David Johnson, Global Product Manager, Chromatography Hardware and Systems, Pall Life Sciences, speaks on the need for automated packing systems to improve column chromatography
A consistent and long-lasting packed bed is crucial when using a chromatography column for purifying biopharmaceutical products. An improperly packed column can lead to ineffective separation, time-consuming repacks and, ultimately, to products that fail to meet specifications. It can also increase costs, not least because some of the solid-phase media, when packed with wasteful methodologies, can result in large unnecessary expense.
Using traditional packing methods takes experience and practice. Until the recent advent of fully automated packing systems, the process was largely manual, and operator-driven packing still predominates today. Yet the single biggest cause of packing failure remains operator error. In fact, most pharma manufacturing operators highlight packing as one of the biggest risks in their downstream processing. Ideally, they would remove chromatography entirely from the manufacturing train and replace it with something easier to predict and control, but this is rarely possible given the effectiveness of column chromatography when in process.
Historically, a group of four or five operators work together to pack a column, in a routine that is closely choreographed by a SOP or batch record. This lays down detailed instructions for how the column should be packed. A suspension of particles of the separation medium is transferred from a slurry tank into the column, where it must be evenly distributed and then compressed to achieve optimal performance.
At the heart of the difficulties is the ability to execute the SOP’s list of instructions in a reliable and reproducible way. All the team operators must be absolutely certain of what they are doing, and do everything both correctly and at exactly the right time. This is not as simple as it sounds and the chance of a mistake being made is high.
A knock-on risk to the business is that operators have to become very skilled to achieve the desired outcomes. If they are not available on any given day, or even choose to move to a job at another company, reliance on key operators means the packing will have to be led by a less experienced operator or cannot be done at all.
Even if a detailed paper SOP is followed to the letter, transferring the process to another plant can be fraught with difficulties. It has been shown that the way the same SOP is executed by teams of people in separate locations often differs. Maintaining consistency is a real issue, teams may execute the SOP down to the last full stop, but it is almost inevitable that something will be changed or omitted, however minor, because a paper SOP cannot be perfectly interpreted. Full automation of the unit operation presents the opportunity to remove most, if not all, of this risk, because automation involves machines. A machine will do exactly the same thing every time, which is a real advantage in achieving consistency when transferring technology.
Many SOPs are designed to leave an excess of the slurried medium in the tank after column packing. This is usually done to avoid breaking the prime inside the column and introducing air. Removing air bubbles not only adds an additional layer of difficulty, but Protein A media, for example, one of the most common solid phases in protein purification, can cost $15,000 a litre. 50 litres remaining in the tank after packing is a significant amount of extremely costly waste. The latest generation of automation and column geometry allows the column to be packed without the requirement of an excess, leaving empty tanks and pipework.
If a packing procedure fails, then the manual method requires the medium to be removed from the column and put back into a slurry tank before it is repacked. Not only does this take time, it increases buffer use and adds to hardware cleaning demands. Carefully designed automation ought to be able to repack the column without the need for removing the column’s contents, saving both time and costs associated with consumables.
Full automation greatly reduces the opportunity for problems to occur. The ideal automation would implement ‘one touch’ operation for packing, unpacking and the cleaning in place of the empty column, making the involvement of the operator minimal. Further benefits would include lower maintenance and cleaning costs, and smaller cleanroom and storage footprints. Properly implemented, no additional pumps or packing skids should be required.
In order to automate column packing, an actuated column is required, with a piston that can be moved up and down, and additional process valving that can be controlled remotely. The automated system must be able to control both the piston’s movement and the valving. Additional sensors are required to detect parameters such as liquid levels and pressures, which enables the pack to be monitored in real time. This is essential to achieve optimum conditions, as it allows the behaviour of the machine to be altered in response to the outputs of the sensors. The sensors also act as signals that individual parts of the packing process have been completed.
The operation of a fully automated system should be simple. Once the operator presses the ‘start’ button for packing, it will agitate the tank, prime the filler and then suction-transfer the slurry into the column. The system then packs the optimal bed and informs the operator that the process is complete.
The software driving the process should include multiple fully parameterised method templates, with a highly adaptable configuration. The interface should be easy to use, and suitable for a general operator rather than a highly trained specialist. The method templates must be robust and must have been validated by the equipment manufacturer or the user’s engineering team, and be configurable to the precise requirements of the process run. It is a bit analogous to baking a cake, small adjustments can be made to create a chocolate cake instead of a lemon cake, but the actual baking process essentially remains the same. Using method templates is much more straightforward than validating a paper SOP, and once they are set, the operator is only able to execute the recipes, not alter them.
Process chromatography columns are essentially pressure vessels, and as columns get larger, the forces required to contain the pressure during packing and operation get exponentially greater. Including more than one actuator for piston control is an advantage in terms of gaining the ability to contain these forces, particularly when columns can be as wide as 2m. Three separate actuators, spaced out around the piston, allow the precise levels of the piston to be monitored and controlled. The result is a decrease in the likelihood of damage being caused to the column during the packing process.
Should packing fail for any reason, an automatic repacking process allows it to be fixed without emptying the medium into a separate slurry tank, rendering a faster turnaround with no tank cleaning or other equipment required. The buffer consumption for repacking is very low; for example, around 150 litres of buffer are required for a 1400 mm ID packed bed with a 190-litre bed volume.
Ideally, the system will be designed such that the only time the media is removed from the column is when it has reached the end of its life, or is no longer required for that particular process.
When the media needs to be removed from the column, it should be unpacked using as little liquid as possible. An automated process facilitates predictable usage of unpacking liquid and efficient use of buffers, resulting in equivalent or fewer slurry volumes to manual unpacking by an experienced operator. Importantly, almost all the sorbent is recovered. In equipment terms, no pump is required and lower media shear is created.
When the operator presses the start button for the unpack, the system initiates the suction of the unpacking liquid, and the medium is then re-suspended before the slurry is pushed back to the tank. Then, another aliquot of unpacking liquid is suctioned, allowing the remaining media to be re-suspended and pushed back to the tank in slurry format. An optional further rinse can follow, if necessary.
Further ease of use can be delivered via an automated clean-in-place (CIP) process. First, the CIP liquid is suctioned into the column. The pipework and valves are turbulent-flow-cleaned, and the CIP liquid is then held at full bed height before the column is emptied. Once the operator changes the tanks, a mandatory manual step between two fully automated sequences, neutralising liquid is sucked into the system and the column is rinsed. Again, there is the option of adding an additional rinse.
Not only does it clean the empty column, all pipework and valve weirs are cleaned independently. There is a single exit for waste flows, facilitating the monitoring of pH or conductivity. All velocities and hold times are configurable to ensure optimised liquid consumption. And, again, no pump is required, saving on equipment costs.
With the modern trend toward single-use equipment to reduce cleaning costs and minimise the risk of carryover, it is possible to include at least some disposable technology within the automated column packing process. It is still early days, but by using an integrated pinch-valve block that can be transferred between columns, fully automated operation is possible using a disposable manifold in conjunction with single-use buffer biocontainers. A mixing tote can also be incorporated, to provide a fully disposable slurry handling train. This removes much of the need for cleaning, and incorporates minimal servicing requirements.
As an example, the driving force behind automatically emptying a slurry tank is to reduce operating expenses resulting from the cost of the separation medium. By modelling a customer’s process information, it was possible to design a system that would save them money. Given its normal operator procedures and the number of times a year a column was packed, we showed that the automated system would be significantly cheaper to run, even given the equipment and implementation costs.
The biggest reduction in expenditure comes in reduced media costs. Savings of $1.3 million a year might be expected for a typical 2000 mm Protein A capture step compared with existing technologies. The savings rise to an annual $1.9 million over a three-column train. A further advantage is the reduction in the documentation required to address sorbent carryover. It is the ability to empty the slurry tank—something that it is not practical to do when packing manually—that leads to these substantial cost savings.
In a second example, the customer was concerned about the potential for operator error in a vaccine separation process. The frequency of column repacks was very high, and the customer wanted to reduce errors, with knock-on reductions in packing labour costs and sorbent consumption. Upon implementation of automated packing, the labour costs were modelled at 40 per cent and, importantly, the packing success rate was forecast to rise from 62 per cent for a manual pack to an anticipated 99 per cent for automated packing.
Finally, in a third example, retrofitting automated packing technology into an existing column set up also saved money. The original highly engineered packing system was being used to purify a biosimilar product. With the investment having already been made in columns, the customer did not want to start again, but it was possible to retrofit an automated system into the columns so that capital cost was not wasted.
In summary, manually intensive column packing remains a productivity bottleneck, and a significant pressure point where errors are made. With the ability to fully automate sorbent handling and column packing, as well as unpacking and CIP of the empty column, this bottleneck can be relieved. The ability to get expert levels of packing repeatedly without the need for specialised personnel eases staffing challenges, facilitates technology transfer and, importantly, maximises usage of the expensive medium.