SMB Takes On Paclitaxel Purification

Simulated moving bed chromatography, widely used in chiral separations, can also be used to remove structurally similar impurities from target APIs.

By Paul A. Metz, Bioxel Pharma, Inc., and Olivier Dapremont, Aerojet Fine Chemicals LLC

Simulated Moving Bed (SMB) chromatography is an important technology for separating chiral compounds, but it can also be used in challenging purifications to remove impurities that are structurally similar to target compounds. Bioxel Pharma, Inc. has developed and implemented a fully automated, high throughput SMB process to remove cephalomannine from paclitaxel, an active pharmaceutical ingredient used in cancer drugs. The novel purification process has been running for six months at the FDA inspected site of Aerojet Fine Chemicals in Rancho Cordova, Calif.

Paclitaxel and cephalomannine are both natural products isolated from the Canada yew. Removing cephalomannine from paclitaxel is particularly challenging, since the molecular structures of the two compounds are very similar. The overall process, from starting biomass to finished product, is outlined in Figure 1.

Bioxel has been producing paclitaxel since 1998 using single-column batch chromatography to remove cephalomannine. However, as demand for paclitaxel increased, the company sought to increase its production scale. While single-column chromatography was selective and produced a high purity product, it lacked the capacity, throughput and yield to meet growing demands.

To improve its paclitaxel output and production economics, Bioxel recently installed an SMB purification process, using equipment built by Groupe NovaSep (Pompey, France). The process has been fully validated following the ICH Q7A cGMP guidance for active pharmaceutical ingredients. The facility now running at Aerojet Fine Chemicals can produce up to 120 kg/yr of USP-grade cGMP paclitaxel. U.S., Canadian and European Drug Master Files have been obtained for the new process.

The large-scale separation of paclitaxel from cephalomannine poses several challenges, involving tradeoffs in equipment costs, separation efficiency, capacity, product yield and throughput. In addition, because paclitaxel is a potent cellular toxin, and separations involve the use of volatile organic solvents, any process has to be engineered to meet stringent safety and containment requirements.

Thus far, Bioxel’s new SMB process has increased product yield and throughput, removed other trace impurities and has reduced or eliminated undesired solvents. This article will summarize the SMB project, highlighting the quality and economic results that have been achieved with the new process. It further compares these results with those expected using single-column batch chromatography.

The processing steps used to make USP-grade paclitaxel are shown in Figure 1. The API is present in the yew biomass at less than 0.1%. The final product is purified to a concentration of greater than 98.5%. Cephalomannine content in the final product must be less than 0.5%. Because paclitaxel is cytotoxic, the final process must be run in a high containment environment.

The Canada yew is harvested from forest lands throughout eastern Canada. Moisture is removed from the biomass, which is then ground into a powder. What follows is a series of fine chemical processing steps including liquid-solid extraction, solid phase extraction, adsorption and precipitation steps that remove paclitaxel and other taxanes from cellulose fiber, waxes, pigments and other materials in the biomass. These steps produce a concentrated intermediate extract powder containing paclitaxel, cephalomannine and other taxanes with side chain substitution on the C-13 position of the tricyclic baccatin ring system. Most of the processing involves proprietary technology using common engineering unit operations.

Chromatography to Remove Impurities

While a few chemical processing approaches [1-2] have been applied to remove cephalomannine from paclitaxel, chromatography is most commonly used. Several chromatographic approaches have been reported for paclitaxel purification at various process points and at different production scales [3-6].

After working with paclitaxel using single-column batch chromatography for several years, Bioxel found that the best strategy for purification was to push cephalomannine together with paclitaxel as far as possible through the process, favoring high paclitaxel yield along the way. Then, the goal was to identify the most efficient way to separate the unwanted impurity from paclitaxel at the end. The company settled on SMB for final purification following computer modeling and successful small-scale process demonstration runs.

Figure 2 shows the SMB system, which is a continuous, closed-loop process. Paclitaxel intermediate extract is fed through a pre-filter and then into the SMB system. The intermediate is separated into two streams, paclitaxel (extract) and cephalomannine (raffinate). Two automated rotary evaporators are used to recycle the eluent collected in the extract and raffinate lines and to dry the isolated fractions. The unit is operated in the normal phase, with a proprietary column packing material.

The SMB operation is followed by crystallization to remove remaining trace impurities, yielding a paclitaxel product that is greater than 98.5% pure. A very mild high-efficiency filter drying process removes residual solvents prior to packaging. The SMB, crystallizer and filter dryer are housed in a dedicated high containment facility. With low solvent utilization, the SMB is operated in a general-purpose area. The solvent tank farm, crystallization vessel and filter dryer are operated in a Class I, Division I, Group C&D area.

Results Favor SMB

Bioxel carried out feasibility studies to assess the use of SMB on a commercial scale targeted at producing 100 kilograms of pure paclitaxel per year. Data from the studies along with historical chromatography data were used to model and compare an SMB process with a single column approach. Studies included column screening and mass overload experiments that were carried out using analytical HPLC and commercially available preparative chromatography packing materials. Sixteen stationary phases were screened with five organic solvents in varying ratios. System selection was based on the following criteria:

  • Paclitaxel solubility in the mobile phase
  • Paclitaxel and cephalomannine resolution
  • Peak retention and elution order
  • Loading
  • Cost and availability of the stationary phase
  • Product stability
The company then ran column overload studies at commercial scale, in which increasing concentrations of paclitaxel intermediate extract were injected onto a single column with the proposed SMB packing and mobile phase composition. It was found that paclitaxel exhibited a counter intuitive behavior in which its retention actually increased with increasing injection mass. This is known as “Anti-Langmuir” behavior [7]. Cephalomannine, on the other hand, showed a typical linear isotherm in which an increase in loading led to a decrease in retention time. This difference in retention phenomena could be exploited since it would result in an increase in selectivity with greater mass loaded onto the column. That is, a higher SMB feed flow would increase production rate without sacrificing purity, since selectivity is increased as well.

With the screening and mass overload data, an SMB process was modeled. This involved the use of Varicol, a patented process variation of SMB in which the injection and collection points are moved asynchronously across the ring of chromatographic columns. A six-column SMB configuration was established with solvent flow rates, operating pressures, switching times, feed concentration and feed flow rate. The model was used to project the economics of a large-scale SMB separation process, and to compare this to the expected output from a single column batch chromatography process at the same scale.

Before the commercial scale SMB was built, demonstration runs were carried out on a laboratory unit with columns that were 1/5 scale compared to the proposed process. Initial conditions for SMB operation were confirmed and further optimized. Stationary phase packing lifetime studies were carried out and critical operating parameters and ranges were identified. Approximately 500 grams of intermediate paclitaxel extract were processed. The ability of SMB to separate paclitaxel from cephalomannine was confirmed. In addition, operating data were collected that could be used to make an economic evaluation of the process. The data were further used to design the large-scale SMB equipment and to set the initial process conditions for SMB start up.

Table 1 lists the projected cost comparison of SMB to single column batch chromatography (SCBC). The relative results were obtained from SMB modeling studies and historical batch chromatography data.

While SMB involves higher up-front capital costs than the single column batch chromatography system, its ongoing operating costs are lower. The most notable savings are in the projected labor rate, which is reduced by 78%, and solvent utilization, which is cut by 68%. Since the SMB process is automated and continuous, it requires fewer personnel for normal operation.  Solvent utilization is much lower, and the solvent that is used is continuously recycled. While energy cost is projected to be lower with the SMB system, the absolute savings are not significant compared with labor and solvent savings.

With the capital investment made and six months of full scale SMB operation complete, the economic projections have been largely corroborated by full-scale process results. However, labor costs have been higher than projected. This is a result of the high level of effort involved in system start-up and process validation. On an ongoing basis, the labor costs are expected to approach the model estimates. Because the stationary phase has not yet been replaced, it is difficult to project the ongoing cost for this replacement item. Solvent costs are in line with estimates. However, solvent utilization rates are very low, making solvent recycling unnecessary.

Table 2 lists a comparison of SMB and single column batch chromatography with respect to product throughput, yield and productivity. The comparison is based on the SMB simulation, actual large-scale process results, and historical pilot scale batch chromatography processing data. For about the same level of product purity, SMB provides a 15% improvement in yield and a projected four-fold improvement in productivity. To put this in perspective, the yield improvement alone represents a processing cost reduction of $45,000 per kilogram of pure paclitaxel.

Yield and throughput results for SMB have been corroborated by empirical process data at full scale. Although a batch chromatography process was not scaled up or optimized, specific productivity would not be expected to improve significantly considering known solubility, loading and selectivity limitation for paclitaxel and cephalomannine.

SMB in an Integrated Process

The SMB simulation model and SMB demonstration runs were used to design the SMB system and provide a starting point for scale-up and process optimization. But the SMB was only one part in an integrated process. The SMB separation was optimized in conjunction with a pre-SMB filtration step, final crystallization and filter drying as an integrated process to meet final product purity requirements. The quality requirements are defined in the U.S. Pharmacopeial Convention (USP 27-NF22). The combined optimization studies led to modifications in the amount of stationary phase packed in the SMB columns, along with solvent composition, loading concentration, feed and take-off rates.

The SMB process is controlled through real-time, near-line, fast HPLC analysis of the extract and raffinate evaporate streams. Near-line data allows for quick start-up times and ultimately assures SMB output purity levels.

While the focus of the SMB was to separate paclitaxel from cephalomannine, the process provided the additional benefit of reducing other minor taxane impurities. Table 3 lists 16 taxane impurities and their input and output concentrations following the SMB process. The impurities listed are only those present in the starting paclitaxel feed (input) and that are influenced by the SMB process. Also listed in Table 3 are the specification concentrations and the percent reduction realized for each impurity following the SMB process. Results in Table 3 reflect the average of seven process engineering runs carried out at full scale.

Cephalomannine content in the starting intermediate extract (input) ranged from 3.5% to 4.9% and averaged 3.8%. As expected, the SMB process reduced cephalomannine by 95% to an average of 170 ppm, lower than the 500 ppm specification limit. It also lowered the amount of other taxane impurities.

For example, one other significant impurity, Taxol C, was chromatographically retained with cephalomannine and was reduced by 93%. Reducing these impurities in the SMB step allowed for increased flexibility and higher yield from the final crystallization. The only exception was for the impurity 7-epipaclitaxel. Epimerization of paclitaxel at the 7-position is well known, catalyzed primarily by heat. Although 7-epipaclitaxel levels remain well within the product specification, further development is underway to identify the source of the epimerization and to control it.

For separation of paclitaxel and cephalomannine, both computer models and empirical process data show that the SMB process is higher yielding and provides greater throughput than single column batch chromatography. In addition to achieving the desired separation of paclitaxel and cephalomannine, the SMB process removes other trace impurities and eliminates the use of undesired solvents. Application of SMB to the removal of structurally similar taxane impurities from paclitaxel has allowed the company to achieve both its product quality and process economic goals.

SMB Basics

SMB is a type of multi-column chromatography that has been applied to difficult separations of chiral compounds [8]. The basis for SMB is to simulate a counter-current contact between a solid stationary phase and a mobile liquid phase. Unlike single column batch chromatography, SMB is a continuous process in which the mixture to be separated is injected onto a ring of chromatographic columns at rotating points between the columns. At the same time, streams are withdrawn from the ring at rotating points, simulating the movement of the stationary bed. SMB has been applied to pharmaceutical separations [9-10] and is now in commercial use in FDA approved processes [11].

Varicol is the trade name for the patented process variation of SMB in which the injection and collection points are moved asynchronously across the ring of chromatographic columns. Normal SMB has four zones. The counter-current movement of the solid phase with respect to the liquid mobile phase is simulated by the periodic and simultaneous switching of the product feed and mobile phase make-up lines (inlets) and the extract and raffinate lines (outlets).

In SMB, the number of columns in each zone is constant and is an integer. Varicol uses an asynchronous shift of the inlet and outlet lines. In this way, the column distribution between zones varies with time, allowing for optimization of column distribution. Thus, the number of columns in each zone does not have to be an integer. This design promotes even more efficient use of the solid and liquid mobile phases as compared to SMB.

This results in lower operating costs. The unit is fully automated and computer controlled using validated software developed according to GAMP4 guidelines.


References
  1. Pandey, R.C. and L. Yankov. "Isolation and Purification of Paclitaxel from Organic Matter Containing Paclitaxel, Cephalomannine and Other Related Taxanes." U.S. Patent #5,654,448 (1997).

  2. Murray, C., Ziebarth, T., Beckvermit, J., “Oxidation of Cephalomannine with Ozone in the Presence of Taxol.” U.S. Patent # 5,334,732 (1994).

  3. Rao, K. V. “Method for the isolation and purification of Taxol and its natural analogues.” U.S. Patent # 5,670,673 (1997).

  4. Carver, David R., et al. “Method of using ion exchange media to increase taxane yields.” U.S. Patent # 5,281,727 (1992)

  5. Page, M., Perron, M. “Method for the purification of taxanes.” US Patent # 6,002,025 (1999)

  6. Rao, K.V. et. al. “A New Large Scale Process for Taxol and Related Taxanes from Taxus Brevifolia.” Pharm Res. 12(7) (1995) 1003-1010.

  7. Seidel-Morgenstern, A. “Experimental Determination of Single Solute and Competitive Adsorption Isotherms.” J. Chrom. A, 1037, 255-272 (2002).

  8. McCoy, M. "SMB Emerges as Chiral Technique." C&E News. 78(25) (2000).

  9. Francotte, E.R. “Application of Simulated Moving Bed Chromatography to the Separation of the Enantiomers of Chiral Drugs.” J. Chromatography A, 796 (1997) 101-107.

  10. Miller, L. et. al. “Chromatographic Resolution of the Enantiomers of a Pharmaceutical Intermediate from the Milligram to the Kilogram Scale.” J. Chromatography A, 849 (1999) 309-317.

  11. “Chiral Separations are Enduring Item in the Toolbox." C&E News. 81(18) (2003).

About the Authors

Paul Metz is president of Bioxel International, a division of Bioxel Pharma, Inc. (Lafayette, Colo.) Mr. Metz has a B.S. in Biology and Chemistry from Susquehanna University (Selinsgrove, Penn.) and an M.S. in Chemistry from Bucknell University (Lewisburg, Penn.). Before joining Bioxel, he held positions in pharmaceutical development, engineering, manufacturing and management at Merck & Co., Lonza, Inc. and Hauser, Inc. He can be reached via email at pmetz53575@aol.com.

Dr. Olivier Dapremont is manager of SMB Operations at Aerojet Fine Chemicals. He received his M.S.Ch.E. from the University of Toulouse, France, and his Ph.D. degree in Chemical Engineering and Applied Chemistry in SMB technology and chiral applications in 1997 from the University of Pierre and Marie Curie in Paris. He has worked for Prochrom R&D (Champigneulles, France) and Chiral Technologies, and joined Aerojet Fine Chemicals in 2001, where he has developed and scaled-up for production several SMB separations. He can be reached via email at Olivier.dapremont@aerojet.com.

Editor's Note: To view the graphics that accompany this article, please click the "Download Now" button below.

Show Comments
Hide Comments

Join the discussion

We welcome your thoughtful comments.
All comments will display your user name.

Want to participate in the discussion?

Register for free

Log in for complete access.

Comments

No one has commented on this page yet.

RSS feed for comments on this page | RSS feed for all comments