The pressure to bring a new drug to market has increased significantly but the develop time needed has also increased because molecules are more complex to make. Chemists need to compress time by working in parallel on difficult steps. Process optimization is not always completed in time and a less desirable manufacturing processes is used to deliver clinical trials material. Further optimization can happen later, but once filed, a registered process is difficult to change. Therefore, using scalable technologies early in development to reduce the need to redesign the process later is critical. However, technologies are not always available, well understood, or too expensive to be considered.
Chromatography is one of these. This unit operation is used at all development stages but is often discarded as a solution for commercial production.
Chromatography is mostly known and used for being a very powerful analytical tool. All development chemists check reaction completion and impurity profiles using an HPLC method. Every API is tested for purity using a chromatography method (HPLC or UPLC). Residual solvents are also identified and quantified by chromatography (GC). It is a very powerful tool that can separate the most difficult mixtures (enantiomers, peptides, polymers…). It is equally a formidable production tool used in many industries. For example, continuous ion exchange SMB processes are used to purify the sugar we consume (1). If the food industry is using SMB for manufacturing a product with tremendous price pressure, it is because the technology is robust, reliable, scalable, and very efficient, resulting in a very low operating cost at commercial scale. However, when it comes to API manufacturing, process development chemists are reluctant to use chromatography as a long-term solution and will spend a significant amount of development time to remove chromatography from the synthetic route used in early development. This is likely due to a lack of understanding the optimization process required to bring this technology to commercial scale efficiently and cost effectively.
Chromatography is a large family of processes relying on the interaction of solutes with a solid phase. The expectation is that different solutes have different interactions thus providing a separation (2). Typically, the solid phase is packed in a column and the solutes are introduced and eluted forward with a mobile phase (solvents). The differences in interactions result in a difference in residence times and solutes can be recovered individually. Obviously, optimization of multiple parameters is needed (concentration, particle size, surface chemistry…). When used for production, Preparative chromatography (Prep) is optimized to achieve the triple objective: Purity, Yield, and Throughput. While Purity and Yield are somewhat understood, Throughput is challenging as it is the realm of non-linear chromatography (3). Chromatograms look extremely distorted in an apparent chaos. Therefore, understanding the process fundamentals is important to take advantage of overloading effects to optimize the separation (2)(3)(4). Unfortunately, most chemists are not sufficiently trained in Prep optimization and the knowledge is often limited to operation of inefficient low-pressure columns. While it is accepted that a “med-chem” route will never be economical to produce commercial quantities, this kind of chromatography, should not be scaled-up either. The “med-chem” route is redesigned to a scalable route that ultimately provides a cost-effective process. This should be the same for chromatography. Fortunately, to support early development, several equipment manufacturers have developed “easy to use” solutions to replace the glass column by using more advanced pre-packed cartridges and semi-automated injection/collection systems. These devices have provided chemists with a quick and simple purification solution without the required expertise in optimization. However solvent consumption remains high, yield is mediocre, and throughput is limited because the separation is not properly optimized. However, because of the small scale, the column will be oversized to overcompensate for the poor performance. Unfortunately, this contributes to the “bad” reputation of the technology and development chemists work hard to eliminate it.
If the chromatography used at early stage is scaled-up without optimization, massive quantity of solvents is required with limited options for recycling. The Prep needs to be optimized with the rest of the process. While a gradient is useful in analytical chromatography to ensure complete separation of all species in the sample, isocratic elution is preferred in Prep. Converting the gradient to isocratic provides an amazing opportunity for solvent recycling and a shorter cycle time. This means that the same eluent must be used for the feed to simplify the solvent mix to recycle. This can pose some challenges as solubility is an issue that affects the separation throughput. Assessing strongly adsorbed impurities is important as they can affect column performance by gradually degrading the separation capacity. Using isocratic elution also favors solvent composition control using PAT solutions (capacitance probe, Near IR, density, pH…). Unfortunately, not all processes can be converted to isocratic (oligonucleotides, peptides, biologics for example) and solvent usage will remain an issue, but more complex chromatographic separation solutions can be explored (SMB, CSSR, MCSGP, twin columns…).
While reverse phase chromatography (i.e. aqueous mobile phase) is usually preferred for analytical, developing a normal phase process (i.e. mixture of organic solvents) is preferred for Prep. This offers more favorable solvent recycling options and an easier transition to the next synthesis step.
The development of supercritical chromatography (SFC) has contributed to a significant reduction in solvent usage, bringing “greener” processes to early development (5). Unfortunately, large-scale SFC units, for now, are not available as the CAPEX required is high.
Optimizing the load injected is essential to reduce solvent use and increase throughput. Too often separations are “optimized” using a “touching-band” approach. However, pushing the injection to larger loads can provide significant benefits due to displacement effects (3). Also, maximizing the load per injection reduces the size of equipment required (i.e. column) reducing the CAPEX requirements.
Reducing cycle time increases throughput; this is achieved by operating at the highest possible flow rate resulting in high operating pressures.
Typically, Prep columns are rated for 50 bars or more. Back pressure is driven by the viscosity, and the chromatographic media size. The latter is driving the column efficiency. Small particles provide high efficiency but high back pressure. Thus, a compromise between high pressure and efficiency is achieved by selecting the right particle size (2, 3).
Prep has been intensively studied and modelled efficiently for in-silico optimization (3)(4). It hinges on measuring the equilibrium of the various compounds in solution with the solid phase. The mathematical model for this equilibrium called the isotherm, can vary in complexity. Simple models are used for early development, requiring only a small amount of product. The data generated provides an evaluation of the purification cost. Once the process reaches commercialization, more complex, realistic, and expensive models (digital twins) are used. Overall, Prep is very easy to scale-up. The liquid/solid equilibrium must remain the same. This scales with the ratio of the columns surface areas. For example, a separation on a 200 mm diameter column produce 16 times more than the same separation on a 50 mm column as the cross section ratio is 16 (2002/502). Therefore, lab scale modeling and in-silico optimization, provides the data to predict larger scale systems. More advanced chromatography techniques can be explored depending on the separation. For example, a batch chiral separation can be optimized using stacked injections (Figure 1) or the cyclo-jet process (Figure 2) or SMB (Figure 3) (7). With modeling and simulations, one can calculate these processes and obtain the data for cost evaluation purpose.
Finally, a compromise between the cost associated with the equipment and the cycle time is needed. Purification could be done using multiple injections on a small column or fewer injections using a larger column. The latter provides savings in processing time and in QC as fewer injections are tested. However, more media is required, and this could be expensive.
Ultimately, the cost of the final product in $/kg should be considered and compared to other techniques.
CONCLUSION
Developing an efficient synthetic API route is complex and arduous. Many technologies used at small scale are often removed because of unavailability at larger scale or lack of proper optimization. For any technology, phase appropriate optimization should be considered to mitigate development time and cost. Undervalued techniques such as Prep chromatography can be implemented rapidly and economically at all scales providing proper optimization. Using technologies that are easy to scale, like chromatography, can significantly compress development time and provide economical solutions solving the time conundrum for the Pharma industry. These technologies exist and are well understood. Let’s not be afraid to use them.
GLOSSARY
GC: Gas Chromatography
HPLC: High Performance Liquid Chromatography
IEC: Ion Exchange Chromatography
MCSGP: Multicolumn Countercurrent Solvent Gradient Purification
PAT: Process Analytical Technologies
SFC: Supercritical Fluid Chromatography
SMB: Simulated Moving Bed
UPLC: Ultra Performance Liquid Chromatography
Figure 1. Stacked injection chromatography.
Figure 2. Steady state recycling internal profile (Figure from [7]).
Figure 3. SMB internal concentration profile.
REFERENCES AND NOTES
- F. Boon Process innovation in the Sugar Industry: Chromatographic Sugar Separation Using SMB Technology American Institute of Chemical Engineers, 2006 Annual Meeting
- J. Cazes, R. P. W. Scott “Chromatography Theory “ Marcel Dekker, 2002
- G. Guiochon and al. “Fundamentals of Preparative and Nonlinear Chromatography” Academic Press Inc., U.S., 2006
- R. M. Nicoud “Chromatographic Processes; Modeling, Simulation and Design”, Cambridge University Press, 2015
- L. Miller “Use of dichloromethane for preparative supercritical fluid chromatographic enantioseparations” Journal of Chromatography A, 1363 (2014) pp 323-330
- M. Kaspereit and al. “Simplified design of steady-state recycling chromatography under ideal and nonideal conditions” Chemical Engineering Science, Volume 66, Issue 21, 2011, pp 5428-5438
- C. Grill and al. “Resolution of a racemic pharmaceutical intermediate – A comparison of preparative HPLC, steady state recycling, and simulated moving bed” Journal of chromatography. A. 1026. pp101-108.
