Purification Of Synthetic Peptides

Most therapeutic peptides are synthesized through chemical solid-phase peptide synthesis (SPPS), generating synthesis-related product impurities that need to be removed by chromatographic purification steps. Reverse-phase high-performance liquid chromatography (RP-HPLC) is the method of choice to remove the impurities, often employing automated batch rechromatography steps (i.e., separate purification runs to recover the product from impure side fractions) to increase yield while maintaining target purity. Advances in chromatography separation have mostly relied on developing better chromatography stationary phases and finding the optimal combination of stationary phases, mobile phase (solvent) and load of feed material in order to improve resolution between product and impurities. However, single-column batch chromatography faces problems even with advanced stationary phases when it comes to separating product and impurities having very similar adsorptive properties. The general trade-off between yield and purity is an intrinsic feature of single-column batch chromatography with the consequence of losing yield at target purity. The yield-purity trade-off is accentuated when too much feed is loaded; thus, an optimal load is also needed to obtain sufficient resolution. Rechromatography as a means to improve yield leads to an accumulation of feed impurities and does not alleviate the yield-purity trade-off obtaining a better product yield without compromising target purity. Although rechromatography can be automated, 1 ^, 2 ^, 3 ^, 4 the overall purification time, including rechromatography, is prolonged by repetitive analysis of the side fractions confirming target purity prior to proceeding to the next rechromatography sequence. The feed of each rechromatography run, being composed of various side fractions, has a different composition than the original load material, caused by the accumulation of different impurities from each rechromatography sequence.

Improving the product separation from impurities by prolonging elution time, reducing feed load, and/or increasing column length will decrease productivity. With the intrinsic batch process constraints, good baseline separation of product and impurities remains a significant challenge for the preparative-scale single-column batch process. A way to overcome the yield-purity trade-off and to desensitize loading constraints is to use a continuous process that operates outside the yield-purity constraint and combines the separation power of advanced stationary phases with enhanced process capacities, accentuating the separation power of the stationary phase. Figure 1 shows the yield-purity trade-off relationship of conventional batch chromatography.

The MCSGP process is a unique ternary continuous separation process that can unlock the yield-purity trade-off by providing high yield at target purity, at up to 10-fold higher productivity. MCSGP, being a continuous process, also minimizes scale-up constraints and is suitable for large-volume processing applications. The process eliminates the need for rechromatography and the associated time and resources for sampling and testing. MCSGP has been used successfully in many applications, including purification of therapeutic proteins and peptides, protein isoforms, conjugated proteins, small molecules, macrocycles, and fatty acids.

This paper focuses on the applications and process economics of MCSGP in peptide purification. It summarizes the MCSGP process principle and design, outlines process advantages and productivity gains, shows case studies, describes the power of MCSGP for scale-up, and concludes on process economics.

MCSGP Process Principle and Design

The MCSGP process is a cyclic process applying internal recycling of impure side fractions to improve the yield,5 ^, 6 ^, 7 ^, 8 ^,  enabling high yield at target purity.

MCSGP design starts with a single-column batch process, which does not need to be optimized and is designed as an isocratic or gradient run. Obtaining a reasonable separation between product and impurity peaks in the batch run used as basis for MCSGP design is sufficient. Impurities that co-elute exactly under the product peak cannot be satisfactorily separated by MCSGP either. Therefore, it is important to ensure that some separation between product and impurities occurs in the batch run. The batch process is automatically transformed into an MCSGP process using an MCSGP process design tool embedded in the operating software of the continuous chromatography system, as explained further next.

The MCSGP process operates with two or more identical columns with the same stationary phase. MCSGP processes have been operated with up to eight column configurations.7 ^, 8 ^,9  MCSGP technology evolution reduced the required column configuration number to two, greatly reducing hardware and process complexity5  without loss in performance.

The process uses a minimum of two columns to operate several subprocesses continuously: feed containing the complex mixture with product and impurities is loaded on a first column and separation occurs as in single-column batch chromatography. While pure product is eluted for collection and impurities are discarded, both in a cyclically continuous manner, product-containing side fractions are kept in the process and are loaded on the columns followed by fresh feed until the most pure product is extracted. The columns are also cleaned and reconditioned as part of the process, thereby avoiding any accumulation of impurities or fouling of the stationary phase. Product elution is done by isocratic mode or linear gradient.

• 1Lee, J. W., and P. C. Wankat. “Comparison of Recycle Chromatography and Simulated Moving Bed for Pseudobinary Separations.” Industrial & Engineering Chemistry Research 48, no. 16 (2009): 7724–32.
• 2Bailly, M., and D. Tondeur. “Recycle Optimization in Non-Linear Productive Chromatography—I Mixing Recycle with Fresh Feed.” Chemical Engineering Science 37, no. 8 (1982): 1199–212.
• 3Tarafder, Abhijit, Lars Aumann, and Massimo Morbidelli. “Improvement in Industrial Re-Chromatography (Recycling) Procedure in Solvent Gradient Bio-Separation Processes.” Journal of Chromatography A 1195, no. 1–2 (June 2008): 67–77.
• 4Grill, Charles M., Larry Miller, and Tony Q. Yan. “Resolution of a Racemic Pharmaceutical Intermediate: A Comparison of Preparative HPLC, Steady State Recycling, and Simulated Moving Bed.” Journal of Chromatography A 1026, no. 1–2 (February 2004): 101–8.
• 5 a b Steinebach, F., N. Ulmer, L. Decker, L. Aumann, and M. Morbidelli. “Experimental Design of a Twin-Column Countercurrent Gradient Purification Process.” Journal of Chromatography A 1492 (April 2017): 19–26. doi:10.1016/j.chroma.2017.02.049
• 6Müller-Späth, T., M. Krättli, L. Aumann, and M. Morbidelli. “Increasing the Activity of Monoclonal Antibody Therapeutics by Continuous Chromatography (MCSGP).” Biotechnology and Bioengineering 107, no. 4 (November 2010): 652–62. doi:10.1002/bit.22843
• 7 a b Aumann, L., and M. Morbidelli. “A Continuous Multicolumn Countercurrent Solvent Gradient Purification (MCSGP) Process.” Biotechnology and Bioengineering 98, no. 5 (December 2007): 1043–55. doi:10.1002/bit.21527
• 8 a b Aumann, L., and M. Morbidelli. “A Semicontinuous 3-Column Countercurrent Solvent Gradient Purification (MCSGP) Process.” Biotechnology and Bioengineering 99, no. 3 (February 2008): 728–33. doi:10.1002/bit.21585
• 9

The process sequence of two-column MCSGP is shown in Figure 2. The MCSGP process is designed by starting with single-column batch chromatography (design chromatogram) using a linear gradient or isocratic conditions to identify and visualize the positions of the product at target purity, impurities, and overlapping sections of product and impurity. The MCSGP process design software executes the process flowsheet by dividing subprocesses into zones according to the presence of product and impurities. Each section of the design chromatogram corresponds to a specific task (zone) of the MCSGP process. I1, B1, I2, and B2 are subprocesses showing the tasks of the two columns (depicted as numbered barrels) in the different zones. Lines between columns show when columns are connected and arrows indicate the direction of flow. After having completed all sequential tasks (I1, B1, I2, and B2 shown in Figure 2), the columns switch positions and the formerly upstream column now becomes the downstream column of a new sequence (I1, B1, I2 and B2). Once the new sequence has been completed with the columns in the opposite order, one cycle is complete, and the first sequence is initiated again (note that Figure 2 only shows the first part of a single MCSGP cycle).

The required product purity can be adjusted by defining the width of the product elution window where predominantly pure product is found and collected. The average residence time of product in the system depends on the ratio of product being eluted and internally recycled. Typically, the average residence time of the product in the MCSGP process is three times larger than the residence time in batch chromatography.

The design of any MCSGP process requires one to initially run a single-column conventional batch chromatography under linear gradient or isocratic conditions to visualize and identify offline the positions of product at target purity and of impurities.5  This initial chromatogram is called the design chromatogram. Using the MCSGP process design software, the design chromatogram is divided into sections according to the presence of product and impurities (Figure 2). The segmentation typically yields one section where only weakly adsorbing impurities (W) are present, followed by a section of weakly adsorbing impurities overlapping with the product (W/P), then followed by a section of pure product (P), and then by a section where the product is overlapping with strongly adsorbing impurities (P/S), and finally a section of only strongly adsorbing impurities (S). All MCSGP process parameters including inline dilution, feed ow rates, and volumes are calculated based on the design chromatogram by the MCSGP design software. The required feed volume is calculated based on replacement of the amount of peptide in the pure product section, which ensures that the overall peptide mass adsorbing onto the columns in each cycle remains constant and the process rapidly reaches a cyclic steady state.

MCSGP Advantages and Productivity Gains

The MCSGP process has significant advantages compared to single-column batch chromatography, including the following.

• Higher yield: Depending on feed purity, product purity specification, and the extent of overlap of product and impurities, a 40%–90% higher yield can be obtained without compromising target purity.
• Higher productivity: Yield improvement and product overloading without process performance decline leads to higher productivity. The process principle allows running steep gradients and applying higher ow rates. Productivity increase can be up to 10-fold higher than with conventional batch processes (Figure 3).
• Lower solvent consumption: The increased yield leads to reduced solvent consumption (the solvent consumption is expressed as liters of solvent consumed per gram of peptide produced). Solvent consumption is typically reduced by 70%, and solvent savings are even more accentuated under conditions of feed material overloading.

MCSGP Peptide Purification Case Studies

In an early case study, MCSGP was used successfully for the purification of calcitonin produced by chemical synthesis using a six-column MCSGP process.7 ^, 8 ^,9   Because of its reduced equipment complexity and its increased operational flexibility, the two-column MCSGP process design has completely replaced six-column MCSGP.

In another case study, MCSGP was used for the purification of a therapeutic peptide from a starting material with a product purity of 66.1%,10  achieving a target purity of 98.7%. MCSGP displayed a yield improvement from 19% (batch) to 94% (MCSGP); a 10-fold increase in productivity, from 3 g/L/h (batch) to 30 g/L/h (MCSGP); and a decrease of the solvent consumption by 70%, from 3.5 L/g (batch) to 1.0 L/g (MCSGP). These improvements had a significant impact on the operating cost and capital expenditure, as will be described in the forthcoming paragraphs.

Recently, MCSGP has been used for the purification of liraglutide, an acylated glucagon-like peptide-1 (GLP-1) agonist having a fatty acid moiety attached to the peptide chain. The study is not publicly available, yet positive results have been reported in the published study abstract.11  The abstract states that, “The implemented twin-column MCSGP process has shown to be able to achieve very high yield (99.6 %), high purity (97.3 %), and very good performance in terms of productivity (0.24 g/L solvent or 5.8 kg/hr·m3 column).” This case study concluded that “the higher the required purity, the more favorable the MCSGP process is with respect to a conventional batch process.” MCSGP has also been used successfully for insulin purification, obtaining a 3-fold higher productivity and 60% reduction in solvent consumption compared to a standard industrial batch insulin purification process (unpublished data).

• 5
• 7
• 8
• 9
• 10Müller-Späth, Thomas, G. Ströhlein, O. Lyngberg, and D. Maclean. “Enabling High Purities and Yields in Therapeutic Peptide Purification Using Multicolumn Countercurrent Solvent Gradient Purification.” Chimica Oggi—Chemistry Today 31, no. 5 (2013): 56–61.
• 11Gomis Fons, J. Chemical Engineering. Lund: Lund University, 2017. https://lup.lub.lu.se/student-papers/search/publication/8903202

Considerations For Scale-Up

Equipment

MCSGP for peptide purification is best operated with HPLC systems having a flowsheet configuration that supports an MCSGP process. Entry-stage systems (Contichrom HPLC by ChromaCon) and scale-up systems (Ecoprime Twin HPLC by YMC) use the optimized twin-column system configuration. For scale-up, the quality of pump performance (being able to perform smooth gradients) is very important.

Dynamic Process Control for MCSGP

A dynamic process control tool, MControl, has been developed for the MCSGP process. MControl is very important for robust MCSGP operation on both benchtop and GMP scale. MControl is capable of adjusting the MCSGP process in response to changes in temperature, solvent composition, and, to some extent, column performance—assuming that changes in these parameters lead to a shift in the chromatogram, yet do not have a significant impact on the resolution of product and impurities.

MControl is capable of adjusting the position of the product elution window (i.e., the start of phases B1, see Figure 2) based on UV thresholds that are reached during the interconnected and batch phases of the MCSGP process. Any peak shift is recognized and the product is collected by UV threshold, ensuring a robust autocorrection, rather than based on a fixed volume or time. For peptide purification, MCSGP is operated with RP chromatography. RP chromatography is very sensitive to temperature change, leading to a shift in the chromatographic elution profile with the risk of collecting incorrect elution fractions. As an example, lowering the temperature of solvents by just a few degrees will lead to a later product elution. MControl will autocorrect this deviation by recognizing the peak shift and thereby simultaneously shifting the product elution window, helping collect the right product fractions. MControl can be also operated with specified delay periods by ignoring an initial elution profile prior to operating with a threshold collection. Figure 4 provides an overview of the principle of MControl use to operate the MCSGP processes in a robust manner with minimal supervision.

MControl is the simplest and fastest type of feedback control presented for MCSGP, as it provides a direct feedback affecting an ongoing peak elution. Other feedback control methods include control based on elution peak retention times,12  feedback control based on at-line HPLC analysis,13  and model predictive control (MPC).12 ^, 13 ^, 14

Process Modeling

Although chromatographic process modeling is not required to design an MCSGP process, it can be used to speed up process development. A chromatographic process model consists of equations describing mass balances, isotherm, and mass transfer that are solved numerically. The model parameters can be determined by fitting tools that automatically calibrate the model parameters based on a number of linear gradient experiments.

The chromatographic model can then be utilized to simulate single and multicolumn runs, without actually having to carry out the runs experimentally, and predicts their performance in terms of yield, purity, productivity, solvent consumption, and product concentration. Moreover, the model can be used to optimize the MCSGP process within user-specified design space borders.

Process Validation

Process validation concepts have been developed for twin-column countercurrent processes. Similarly, as in single-column chromatography, a risk-based approach is used, testing process parameters and multiple sets of operating conditions, corresponding to different steady states to characterize the process and identify critical process parameters. Simulation and optimization software based on a chromatographic model helps reduce the number of design of experiments (DoE) needed to define the operating space, as indicated in the previous section. A summary of these process validation approaches for a twin-column capture process has been presented and could be adapted to twin-column MCSGP to a large extent.15  Simulated moving bed (SMB) chromatography processes using four to eight columns and countercurrent principles are in use for production of chiral molecules and comply with FDA requirements. Thus, the validation for MCSGP using only two columns is considered a feasible task.

MCSGP Economic Analysis

Based on our own case studies, published data, and user feedback, an economic analysis for twin-column MCSGP chromatography of therapeutic peptides has been carried out, including a comparison with existing single-column chromatography. The assumptions are listed next.

Modeling Assumptions

Single-column batch reference processes with an assumed achievable yield of 40%, 50%, 60%, and 70% were modeled and compared to a two-column MCSGP process with an assumed achievable yield of 95%. The different achievable yield in batch chromatography reflects the degree of difficulty in the separation.

We assumed that the longer the peptide, the more impurities accumulate and the lower achievable yield is obtained under constant load due to narrower peak pooling (Table B). MCSGP has been shown to achieve substantially higher yields than batch processes under given purity constraints. We did not model iterative synthesis optimization steps to deplete individual impurities resulting in improved yield. It was assumed that rechromatography was used to recover 25% of the yield loss in batch chromatography. Because of its internal recycling principle, the achievable yield of MCSGP was assumed to be 95% and independent of peptide impurity content or size.

Additional assumptions of the batch reference run and the MCSGP run are reported in Table A. Additional assumptions of the chromatographic process are summarized in Table B. The processes are operated at different linear ow rates of 181 cm/h (batch process) and 271 cm/h (MCSGP process), respectively. A larger flow rate in MCSGP is assumed because the process, due to its internal recycling capabilities, can achieve high product yield at high flow rates without compromising yield and purity. For batch processes, lower flow rates have to be used to obtain better mass transfer and a reasonable product yield. The loads and cycle times are also summarized in Table B.

• 12 a b Grossmann, C., G. Ströhlein, M. Morari, and M. Morbidelli. “Optimizing Model Predictive Control of the Chromatographic Multi-Column Solvent Gradient Purification (MCSGP) Process.” Journal of Process Control 20, no. 5 (June 2010): 618–29. doi:10.1016/j.jprocont.2010.02.013
• 13 a b Krättli, M., F. Steinebach, and M. Morbidelli. “Online Control of the Twin-Column Countercurrent Solvent Gradient Process for Biochromatography.” Journal of Chromatography A 1293 (June7, 2013): 51–9. doi:10.1016/j.chroma.2013.03.069
• 14Papathanasiou, M. M., F. Steinebach, G. Stroehlein, T. Müller-Späth, I. Nascu, R. Oberdieck, M. Morbidelli, A. Mantalaris, and E. N. Pistikopoulos. “A Control Strategy for Periodic Systems— Application to the Twin-Column MCSGP.” In Computer Aided Chemical Engineering, vol.37, edited by K. V. Gernaey, J. K. Huusom, and R. Gani. Elsevier, 2015: 1505–10. doi:10.1016/B978-0-444-63577-8.50096-6
• 15Baur, Daniel, James Angelo, Srinivas Chollangi, Thomas Müller-Späth, Xuankuo Xu, Sanchayita Ghose, Zheng Jian Li, and Massimo Morbidelli. “Model-Assisted Process Characterization and Validation for a Continuous Two-Column Protein A Capture Process.” Biotechnology and Bioengineering (October9, 2018). doi:10.1002/bit.26849

Results

Total Costs Comparison

Using the abovementioned assumptions, the total costs for the production of 10 kg peptide annually was calculated. The results are shown in Figure 5, where total peptide production costs are shown as a function of the yield in batch and MCSGP chromatography. The overall costs, including synthesis and purification, are dominated by the synthesis costs, which represent 80% to 90% of the total costs. As for batch processes, the chromatography yield increases (40%–70%) and the synthesis costs decrease because fewer synthesis batches have to be produced to obtain the targeted production output of 10 kg per year. Because the MCSGP process provides consistently higher yields (95%) independent of peptide length and impurity content, no additional synthesis batches are necessary to obtain the target production amounts.

Although for a chromatography yield of 40% the overall costs are $4.5 million USD, costs decrease with increasing yield, reaching $3.1 million USD for 70% yield. The overall costs for MCSGP are $2.5 million USD, indicating a savings potential of $2.1 million USD with respect to the batch process with 40% yield and $0.6 million USD annually with respect to the batch process with 70% yield.

Downstream Cost Comparison

The downstream processing costs for batch chromatography, excluding the synthesis costs, are dominated by solvent costs and plant operating costs, followed by stationary phase costs and quality assurance/quality control (QA/QC) costs (Figure 6). Equipment costs (capital expenditure [CAPEX]/year with 10-year depreciation) are the smallest cost contributor. The overall downstream processing costs decrease with increasing yield, due to the lower number of additional batches that need to be synthesized to reach the annual production target of 10 kg peptide. As the yield increases from 40% to 70%, the annual downstream processing costs decrease from $760,000 to $520,000 USD for batch purification.

The downstream processing costs for the MCSGP scenario are $280,000 USD, representing at least 40% cost savings compared to the batch chromatography scenarios. The costs of MCSGP-based downstream processing are dominated by the CAPEX, which accounts for 50%. Smaller contributions are from plant operating costs, QA/QC, and solvent costs. For the MCSGP case, stationary phase costs are negligible, contributing around 3% to the downstream processing costs because significantly smaller columns are used (70.7 L (batch) vs. 2x 7.1 L [MCSGP]). Columns are typically repacked annually with 30% stationary phase replacement. Thus, the cost of goods sold (COGS) elements for the batch are different than for MCSGP and the absolute COGS for MCSGP are significantly lower.

Payback Period

MCSGP total cost savings compared to batch chromatography scenarios are between $0.6 million to $2.1 million USD for 10 kg target production (Figure 5). The CAPEX cost difference of MCSGP HPLC systems vs. batch HPLC systems is estimated to be $1 million USD (Table A). Based on the preceding parameters, a payback period can be calculated as follows: Payback = (CAPEX difference batch vs. MCSGP)/(total cost savings by MCSGP). In comparison with batch processes with 40% or 70% yield, the payback period of MCSGP lies between 6 months and 19 months, respectively, as shown in Figure 6. The cost savings through MCSGP are dependent on the annual production amount and the batch yield. Higher target production quantities result in a corresponding shorter payback period, as equipment utilization is increased. See Figure 7 for payback period for investment in MCSGP compared to batch processes with different yields.

Operational Aspects of MCSGP

Besides the capital expense and payback consideration shown previously, MCSGP has benefits with respect to operational aspects, including the following.

• Reduced equipment requirements for pumps, pressure rating
• Reduced column dimensions
• Reduced stationary phase costs
• Reduced solvent consumption
• Overloading with MCSGP possible without performance loss for overlapping product/impurity peaks
• Reduction in QC sampling and testing

In the simulated cases, batch chromatography requires a 60 cm inner diameter column and 25 cm bed height, leading to a total column volume of 70.7 L, whereas MCSGP operates with two columns of 30 cm inner diameter and 10 cm bed height, resulting in a total bed volume of 14.2 L, leading to lower stationary phase costs. The use of smaller columns and resin volumes in MCSGP is possible due to the increased yield of the chromatographic process and to the larger linear flow rates of MCSGP, which strongly reduce the cycle time compared to batch chromatography. A summary of the operational results is provided in Table C. Another factor contributing to cycle time reduction is that during the disconnected states of the process, the columns are operated at half the residence time. Lower column hardware costs and facilitated packing of the smaller columns were not included in the calculations but would be also in favor of MCSGP.

Smaller columns also reduce the required pump dimensions despite the larger linear ow rates that are used. Although an 8.5 L/min (= 510 L/h) pump is required on a batch skid, the MCSGP skid would only require 3.2 L/min (=190 L/h) pumps, resulting in smaller piping and components and a smaller equipment footprint despite using two columns. The higher load and yield of MCSGP leads to reduced solvent consumption in chromatography, measured in liters of solvent used per gram of peptide purified.

The solvent consumption decreases strongly with increasing yield: For MCSGP it is 0.9 L/g, whereas the consumption for a batch process with 40% yield is 6.6 L/g and the consumption for a batch process with 70% yield is 3.6 L/g (see Figure 8). Thus, MCSGP is capable of reducing solvent consumption by up to 85%, corresponding to up to a 56,000 L savings per year for 10 kg peptide produced. Although direct solvent cost savings have been quantified in the preceding cost calculation, indirect costs such as additional supporting infrastructure, solvent preparation, handling, and disposal have not be included. Because of the reduced solvent consumption of MCSGP, the latter factors would have numbers in favor of MCSGP.

MCSGP has a larger number of feed injections per run than batch chromatography but a lower number of QC samples than batch because for each cycle only a single pool is being sampled and analyzed. With batch chromatography, repetitive QC sampling and analysis is required because rechromatography results in more QC sampling. In this study, we assume that 10 analyses are required per single-column batch cycle and one analysis per MCSGP cycle. The smaller number of fractions in MCSGP leads to an overall reduction of QA/QC costs.

A sensitivity analysis was carried out to examine the impact of reduced column size on the downstream costs in dependence of the process/yield. For batch chromatography, the investigated column diameters were 60 cm, 45 cm, and 30 cm, whereas for MCSGP they were 30 cm, 20 cm, and 15 cm. The results are provided in Figure 9 and results show that for all cases (batch 40%–70% yield vs. MCSGP) stationary phase costs decrease as expected, but plant operating costs and QA/ QC cost rise and surpass the cost savings obtained using smaller columns. Smaller columns require larger operating times, as indicated in the figure, because more cycles (injections) need to be performed.

Conclusion

The economic evaluation of twin-column countercurrent chromatography (MCSGP) for the purification of peptides produced by chemical synthesis shows significant cost advantages of MCSGP, with a payback period for MCSGP compared to different batch scenarios of between 6 and 19 months for an annual production of 10 kg of peptide. Different cases of MCSGP with 95% yield and single-column batch processes with 40%–70% yield were compared, simulating purifications of varying difficulty due to variable impurity content or peptide size. Rechromatography was included in the calculations for single-column chromatography. MCSGP does not require rechromatography due to its high yield and has a total cost advantage (mainly by reducing the number of upstream synthesis batches required to reach target production quantity). Other advantages include significant reduction in solvent consumption.

The cost savings through MCSGP vary between $0.6 million and $2.1 million USD, depending on the yield of the single-column reference process. Thereby, the annual downstream processing costs range from $760,000 to $520,000 USD for single-column batch chromatography purification, depending on the yield, and only $208,000 USD for MCSGP, showing a cost savings of at least 40%. The analysis revealed that the use of larger columns was favorable due to the reduction in plant operating time and number of injections, leading to smaller QA/QC effort, which offset the larger stationary phase costs of larger columns. Indirect solvent costs such as additional supporting infrastructure, solvent preparation, handling, and disposal have not been included in the comparison but the numbers would be in favor of MCSGP due to its reduced solvent consumption.

Regulatory authorities are supportive of continuous manufacturing for pharmaceuticals, which also covers continuous chromatography techniques such as MCSGP. Available simulation and optimization tools allow a reduction in the number of designed experiments for defining the operating space and facilitate process validation.

Acknowledgments

Input from and discussions with Dr. Alessandro Butte (Datahow AG, Zurich, Switzerland) and Dr. Ralf Eisenhuth (Bachem AG, Bubendorf, Switzerland) are highly appreciated.