(669e) Optimal Design of Multicolumn Countercurrent Solvent Gradient Purification

Chromatography is one of the most common separation techniques in the pharmaceutical industry. Single-column batch chromatography is generally used, and while it may be the most straightforward operation mode, it suffers from a yield-purity trade-off problem when separating a close-eluting mixture (Steinebach et al., 2016). To overcome this, a multicolumn counter-current solvent gradient purification (MCSGP) process was proposed by Strohlein et al. (2006), which aims to efficiently separate the early- and late-eluting impurities from the product; the simplest example would be a ternary mixture containing a weak-adsorbing (early-eluting) impurity, the product, and a strong-adsorbing (late-eluting) impurity. The resemblance of counter-current flow in MCSGP is achieved by periodically switching the valves in the direction of fluid flow, thus simulating continuous flow similarly to simulated moving bed (SMB) operation. MCSGP is a semi-continuous process where the columns alternate between batch mode and interconnected (continuous) mode; the batch mode separating the “pure" product and impurities, while the interconnected mode recycles the overlapping regions. The initial MCSGP process consisted of six columns (e.g., columns 1, 3, and 5 operate in batch mode while columns 2, 4, and 6 operate in interconnected mode). A solvent pump is installed before each column to control the gradient in each column so that the gradients can form, for example, a linear gradient throughout the entire process (Aumann and Morbidelli, 2007). To reduce capital costs, Aumann and Morbidelli (2008) proposed a three-column configuration, where all three columns first operate in batch mode, and after some time (the switch time), they switch to interconnected mode. Later, Angarita et al. (2015) and Steinebach et al. (2017) proposed an even more compact form based on a twin-column configuration. Half of the cycle of the twin-column MCSGP is made up of four steps alternating between interconnected and batch modes: in the first three steps, column 1 is responsible for separating the product from the mixture while column 2 is being prepared (feed and overlapping regions are fed into column 2 and are retained at the column inlet); in the fourth step, column 1 is being cleaned while column 2 separates the early-eluting impurity. In the next half of the cycle, these steps are repeated, but the roles of columns 1 and 2 are now switched. Both studies reported that the twin-column configuration was superior to a regular batch process in terms of separation performance, as well as reduced requirements for hardware and provided simpler operation compared to the three- and six-column configurations.

Although the MCSGP concept has been around for a while, modeling and optimization of such a complex operation is still limited in the open literature, although MCSGP has already been successfully implemented in laboratories and has shown promising performance (Muller-Spath et al. (2010), Steinebach et al. (2017), Menza et al. (2024)). As the MCSGP concept was originally developed based on the batch process, the batch design with specific features was transformed into the MCSGP design using a few simple equations to obtain the key MCSGP parameters and design variables (Steinebach et al. (2017), Hooshyari Ardakani et al. (2024)). It was found that, after some fine-tuning, the transformed MCSGP process usually gives a better yield than the batch process while still satisfying the purity constraints (Muller-Spath et al. (2010), Steinebach et al. (2017)). However, the quality of those MCSGP designs remains unclear. If the batch process is optimal, can the transformed MCSGP be further improved, and is this MCSGP process close to the optimal design that would be obtained by directly optimizing the MCSGP process? Unfortunately, optimization studies of a complete MCSGP are rather limited in the open literature and the true performance of MCSGP is not fully understood. To fill this research gap, and also to provide guidance on optimal performance of MCSGP, we consider and compare: (1) an optimal batch design obtained from a proper rigorous optimization, (2) the MCSGP design transformed from the optimal batch design, and (3) an optimal MCSGP design obtained from directly optimizing the MCSGP process by rigorous optimization.

To achieve these objectives, we developed the mathematical models for batch and twin-column MCSGP systems utilizing the same basic mass transfer model (equilibrium dispersive model) and isotherm equations (linear isotherm with linear solvent strength theory), i.e., only the operational logic is different between the operating modes. For the optimization, an objective function to maximize the product yield was chosen. Key design variables for a batch system include eluting flow rate, eluting duration, and gradient information of the eluting period, which are optimized using a particle swarm optimization (PSO) method. For the twin-column MCSGP, recall that one cycle contains eight steps, where the operation of the last four steps is the same as the first four steps but with the roles of the columns switched, meaning optimization of design variables for half of the cycle is sufficient for an optimal design. Recall that for a twin-column system, each step involves two pumps for adjusting flow rates and gradients. Therefore, for half a cycle, eight flow rates must be optimized. In addition, the solvent gradient of the gradient pump must be optimized, and the solvent composition of other pumps (feed pump, dilution pump, and spike pump) must also be optimized or be pre-defined. Furthermore, the switch times of each step are also optimization variables; important as switching time directly affects the product collection period, hence the product yield. The optimal MCSGP design is compared to the optimal batch design as well as the MCSGP design that is transformed from the optimal batch design. This work will provide guidance for how to optimize MCSGP design and operation, but more importantly, evaluate the true performance of batch-to-MCSGP transformation, and finally consider the performance between MCSGP and regular batch operation for selected case studies.

References

Angarita, M., Muller-Spath, T., Baur, D., Lievrouw, R., Lissens, G., Morbidelli, M., 2015. Twin-column CaptureSMB: A novel cyclic process for protein A affinity chromatography. Journal of Chromatography A 1389, 85–95. doi:10.1016/j.chroma.2015.02.046.

Aumann, L., Morbidelli, M., 2007. A continuous multicolumn countercurrent solvent gradient purification (MCSGP) process. Biotechnology and Bioengineering 98, 1043–1055. doi:10.1002/bit.21527.

Aumann, L., Morbidelli, M., 2008. A semicontinuous 3-column countercurrent solvent gradient purification (MCSGP) process. Biotechnology and Bioengineering 99, 728–733. doi:10.1002/bit.21585.

Hooshyari Ardakani, M., Nosengo, C., Felletti, S., Catani, M., Cavazzini, A., De Luca, C., Rezadoost, H., 2024. Enhancing the purification of crocin-I from saffron through the combination of multicolumn countercurrent chromatography and green solvents. Analytical and Bioanalytical Chemistry 416, 2553–2564. doi:10.1007/s00216-024-05228-6.

Menza, S.T., Prestia, R., Fioretti, I., Sponchioni, M., 2024. Model-Based Optimization Strategy for Intensification in the Chromatographic Purification of Oligonucleotides. Journal of Chromatography A, 465321doi:10.1016/j.chroma.2024.465321.

Muller-Spath, T., Aumann, L., Strohlein, G., Kornmann, H., Valax, P., Delegrange, L., Charbaut, E., Baer, G., Lamproye, A., Johnck, M., Schulte, M., Morbidelli, M., 2010. Two step capture and purification of IgG2 using multicolumn countercurrent solvent gradient purification (MCSGP). Biotechnology and Bioengineering 107, 974–984. doi:10.1002/bit.22887.

Steinebach, F., Muller-Spath, T., Morbidelli, M., 2016. Continuous counter-current chromatography for capture and polishing steps in biopharmaceutical production. Biotechnology Journal 11, 1126–1141. doi:10.1002/biot.201500354.

Steinebach, F., Ulmer, N., Decker, L., Aumann, L., Morbidelli, M., 2017. Experimental design of a twin-column countercurrent gradient purification process. Journal of Chromatography A 1492, 19–26. doi:10.1016/j.chroma.2017.02.049.

Strohlein, G., Aumann, L., Mazzotti, M., Morbidelli, M., 2006. A continuous, counter-current multi-column chromatographic process incorporating modifier gradients for ternary separations. Journal of Chromatography A 1126, 338–346. doi:10.1016/j.chroma.2006.05.011.