Continuous processing is increasingly seen as having the potential to address the need for reduced manufacturing costs, improved product quality and increased flexibility, while at the same time reducing technology transfer risks. These benefits are also expected to be delivered with an ever-smaller facility footprint as well as being faster and more cost effective. Easier said than done!
In downstream processing, the barriers to continuous processing used to revolve around the limitations of the technology available, while perfusion bioreactor-based processes were viewed as taking a long time to develop and the economics for their use was not clear. As technologies have become more advanced, one of the primary factors holding companies back from making the shift to continuous processing is the lack of prior insight as to what results it will deliver. No company has yet adopted fully continuous processing for commercial biomanufacturing process, and the new technologies required for doing so are not mature and have not yet been evaluated based on cost performance.
Historically, the data simply have not been available for biopharma companies to evaluate the cost-effectiveness of continuous processing for their business. But today, using modelling and analysis tools, such as BioSolve Process, many of the major potential end users of continuous processing technologies have modelled their technologies in the software. They want the impact of continuous processing to be accurately assessed for the potential benefit to the manufacture of new processes in development. It is early days, but companies are starting to realise the value, and explore ways to assess risk mitigation.
Perfusion vs fed-batch
Some biopharma companies have been running perfusion bioreactors in a continuous mode for several decades. The choice of whether to use continuous or not was dictated by the product, and little has been published on the economics of continuous perfusion processes. By modelling perfusion and fed batch processes at different scales and analysing different manufacturing configurations in BioSolve Process, the economics of perfusion and how it changes with scale were explored.
To understand the impact of perfusion costs in isolation, three process configurations were defined:
- Conventional (fed batch upstream, batch downstream)
- Hybrid (fed batch upstream, continuous downstream)
- Continuous (perfusion upstream, continuous downstream)
As the basis for this analysis a monoclonal antibody process (mAb) was looked at for three scales – 100kg/year, 500kg/year and 2000kg/year – to represent the range typically seen in manufacturing. In the first case, a typical high titre mAb at 4.5g/L and the equivalent ‘traditional’ perfusion process at 0.9g/L, were looked at.
Upstream modelling provides insight into the impact of titres and media usage/cost on the comparison of fed batch versus perfusion. Two scenarios are presented in Table 1: one based on a stainless steel facility and the second based on a single-use technology. The table, showing capital expenditure and operating costs (CoGs), provides interesting insights. First, the impact of perfusion on the process savings can be distinguished by looking at the savings associated with the hybrid process compared with the fully continuous process. The main difference between these two configurations is that the hybrid option uses batch bioreactors for the upstream operation whereas the continuous is based on perfusion.
| Table 1: Capital expenditure and operating costs (CoGs) | ||||||
| 100kg/year | 500kg/year | 2000kg/year | ||||
| CAPEX | CoG | CAPEX | CoG | CAPEX | CoG | |
| ($106) | ($/g) | ($106) | ($/g) | ($106) | ($/g) | |
| Stainless | ||||||
| Batch | 79 | 318 | 131 | 107 | 186 | 45 |
| Hybrid | 81 | 306 | 116 | 85 | 163 | 35 |
| Continuous | 59 | 217 | 79 | 72 | 119 | 42 |
| Single use | ||||||
| Batch | 21 | 163 | 47 | 65 | 164 | 42 |
| Hybrid | 24 | 147 | 45 | 49 | 141 | 31 |
| Continuous | 32 | 144 | 48 | 56 | 102 | 40 |
Key observations include:
- At lower throughputs, the economics favour the perfusion process.
- At a very large scale (2000kg/year), the batch-based bioreactor is more economic.
- In this case, at about 500kg/year, there is a pivot point where for the stainless process the perfusion bioreactor is favoured (although the savings are diminished), whereas for the single-use scenario the batch reactors are favoured.
So what is happening? Further insight can be obtained when using BioSolve Process to look at the cost of goods breakdown at the extremities of the capacity (see Figure 1).
The material costs for the low-throughput process are not a significant contributor but, for a large-scale operation, the materials costs dominate for the perfusion-based process. Drilling into the numbers further, the perfusion bioreactors in the continuous process are found to consume a large amount of media, used at a rate of two volumes of media per bioreactor volume per day (VVD), and the cost of the media is seen to be a significant driver of CoGs at the larger scale.
This provides useful insight as it identifies targets for process improvement. It can be concluded that perfusion bioreactors will be more attractive at lower throughputs, but there is potential for using perfusion in larger scale processing if:
- The cost of the media can be reduced
- The media use efficiency can be reduced by:
- Reducing VVD for a given titre
- Using a modern high cell density cell line producing higher product titres (this has to take account of increased costs associated with more complex media).
What is happening in downstream processing?
Increasingly, the focus of continuous processing has moved downstream and, in recent years, the industry has seen the development of operations that can be run in a continuous process. But first what is meant by continuous operation must be considered.
From a process perspective, this involves moving from a discrete batch mode, where the batch size and frequency are used to define throughput, to a continuous processing environment, where there is a flow of material through a production line. The important concept here is that there is a continuous flow of material in and out of the unit operation.
What happens in the unit operation can be treated as a black box as long as there is a continuous flow in and out. Purists would argue that simulated moving bed chromatography is the only true continuous way of operation; however, all the continuous chromatography operations on offer are variants of using switched mode columns, with between two and more operated in sequence. In fact, any batch operation can be run in a continuous flow line and hence realise many of the benefits of continuous operation.
Figure 2: Two options for downstream continuous operation using membrane absorbers
There are two basic options that are illustrated using membrane absorbers (see Figure 2).
Option 1: the batch operation is duplicated. In this case the membrane absorber is run in parallel – one on standby and one in operation. The minimum size is based on the time taken to switch from standby to operation and the user can select any suitable size above the minimum.
Option 2: a surge vessel is sized to accommodate the line flow during the switch from an exhausted membrane to a fresh unit. Surge vessels are often used to dampen flow fluctuations in the line. There is again flexibility in sizing the absorber.
The data: By applying these principles and employing technologies, such as switched mode multi column chromatography, which can be run in a continuous fashion, the economics of continuous operation can be analysed. The data presented below used BioSolve Process to compare conventional batch processing with continuous operation.
Consider first the downstream operation at the 500kg/yr scale. In Table 2, a significant reduction in both capital (39%) and operating costs (33%) is seen when moving from batch to continuous operation. Though interesting, it would be better to understand how these savings are obtained.
| Table 2: Batch vs continuous costs as a function of scale | ||||||||||||
| Configuration | 100kg/year | 500kg/year | 2000kg/year | |||||||||
| CAPEX ($106) | CoG ($/g) | CAPEX ($106) | CoG ($/g) | CAPEX ($106) | CoG ($/g) | |||||||
| Batch | 79 | 0% | 318 | 0% | 131 | 0% | 107 | 0% | 186 | 0% | 45 | 0% |
| Hybrid | 81 | 3% | 306 | -4% | 116 | -12% | 85 | -20% | 163 | -12% | 35 | -22% |
| Continuous | 59 | -25% | 217 | -32% | 79 | -39% | 72 | -33% | 119 | -36% | 42 | -5% |
In BioSolve Process a batch cost breakdown can be generated by unit operation and cost category. This feature was used to generate Figure 3 – the left hand charts show the absolute costs and the right hand the normalised percentage breakdown. Upon inspection, the cost breakdown is very different:
Figure 3: Comparing downstream costs, batch vs continuous at 500kg/yr
- There are more unit operations contributing to overall cost of the batch process
- There is a more even distribution of costs in the batch process with the three chromatography steps dominating
- The continuous process costs are skewed to the two bind and elute chromatography steps
Comparing the Protein A column in the continuous process with the bind and elute Ion Exchange column in the batch process shows that the continuous column costs are dominated by capital costs associated with capital hardware, with less relative contribution from the resin costs. The implication is that capital hardware costs are relatively high and there could be significant scope for greater cost benefits if the hardware costs reduce as the market develops.
Finally, the total costs and the profile of cost distribution between upstream and downstream are considered in Table 3. For capital costs, most of the continuous benefits are gained in the upstream area. Despite the much smaller scale in continuous downstream, there is little capital reduction. As expected for continuous operation, materials are dominated by upstream – more so than in the batch process. Also, for continuous processing there is significant consumable use upstream, which reflects the use of ultra-filter membranes in the cell separation device.
| Table 3: Total costs – batch vs continuous at 500kg/yr | ||||||||||||
| Capital | Materials | Consumables | Labour | Others | Total | |||||||
| Batch Process | ||||||||||||
| USP | 29.5 | 54% | 3.8 | 81% | 0.1 | 1% | 9.3 | 48% | 7.4 | 53% | 50.1 | 47% |
| DSP | 25.4 | 46% | 0.9 | 19% | 14.0 | 99% | 10.2 | 52% | 6.6 | 47% | 57.1 | 53% |
| Total | 54.9 | 100% | 4.7 | 100% | 14.2 | 100% | 19.5 | 100% | 14.0 | 100% | ||
| Percent | 51% | 4% | 13% | 18% | 13% | 100% | ||||||
| Continuous Process | ||||||||||||
| USP | 10.1 | 30% | 15.8 | 98% | 1.7 | 23% | 2,8 | 38% | 2.6 | 32% | 33.0 | 46% |
| DSP | 23.1 | 70% | 0.3 | 2% | 5.7 | 77% | 4.5 | 62% | 5.7 | 68% | 39.3 | 54% |
| Total | 33.2 | 100% | 16.1 | 100% | 7.4 | 100% | 7.2 | 100% | 8.4 | 100% | 72.3 | 100% |
| Percent | 46% | 22% | 10% | 10% | 12% | -5% | ||||||
In conclusion, at the 500kg/yr scale some real benefit can be seen in moving to continuous processing, despite the much higher material costs associated with the perfusion bioreactor. It would appear that the cost of chromatography hardware is high and that this is restricting the benefit we would expect in downstream processing when running continuously. This reflects an immature equipment market in this area and it can be expected that costs will come down in the future, demonstrating potential benefits for continuous processing.
