(230a) Chromatographic Scale-Up On a Volume Basis

The scale-up of a chromatographic separation is in principle a simple procedure since the process parameters are scalable in a linear fashion. The process is scaled by increasing the column diameter while keeping the bed height, the velocity, and the volumes of the different phases (measured in column volumes) constant. The problem with this simple and well established concept, usually referred to as linear scale-up, is the lack of flexibility.  

For practioners, especially those dealing with large scale operations, it is no secret that chromatography columns come in discrete sizes regarding the diameter. The diameter and the number of columns are the only design variables we have available when going into production scale. As a consequence we most often cannot choose the column volume that matches the desired capacity. We are faced with the problem that for a chosen diameter we need for instance 2.2 columns in order to meet the capacity demands. The solution is either to invest in a third column or go back to the lab and try to optimize the process in order to squeeze it into two columns. Both solutions can be rather costly either in investment or in time consumption in development. Further, we have no guarantee that the optimization will actually succeed (not to speak of the documentation burden if it does). Even when we manage to meet the right capacity we still risk encountering the same problem at a later date if we want to increase the production capacity. In pilot scale the problem can be even more pronounced since we must use existing equipment. The lack of flexibility of the column volume can therefore increase the number of cycles on each column and thereby the duration of the pilot campaign (where the chromatography steps are often the bottleneck). At the other end of the scale, in the development phase, we are met with similar problems. Knowing that the process will end up in production some day and the bed height must be kept constant we aim for a suitable bed height (like 20 cm) when developing the chromatography in lab scale. In order to avoid packing problems and flow irregularities a certain column diameter (~ 1 cm) is required and we therefore end up with a lower volume limit around perhaps 15-20 mL in order to have a representative process step. Especially in the early phase of development where the product quantities are limited this can be a substantial volume which reduces the number of trials and we therefore risk a delay in development. A volume in this range is also certainly not desirable for special tasks like virus validation. 

Fifteen years ago we revisited the scaling principles of chromatography in order to find a more flexible way of scaling up (or down). From a conceptual point of view the solution was quite simple and does not involve any new theory. By scaling the process based on a constant contact time (residence time) we can, within limits, introduce the column bed height as a process design variable rather than a fixed parameter. Since the bed height is a continuous variable, as opposed to the diameter, we can use it to obtain exactly the volume we need in any scale. After proof of concept in lab scale and later in pilot scale the new concept became an established approach which has proven extremely valuable over the entire range from development in lab scale to commercial production. We have previously presented and published some of our early results. Our expectation at the time was that this would become an established approach at least for pharma companies producing significant volumes. But our experience from interactions with companies, including producers of chromatography equipment, and with the regulatory authorities has shown that this is apparently still new territory. Since we find the approach very useful to the field of industrial chromatography we would like to take the opportunity to present ten years of experience in chromatographic scale up on a volume basis. The presentation will include the theoretical background and proof of concept as well as industrial cases with scaling factors of more than 1000. Examples include IEX, HIC, RPC, and SEC processes for the separation of both peptides and large proteins.