprocessing bags are currently available up to 1000-L scale (HyClone Europe catalogue), operation at 300 L should not pose any additional major obstacle. The cells are to be harvested, lysed for protein release, followed by removal of cell debris/empty cells and a first chromatographic or adsorption/desorption separation stage either for product capture (preferably) or for contaminant capture. A detailed process diagram and a complete mass balance were developed and used as the basis for equipment sizing. The process scheme is shown in Figure 2A using conventional equipment and in Figure 2B using equipment configured for use in a disposable fashion. The resulting equipment list was used to establish the capital investment for each option [Eqs. (1) and (2)]. From the mass balance it was also possible to estimate raw materials and disposable equipment consumption which were used in the calculation of the running costs. It was assumed that the disposables-based plant had an identical process sequence and the same yield per unit op- NOVAIS ET AL.: ECONOMIC EVALUATION OF DISPOSABLES-BASED BIOPROCESSING 147 Figure 2. Process diagram of the case study process: E. coli production of an antibody fragment. (A) conventional route, and (B) disposables-based route. In the latter case, the process vessels are disposable bioprocessing containers and the fermentation is achieved with a plunging-jet design. eration as the conventional plant. Both conventional and disposables-based options were assumed to operate 48 batches/year for a project life of 8 years. Hence, no account was taken of the potential for some rapid turnaround of a process batch when using the disposables option. It was assumed that the decision to build the conventional manufacturing plant has to be taken by Phase III clinical trials, that is approximately 3 years before entry to market. A similar constraint was applied for the disposables option. This assumption may result in an underestimate of the benefits of disposables option which, through having a reduced capital expenditure, may enable such financially risky decisions (failure rates being as high as 1 in 3 at early clinical stages; Struck, 1994) to be made somewhat earlier. Capital Investment Based on bioprocessing plant data (A. Sinclair, Biopharm Services, Amersham, Bucks, UK; D. Doyle, Kvaerner Pro148 cess, Portsmouth, UK, personal communication) it is possible to assign values to the fixed capital investment factors ( f1 to f10) in Eq. (1). It is also possible to estimate values for the factors in Eq. (2) that translate conventional into disposables ( f81 to f 810) from the assumptions described in the methodology section. The factor f81 was calculated considering that utilities account for approximately half of the total equipment costs and that they are reduced by a factor of 60% because there is no need for clean steam production, etc. (This factor also assumes that all the process equipment is disposable.) The values for the conventional and disposable factors are summarized in Table I and enable the calculation of the two “Lang” factors. The “Lang” factor for the conventional plant, Lconv, evaluated at 8.1 is close to the range commonly observed for biopharmaceutical plants (6 to 8, D.Doyle, Kvaerner Process, Portsmouth, UK, personal communication). A “Lang” factor is obtained for the disposable plant, Ldisp 4 4.7 based on the equipment cost for a conventional option. BIOTECHNOLOGY AND BIOENGINEERING, VOL. 75, NO. 2, OCTOBER 20, 2001 Table I. Capital investment factors for conventional ( fi) and disposablesbased ( fi f 8i ) bioprocessing plants and corresponding “Lang” factors. Description 1 2 3 4 5 6 7 8 9 10 Equipment (incl. utilities) Pipework and installation Process control Instrumentation Electrical power Building works Detail Engineering Construction and site management Commissioning Validation Contingency factor (c) “Lang” Factor Conventional fi Disposable f8i 1 0.2 0.9 0.33 0.37 0.6 0.24 1.66 0.77 0.4 1 0.66 1 0.8 0.5 0.75 0.07 1.06 1 0.5 1.15 1.15 Lconv 4 8.13 Ldisp 4 4.73 Note: The conventional plant factors ( fi) are derived from standard figures for conventional bioprocessing plants (A. Sinclar, Biopharm Services, Amersham, Bucks, UK; D. Doyle, Kvaerner Process, Portsmouth, UK, personal communication). The factors for the disposable translation ( f i8) are based on assumptions derived from the definition of disposable manufacture (presented in the Methodology). The Lang factors are defined 10 10 as: Lconv = c(i=1 fi and Ldisp = c(i=1 fi f 8i . The values of capital investment for both options were obtained from Eqs. (1) and (2) with the conventional equipment costs based on the case study process and from the values of Lconv and Ldisp shown in Table I. Hence, the conventional capital investment was evaluated at $19.2M as opposed to $11.2M for the disposable option. Running Costs Table II shows the materials costs calculated for both options based on the process requirements. The cost of the raw materials in the conventional plant was estimated from laboratory supplies catalogs (i.e., Sigma Biochemicals and Reagents for Life Science Research, 1998; BDH Laboratory Supplies Catalogue, 1998) costs for the largest available quantities to which a discount factor of 3 was applied. For Table II. Annual materials’ costs estimates for both conventional and disposables-based routes. Materials Costs ($k/year) Conventional Disposable Raw materials Membranes Matrices (IEX) Other disposable equipment Total 36 58 16 0 110 138 1166 319 191 1814 Note: The cost of each item was obtained from a mass balance to the process and from the process flowsheet. The item “Other disposable equipment” includes bioprocessing containers and flexible pipes. It can be noted that the total cost of materials requirements of a disposables-based plant is approximately 16-fold higher than that of an equivalent conventional bioprocessing plant. the disposable plant it was assumed that the media and the buffers are bought preprepared supplied in sterile containers and cost on average $4/L (range of costs suggested by HyClone Europe). The running costs of the membranes and matrices in the conventional plant were calculated from membranes and chromatography manufacturers catalogs [Millipore (UK) Ltd. and Amersham Pharmacia Biotech] taking a conservative estimate that these are used 20 times before being replaced. In a disposable plant both membranes and ion-exchange matrices are disposed of after each batch. The item “other disposable equipment” includes all other equipment not specified above, such as bioprocessing containers and flexible pipes (HyClone Europe Price List). This item is close to zero in a conventional plant. It was assumed that a disposable plant makes use of containers of the same volume as the stainless steel containers in the conventional plant and that it needs approximately 10 m of flexible tubing for each unit operation. The factors x1 to x5 [Eq. (3), Table III] were calculated from the cost breakdown presented by Datar et al. (1993). The cost of depreciation is estimated using the capital investment (FCIconv) and a working life of 8 years for the plant. From there it is possible to calculate the cost of the other individual items of the running costs of the conventional plant through Eq. (3). The running costs of the disposables-based option were estimated through Eq. (4). As a first approach it was assumed that the disposables-based plant has the same staff requirements as its conventional equivalent (y1 4 1). From Table II it can be noted that there is a 16-fold increase in the running costs associated with all materials and consumables in a disposables-based approach, hence y24 16. In a disposable plant the utilities running cost is expected to be halved due to the absence of operations such as clean-in-place (CIP) and steam-in-place (SIP) (y3 4 0.5). Depreciation costs are reduced as a result of the lower capital investment involved as shown in Table I, that is y4 4 0.6 and other costs are likely to remain unchanged (y5 4 1). The resultant comparison for the running costs for conventional as opposed to disposable operation is given in Figure 3, the running costs of a disposable biopharmaceuTable III. Running costs factors derived from a cost distribution presented by Datar et al. (1993) for bacterial process. 1 2 3 4 5 Item xi yi Labor costs Materials Utilities Depreciation Other 0.14 0.06 0.14 0.19 0.47 1 16 0.5 0.6 1 Note: The conventional factors (xi) were obtained from the percentages presented by Datar et al. (1993) for their case study after excluding general expenses from the overall running costs. the item “Other” includes costs such as patents and royalties, waste, indirect manufacturing expenses, etc. The disposable factors (yi) were predicted considering that staff costs remain the same, raw materials including disposable equipment increase 16-fold (based on the calculations in Table II), utilities costs are reduced by half, and other costs remain unchanged. Depreciation costs are reduced as a direct result of the lower capital investment involved. NOVAIS ET AL.: ECONOMIC EVALUATION OF DISPOSABLES-BASED BIOPROCESSING 149 Table IV. Economic analysis results summary. Option Convention Disposable FCI ($M) RC ($M/year) NPV ($M) 19.2 12.8 102.6 11.2 22.0 75.8 Note: The fixed capital investment was obtained from the total cost of conventional equipment and from the “Lang” factors in Table I. The running costs are as presented in Figure 3 and the fixed capital investment (FCI). The net present value (NPV) was calculated according to Eq. (7) assuming annual sales of the Fab8 antibody fragment to be 5 times the conventional running costs (Coopers and Lybrand, 1997. Pharmaceuticals: Creating value by transforming the cost base, company publication). Figure 3. Breakdown of the running costs of the conventional and disposable processes. The conventional breakdown was based on a cost distribution presented by Datar et al. (1993) for a bacterial process and the disposable individual running costs were calculated with the use of the factors presented in Table III and from the conventional capital investment value presented in Table IV. The item “Other” includes costs such as patents and royalties, waste, indirect manufacturing expenses, etc. tical plant being 1.7 times higher than the equivalent conventional costs. A second model originally developed for conventional engineering processes (Sinnott, 1991) was also used for the calculation of the running costs (analysis not presented). According to this model the overall running costs associated with a disposables option would differ by a factor of 0.9 to those of a conventional option compared with a factor of 1.7 given above. This is because the second model places more emphasis on costs such as depreciation and utilities, which are reduced in a disposables approach, rather than on raw materials and consumables. Net Present Value To compare the disposables-based approach with the conventional one the respective NPV was calculated according to Eq. (6). In this equation it was considered that the investment was completed in year zero for both approaches. Both plants were considered to start operating at half capacity in year 1 (RC1 4 1/2 RC), achieving full capacity in year 2. Sales commence in year 3 (m 4 3, S1 4 S2 4 0). Taking the project life span as 8 years (t 4 8) as specified in the description of the case study and the discount rate as r 4 0.2, Eq. (6) becomes: 2 NPV = FCI + ( n=1 − RCn + ~1 + 0.2!n 10 Sn − RCn ( ~1 + 0.2! n=3 n To evaluate the reliability of the disposable cost models, sensitivity analysis was carried out for relevant variables in the disposables approach. The parameter chosen for the comparison was the ratio of disposable NPV over conventional NPV, which is >1 when the disposables-based option is financially the most attractive. The variables studied included capital investment (and consequently also the “Lang” factor for disposable bioprocessing plants, Ldisp), staff costs and materials costs with the aim of evaluating whether the assumptions made for the definition of the disposable cost models affect the final comparison. Figure 4 shows the impact on the NPV ratio of a variation of −25% made to each of these disposable plant variables. (Note when the capital investment is varied, the depreciation costs are also affected as they are, by definition, directly related to the capital investment.) It can be seen that these three variables have only a slight impact on the NPV ratio. A further effect that may influence the way the two processing options compare is the resulting time to market achievable. Figure 4 also shows how a reduction of 9 months in entry to market with the disposables-based approach can affect significantly the NPV ratio. For example, (7) The net present value was obtained having set the annual sales for each case as 5 times the running costs of the conventional plant (Coopers and Lybrand, 1997. Pharmaceuticals: Creating value by transforming the cost base, company publication). The results are summarized in Table IV. 150 Sensitivity Analysis Figure 4. Sensitivity analysis showing the effect on the NPV of a 25% reduction on crucial cost variables. The results are presented as a ratio (NPVdisp/NPVconv) and were obtained by altering the following variables in the disposable case, leaving the conventional case fixed: fixed capital investment, staff costs, materials costs and time to market. BIOTECHNOLOGY AND BIOENGINEERING, VOL. 75, NO. 2, OCTOBER 20, 2001 the NPV ratio is 0.74 when the time to market is the same but increases to 1.00 if disposables allows for a 9 months earlier entry to market. Another important aspect that cannot be overlooked is whether a disposables-based approach may affect the overall yield of the process. The unit operations that may be affected by a switch to disposables operation are the fermentation, where a stirred tank has to be replaced by a plunging jet (Zaidi et al., 1991) or an airlift reactor, and the chromatography step, where the column may be replaced by a column prepacked with cheaper media or by multiple batch adsorption/desorption steps. The microfiltration, and periplasmic release steps are not expected to be affected as they remain intrinsically the same as in the conventional process. The performance of a disposable fermentor may be lower when compared to a conventional fermentor as a result of different factors such as oxygen-transfer difficulties. Work carried out in this research group (Baker et al., to be published) with a disposable fermentation plunging-jet design indicated that oxygen transfer might be a factor affecting the yield. The yield loss may be due to achieving a lower level of biomass or through the cells being intrinsically less productive (lower expression levels). Additionally, in the particular case of a periplasmically expressed product some material may be released into the fermentation broth during the operation of a disposable fermentation as is observed in stirred-tank fermentors (Gill et al., 1998). The effect on the disposable plant NPV of a reduction of 25% in the yield of antibody fragment was studied. It was considered that this reduction in yield could be effected in two different ways: (a) as a 25% reduction in the biomass, and (b) as a 25% fall in the expression level of the cells, but with an overall identical biomass concentration (Fig. 5). Alternative “b” also considers the case where there is a reduction of the amount of product found in the periplasmic space with an equivalent increase in that found in the extracellular me- dium. The NPV of each alternative was calculated as described in the methodology section considering that the final product mass obtained was the same as in the base case. A 25% yield loss in terms of biomass results in a slight drop in the achievable NPV to 91% of that of the base case, while a 25% lower expression level has a higher impact on the NPV, decreasing it to 83% of that of the base case (Fig. 5). Each of these effects may also be examined more closely. Sensitivity analysis was carried out for a range of fermentation yields of 50% to 100% of the conventional base case, combined with sensitivity analysis for the cost of the materials (raw materials and disposable equipment) and is shown in Figure 6A and 6B. Figure 6A illustrates the case where the yield loss is associated with less biomass while Figure 6B analyzes the consequences of yield loss due to less product being produced by the cells. A 50% drop in biomass yield results in a 30% drop in the NPV. This is more accentuated in the case where the yield loss arises from a lower expression level, resulting in a 60% decrease of the NPV. The cost of the materials was examined as it is Figure 5. Impact of lower fermentation performance effects on the NPV of the disposable option. It was considered that the reduction in yield could be effected in two different ways: as a 25% reduction in the biomass (lower biomass) or as a 25% fall in the expression level of the cells, but with an overall same biomass concentration (lower expression). The results are presented as % of the NPV of the base case, that is a disposables-based plant with same yield as a conventional plant. Figure 6. Combined sensitivity analysis on the disposable plant NPV for fermentation yield and cost of materials. The fermentation yield was varied from 50% to 100% of the yield obtainable with the conventional plant. The reduction in the fermentation yield was considered to be a result of (a) lower biomass obtainable, and (b) lower expression level of the cells. A reduction in the cost of the materials from 0 to 50% allowed the impact on the NPV of the disposable plant to be investigated. NOVAIS ET AL.: ECONOMIC EVALUATION OF DISPOSABLES-BASED BIOPROCESSING 151 expected that it will be decreased once a market for disposable equipment has been established. In both cases, a 50% saving in materials costs compensates for the loss in yield, bringing the NPV up to 112% (Fig. 6A) and 96% (Fig. 6B) of that of the base case. The final source of process yield variation considered in this study was that due to a lower yield in the disposables chromatography format. This may be thought to arise as a consequence of less-specific binding resulting in product loss in the wash step. Alternatively, a reduced yield might also be due to a lower capacity of the matrix for the product. The first case requires the use overall of larger process volumes, while the second case results in the need for higher volumes of matrix and of chromatography buffers. The second scenario was analyzed here. Figure 7 shows the results of a sensitivity analysis performed for a reduction in the chromatography yield in combination with a reduction in the cost of the materials. This is a very likely case because a choice of a matrix with a lower capacity would most certainly be driven only by economic considerations. The NPV decrease is small at only 10% when the chromatography yield is 50% lower and this can be compensated for by a 25% saving in materials costs, for which case the NPV would be 107% of that of the base case. DISCUSSION AND CONCLUSIONS The objective of this work was the economic modeling of a fully disposable bioprocessing plant. This was achieved using conventional models and assumptions derived from the definition of disposables-based bioprocessing. Although the NPV values indicate the conventional option to be the most attractive ($76M for disposables and $103M for conventional), the difference at only 25% is probably sufficiently close to make disposables a viable alternative, especially when considering the other advantages of disposable plants outlined in the Introduction. Figure 7. Combined sensitivity analysis on the disposable plant NPV for chromatography yield and cost of materials. The yield of the chromatography step was varied from 50% to 100% of the yield obtainable with a conventional chromatography column. Reduction in yield was considered to be a result of a lower capacity of the disposable matrix for the product. A reduction in the cost of the materials from 0 to 50% allowed the impact on the NPV of the disposable plant to be investigated. 152 It has to be noted, however, that the use of a conventional process engineering model for the calculation of the running costs (analysis not shown) indicated disposable plants as the more attractive option, with a higher NPV. This result is contrary to what could be intuitively predicted from the definition of a disposables-based plant where the increase of the variable associated to disposable equipment would be expected to have a higher impact on the NPV. For this reason, we considered the model presented in this article for estimating conventional-based process costs as more appropriate to describe a biochemical engineering process. This does, however, show how critical it is to identify an appropriate running costs model, which is a common problem found in biotech costing studies. It was also crucial to perform sensitivity analysis on the assumptions made to develop the disposable models to evaluate how much impact they would have on the overall model if they were invalid. The sensitivity analysis done for the case study (Fig. 4) shows that the overall disposable model is not very sensitive to the critical variables. This means that the estimation of these disposable values will not affect crucially the conclusions of any economic comparison. Turning now to consider the differences between the two modes of processing, it is clear that staff costs have a small effect on the way the conventional and the disposable options compare. This means that even if the staff requirements of a disposable plant are less than those of a conventional plant that would only result in a slight increase of the NPV ratio. A decrease of 25% in the capital investment would result in a variation of less than 5% on the NPV ratio, showing insensitivity to this variable. This finding provides more confidence in the “Lang” factor used for the disposables-based option (Ldisp). Materials’ costs are shown to be more influential but their impact on the NPV ratio is still less than 10% achieved by a 25% reduction in materials costs. It can be concluded from Figure 4 that the costing model is not strongly dependent on the assumptions made for the disposables plant and that it is consequently reliable to cost this bioprocessing option. The disposable “Lang” factor (Ldisp) developed in this article is perceived as a useful variable for the future economic evaluation of disposable bioprocessing plants. Its value (60% of Lconv) shows the strong reduction in the capital investment achievable by a disposable approach. This results in increased flexibility for the disposable plants because changes in the process or the product result in less financial loss, which is of great interest for start-up companies. It also allows for an earlier decision to build which may result in earlier entry to market. This effect benefits strongly the disposables option because a reduction in time to market has a high impact on the NPV ratio, as shown in the results of the sensitivity analysis and depicted in Figure 4. The increased flexibility also arises from the concept of “pay as you need,” due to the transfer of capital costs onto running costs, making it less damaging when a product fails BIOTECHNOLOGY AND BIOENGINEERING, VOL. 75, NO. 2, OCTOBER 20, 2001 in clinical trials and advantageous given the current industry trend of ever shorter project lives. A final difficulty encountered in the analysis of disposable bioprocessing is to establish the degree of similarity between disposables-based plants and conventional designs. The fermentation step is an example of how the different engineering features of a disposable plant can have a detrimental impact on product yield. Although the impact on yield may happen in different ways, an analysis of the sensitivity in Figure 5 shows that this affects the obtainable NPV by less than 20%. Figures 6A and 7 show that the reduction in the achievable yield given by a disposables-based option and associated with lower levels of biomass production in the fermentation and with reduced chromatographic performance has only a limited impact on the NPV that is realized. By contrast the loss of yield due to less-productive cells has a dramatic impact on the NPV (Fig. 6B). This is, however, a less likely scenario. The analysis shows that any fall in NPV can easily be overcome when disposable equipment producers start responding with higher production scales and hence, lower prices to an increasing demand for their products. Effectively, a 50% reduction in the yield of the fermentation can be compensated for by a 50% reduction in the cost of the materials (Figs. 6A and 6B). This is even more striking in the case of the chromatography (Fig. 7) where even a 50% loss in yield can be overcome by a saving of 25% in the materials costs, a margin that appears highly probable as the disposables approach starts to gain acceptance. Other questions that could be asked include: Can the number of batches achieved per year in a disposable plant be improved because there is no downtime for CIP and SIP? Does a disposables-based process effectively allow for quicker entry to market? These questions will be addressed in a future article. In conclusion, a disposables-based plant with the same features as its conventional equivalent is economically and conceptually attractive as it may be of easier and quicker implementation with a comparable overall investment required. This work is part of an Innovative Manufacturing Initiative (IMI) project entitled “Bioprocessing With Disposables To Cut costs and Increase Process Efficiency for Biopharmaceuticals” funded by the Biotechnology and Biological Sciences Research Council. The support of Millipore, Kvaerner Process, Lonza Biologics, Hyclone and Biotage is gratefully acknowledged. Joana L. Novais is supported through the Praxis XXI programme of the Portuguese Ministry of Science and Technology. NOMENCLATURE c CF Econv contingency factor annual cash flow, in $ equipment costs (conventional plant), in $ fi f i8 FCI L m n NPV r RC S t xi yi conventional plant factors for calculation of individual FCI items factors which translate conventional plant FCI items into disposable ones fixed capital investment, in $ “Lang” factor year of entry to market project year net present value, in $ discount rate running costs, in $ value of annual sales, in $ project life, in years fractions of conventional running cost conversion factors for calculation of fractions of disposables running cost Subscripts conv disp n refers to the conventional plant refers to the disposable plant refers to year n References Basu PK, Quaadgras J, Mack RA, Noren AR. 1998. Achieve the right balance in pharmaceutical pilot plants. Chem Eng Prog 94:67–74. Burnett MB, Santamarina VG, Omstead DR. 1991. Design of a multipurpose biotech pilot and production facility. Ann NY Acad Sci 646: 357–366. Datar RV, Cartwright T, Rosen C-G. 1993. Process economics of animal cell and bacterial fermentations: A case study of tissue plasminogen activator. Bio/technology 11:349–357. Ernst S, Garro OA, Winkler S, Venkataraman G, Langer R, Cooney CL, Sasisekharan R. 1997. Process simulation for recombinant protein production: Cost estimation and sensitivity analysis for Heparinase I expressed in Escherichia coli. Biotechnol Bioeng 53:575–582. Gamerman GE, Mackler BF. 1994. Winning in today’s biopharmaceutical marketplace. Chem Tech 24:37–41. Gill A, Bracewell DG, Maule CH, Lowe PA, Hoare M. 1998. Bioprocess monitoring: An optical biosensor for rapid bioproduct analysis. J Biotechnol 65:69–80. Hamers MN. 1993. Multiuse biopharmaceutical manufacturing. Bio/ technology 11:561–570. Lang HJ. 1948. Simplified approach to preliminary cost estimates. Chem Engng 55:112–113. Nicholson I. 1998. Outsourcing manufacturing: A strategic rationale. Paper presented at the Biopharm Europe Symposium, October, 1998. Pearl SR, Christy CD. 2000. Heat exchange apparatus. U.S. Patent 6,131,649. Peters MS, Timmerhaus KD. 1991. Plant design and economics for chemical engineers. New York: McGraw Hill. 910 p. Sinnott RK. 1991. Chemical engineering, Volume 6: An introduction to chemical engineering design. Coulson JM, Richardson JF, editors. Oxford: Pergamon Press. 838 p. Struck MM. 1994. Biopharmaceutical R&D success rates and development times. Bio/technology 12:674–677. Zaidi A, Ghosh P, Schumpe A, Deckwer WD. 1991. Xanthan production in a plunging jet reactor. Appl Microbiol Biotechnol 35:330–333. NOVAIS ET AL.: ECONOMIC EVALUATION OF DISPOSABLES-BASED BIOPROCESSING 153