Experimental design of a twin-column countercurrent gradient purification process

Highlights

• The performance of a two-column MCSGP chromatographic process capable of alleviating the purity-yield trade-off of classical batch chromatography is presented.

• A procedure to design the process operating conditions based on a single batch design chromatogram is presented.

• The impact of different process variables on its performance is demonstrated.

• As an example, a single charge isoform of a monoclonal antibody is purified with purity and yield > 90%.

Abstract

As typical for separation processes, single unit batch chromatography exhibits a trade-off between purity and yield. The twin-column MCSGP (multi-column countercurrent solvent gradient purification) process allows alleviating such trade-offs, particularly in the case of difficult separations. In this work an efficient and reliable procedure for the design of the twin-column MCSGP process is developed. This is based on a single batch chromatogram, which is selected as the design chromatogram. The derived MCSGP operation is not intended to provide optimal performance, but it provides the target product in the selected fraction of the batch chromatogram, but with higher yield. The design procedure is illustrated for the isolation of the main charge isoform of a monoclonal antibody from Protein A eluate with ion-exchange chromatography. The main charge isoform was obtained at a purity and yield larger than 90%. At the same time process related impurities such as HCP and leached Protein A as well as aggregates were at least equally well removed. Additionally, the impact of several design parameters on the process performance in terms of purity, yield, productivity and buffer consumption is discussed. The obtained results can be used for further fine-tuning of the process parameters so as to improve its performance.

Introduction

In the downstream processing of biopharmaceutical products, due to the high value of the target product, it is important to achieve large overall yields, while keeping purity within specifications. This is made more difficult by the fact that typically a few different chromatography steps are needed to achieve the desired product purity [1]. Accordingly, each of these steps needs to fulfill stringent constraints in terms of productivity, purity and yield. While process-related impurities such as DNA and HCP exhibit rather different adsorption characteristics and are therefore relatively easy to be removed, product-related impurities such as aggregates and fragments or unwanted isoforms constitute a more sever challenge. These impurities are typically removed in the polishing steps which follow the capture step. Due to their limited selectivity, in single columns operated in the batch mode, such impurities typically overlap with the chromatographic peak of the target product, including species eluting both before and after it. In such a situation batch processes typically exhibit a severe trade-off between purity and yield. In particular, when taking a very narrow product collection window it is possible to achieve high purities but low yield. Of course, the latter can be improved by enlarging the window width but this comes at the expenses of the purity which decreases rapidly, particularly in the case of difficult separations. As the purity often needs to be above certain specifications, this results in a decreased yield and product loss.

In general, there are three possible solutions for this problem. The first one is to use very efficient stationary phases, for example with fast mass transfer rates, therefore sharpening the elution peaks [2]. This can be achieved with smaller resin particles, which has the drawback of an increased pressure drop [3] and therefore calls for low flow velocities, which usually lead to too low productivities. An alternative would be to increase the column length, which would result in similar results [3]. A second possibility is to modify the process operating conditions, for example decreasing the load amount or the gradient slope, thus reducing the overlapping regions [4]. However, also in this case, the productivity decreases and the solvent consumption increases, resulting in high processing costs [5], [6]. A third possibility is to replace the single column batch operation with a multi-column process, where the overlapping region of the chromatogram is recycled within the unit while only the fraction of the target product which satisfies the purity specifications is taken out of the unit. This allows achieving simultaneously a higher purity and a higher yield, thus softening the trade-off mentioned above [7], [8]. Of course, such processes are more challenging to design and operate than single column processes [9]. In this work, we develop a simple procedure to design such processes with specific reference to the twin-column multicolumn countercurrent solvent gradient purification (MCSGP) process. The main idea is to start from a single column batch chromatogram where the operating conditions, such as load amount and gradient slope, have been selected to obtain a region in the chromatogram where the target product purity specifications are satisfied. This is referred to as the “design chromatogram”. Using the developed procedure, such conditions are then transferred to the twin column MCSGP unit where the product purity is maintained, while increasing yield at similar productivity with respect to the single column process. It is worth noting that the developed procedure does not have the ambition to identify the absolute optimum of a given process, but rather to identify reliable operating conditions which satisfy purity constraints with a very limited experimental effort. Other possibilities for process design of multi-column separation processes include mechanistic modelling [10], [11], simplified models like the “triangle theory” for SMB [12], empirical design [13] or online optimization [14], [15]. All these approaches require significantly larger experimental (and modelling) effort, and could be regarded as fine tuning procedures to be applied starting from the operating conditions identified with the empirical procedure developed in this work.

It is worth noting that the three strategies for high purity and yield discussed above do not exclude each other. Actually, optimized performance can be achieved by properly considering all of these and acting on all possible process parameters. In this work, we consider, with illustrative purposes, the isolation of the main charge isoform of a monoclonal antibody (mAb), which is indeed a rather challenging separation task [16]. This is conducted with both batch and multi-column chromatography, using a small particle resin at relatively low load amounts so as to obtain reasonably high purity and yield. It is worth noting that although the design procedure is explained in the following with reference to ion-exchange chromatography, the same principles can be applied also for other types of chromatography, such as reversed phase chromatography, where the gradient in the mobile phase as well as in-line adjustment is realized using a suitable modifier rather than pH or a salt [17].