In the Loop: Continuous Chromatography for Chiral (and Other) Separations

Multicolumn continuous chromatography defies most of the conventional wisdom about commercial-scale chromatographic separation.

Conventional wisdom says process chromatography is expensive. It is necessary for biomolecular separations, but should be avoided in small-molecule manufacturing. It is inherently a batch process, usually requiring multiple injections to process a single reactor-load of product, and it consumes vast amounts of solvent and packing material.

Everything conventional wisdom says about process-scale chromatographic separations is wrong—or at least requires a lengthy footnote to explain the exceptions. And high on the list of exceptions is multicolumn continuous chromatography, the category that includes simulated moving bed (SMB) chromatography and its descendants.

From its origins in the 1950s as a technique for chiral separations of commodity chemicals, continuous chromatography has come into its own in the pharmaceutical world over the past four or five years, as single-enantiomer active pharmaceutical ingredients (APIs) have come to dominate the roster of top-selling drugs. Of 2005's seven top-selling formulations, six are single-enantiomer products (1).

The largest multicolumn continuous chromatography unit in service in North America: the 5 x 1000-mm Novasep Process system which went into service in March 2006 at Ampac Fine Chemicals' Rancho Cordova, California facility.

Tortoises and hares

Continuous chromatography links multiple columns—typically five to eight of them—in a loop, outlet to inlet, with additional inlets for raw product feed and two outlets, one for target product (extract) and the other for discarded product (raffinate). To make the process continuous, the feed, extract, and raffinate ports cycle from column to column, lagging behind the solvent flow. The effect is as though the stationary phase were moving backward relative to the feed port, while the solvent is moving forward. If the stationary phase, mobile phase, and flow rate are chosen properly, the viewer straddling the feed port sees the product stream divide into two continuous streams, with one component flowing continuously downstream and the other flowing continuously upstream (Figure 1).

Figure 1: Operating principles of a four-column continuous chromatography system. Trying to follow the inlet and outlet ports as they cycle around the system (facing page) can be confusing. By focusing on the view from the feed port (right), it becomes apparent that one fraction moves upstream from the feed to the extract port, and the other moves downstream. In this view, switching the feed and outlet ports around the loop makes it appear that the columns "Shift left" as noted.

Here is the standard introductory analogy: Imagine a conveyor belt a few yards long, with the belt moving from right to left at walking speed. There's a wolf chained near the left end of the conveyor belt, which guarantees that any animal loaded onto the middle of the belt will head to the right just as fast as it can. Onto the middle of the belt, an assistant (already fugitive from a warrant issued by the ASPCA) loads tortoises and hares at random. Each of these flees to the right as soon as it catches sight of the wolf. The hares, which run considerably faster than the conveyor belt, all escape off the right end of the conveyor. The hapless tortoises are deposited in a heap at the left end (where, fortunately, the wolf's chain is too short for him to reach them, and they, too, manage to scramble to safety).

Stationary phase selection

The chiral stationary phase (CSP) is the heart of the separation, whether conventional or continuous. CSP products developed by Daicel Chemical Industries (Osaka, Japan, www.daicel.co.jp) are silica beads coated with cellulose or amylose, generally modified with achiral functional groups. As Geoffrey B. Cox explains, the functional groups serve to draw the target molecules into chiral grooves in the polysaccharide backbone, where interactions between the chiral target and the chiral sugars are strong enough to separate the right hand from the left. Cox is vice-president and general manager for separation solutions at Chiral Technologies, Inc. (West Chester, PA, www.chiraltech.com), a subsidiary of Daicel that supplies CSPs and services for chiral separations.

The HPLC laboratory. Precise method development is the foundation for continuous chromatography.

Chirbase (chirbase.u-3mrs.fr) reports more than 50,000 chiral separations of 15,000 molecular species using more than 1000 chiral stationary phases, of which about 200 are commercially available. This information repository was founded in 1993 by a group at the University Aix-Marseille III (France) to consolidate information on chiral separations and stationary phases (2). The database offers structural, bibliographic, and chromatographic data for HPLC, GC, and SFC methods.

Yet there are no reliable rules to guide CSP selection, Cox says. "It's purely screening," he says, and a significant part of the method development consists of running samples through bristling manifolds to dozens of small columns from a CSP library. One selects the medium that performs best, and optimizes from there. And optimization is important. CSP medium can typically cost from $5000–20,000 per kilogram, so it is key to select the optimum medium and make the most efficient possible use of it.

Ampac Fine Chemicals 6 x 800 mm production SMB system, one of three systems installed during 1999–2000.

Efficiency and experience

Novasep Process (Boothwyn, PA), a subsidiary of Groupe Novasep (Pompey, France, www.novasep.com), designs, builds, operates, and sells specialty separation equipment, offering liquid chromatography, multicolumn continuous chromatography (both SMB and the company's proprietary modification, the Varicol process[3]), supercritical fluid chromatography, evaporation, and crystallization—including most of the commercial-scale multicolumn continuous chromatography systems used in the pharmaceutical industry today.

Conventional chromatographic separations use only a small fraction of the column bed for separation. In a process separation, with defined inputs, the chromatographer can increase throughput by running multiple injections through the column, properly spaced. By stacking several injections on the column at the same time, says Novasep technical sales manager Willy Hauck, the process operator can boost the total bed volume actively used in the separation to about 10%, reducing the amount of stationary phase and time required. "It takes a lot of convincing," Hauck says, "but we know it will always produce the right peak at the right time."

Shifting to multicolumn SMB chromatography can markedly increase the active volume, up to 50–70% of the total bed volume, with corresponding improvements in productivity and reductions in solvent consumption.

And, according to Hauck, the proprietary Varicol process (with computer-controlled valve manifolds that dynamically control feed and outlet valves to switch columns into the train where they are most needed) can reduce the number of columns needed and make use of 70–85% of total packing material for separation.

In addition, the closed purification loops of continuous chromatography systems are well suited to recovering and reusing solvent—with recovery rates running above 99%—further reducing up-front solvent costs and back-end environmental impact and disposal costs.

These efficiencies in medium and solvent usage help drive unit separation costs down as volume increases. At the pinnacle of the industrial scale, sugar processors use SMB to purify thousands of tons of fructose for pennies per kilogram—a level of economy necessary for a product that sells for just a few dollars per kilogram.

Operating life and lifetime cost

Feedstocks for chiral separations are relatively clean, which can help extend the useful life of the packing material. For small-molecule processes, says Novasep sales and marketing manager Dan Paradis, engineers originally specified a minimum two-year duty cycle for each packing. In practice, though, he's seen service lives of four and five years in pharmaceutical processes. In other circumstances—for example, where plants campaign multiple products—it might not be unusual to change every six months.

Cox cites calculations for a drug manufacturer who needed 90,000 kg of a chiral intermediate annually. The initial (non-SMB) chromatographic process yielded only about 0.2 kg of pure enantiomer daily per kilogram of chiral stationary phase, with poor (less than 10 mg/mL) solubility. The production target was 1.0 kg (or more) of pure enantiomer per kg of CSP, with solubility of at least 50 mg/mL. After optimization, Chiral Technologies developed an SMB process with six 80-cm-diameter columns, holding a total of 180 kg of chiral medium. This optimized system produced 1.5 kg of pure enantiomer per kg of medium daily, for a total output of 270 kg chirally pure product per day, reaching the annual target of 90,000 kg in 330 days of production.

At $14,000 per kg, the CSP cost $2.5 million, but the columns have a minimum three-year working life (for 270,000 kg of product) in this dedicated process. All told, the medium contributed $9/kg ($0.009/g) to the cost of the intermediate. Capital costs for installing the SMB system and the associated solvent recovery system totaled about $10 million (but would be closer to $15 million today), with a seven-year depreciation. During those seven years, the system would produce 630,000 kg of enantiomer, contributing $16/kg ($0.016/g) to the cost. So, in the final analysis, chromatographic separation added $0.025 to the cost of each gram of product produced. Compare that with bioseparation costs of $1–2 per gram of product for Protein A medium and $20–30 per gram for buffer in a monoclonal antibody purification (4).

Asset Management

In March, AMPAC Fine Chemicals (AFC, Rancho Cordova, CA, www.ampacfinechemicals.com, formerly Aerojet Fine Chemicals)—a contract manufacturing and fine chemicals company with deep experience in multicolumn chromatography—started production of an approved therapeutic using North America's largest pharmaceutical SMB unit, installed at its plant on Highway 50 outside Sacramento. The Novasep-built system (AFC's sixth multicolumn continuous chromatography system) has five 1000-mm diameter by 10-cm tall columns packed with about 200 kg of chiral stationary-phase medium.

The system is currently dedicated to commercial-scale production of a single, approved (but unspecified) therapeutic. It went into production in March 2006, going from ground-breaking to the first drop of product in nine months, including three weeks for validation.

AFC's 5 X 1000-m Varicol system typically runs 300 days a year, leaving a comfortable margin for calibration and scheduled maintenance. Overall, its on-stream factor (time in service/time scheduled to be in service) is greater than 95%. In that time, it produces more than 100,000 kg of near-final API.

AFC campaigns a smaller, 200-mm pilot-scale unit, running one to three months between product changeovers. At changeover, operators must remove and replace the 14 kg of chiral stationary phase, a not-inconsiderable expense. The used medium can be saved however, to be cleaned, regenerated, and re-used when the same product is again purified on the system.

Relative Advantages

In another analysis, Michel Hamende summarized the experiences of UCB Pharma (Brussels, Belgium, www.ucbpharma.com) in optimizing production of a chirally pure therapeutic (5). The project team analyzed three principal production strategies—recrystallization, chromatographic separation, and asymmetric synthesis from chiral precursors—and compared manufacturing costs, investments, environmental impact, and "complexity of industrialization," concluding that the data were "clearly in favour of chromatographic resolution, mainly in terms of manufacturing costs and environmental impact" (Table I).

Table I: Comparison of synthetic routes-cost, investment, environmental impact, and complexity (Hamende, 2005).

The analysis showed that the key process parameters remained constant over a 400-fold scale-up volume range. And, the closed-loop system allowed operators to recover 99.7–99.8% of the solvents tested, losing just 0.36 to 0.68 L of solvent per kilogram of product produced.

Separate early in the process stream

To make the best use of both SMB and production resources, look for an opportunity to make a chiral separation before the final product, say the Chiral Technologies engineers. "It's just a slight exaggeration to say that most APIs aren't soluble in anything," says Cox. At an earlier synthetic step, though, the essential chiral structures may be in place in a soluble compound. A chiral purification here can produce substantial benefits by making the most of subsequent reactions and reducing later-stage purification demands.

Cox gives the example of a poorly soluble (less that 20 g/L) API yielding low selectivity and low productivity (less than 0.3 kg per kilogram of chiral separation medium daily). The solution was to move to the penultimate step in the synthesis, where the intermediate compound was well dissolved (better than 80 mg/mL) and medium optimization could push productivity to about 2 kg of product per day per kilogram of chiral medium. Perhaps more important, using SMB allowed the process developers to produce clinical trial materials within three years of starting the project; waiting for a more elegant synthetic route could have added years to the development time.

Despite progress toward asymmetric chiral syntheses, most of the challenges that reach separation engineers are 50/50 racemates. If the product has been fully synthesized, that means that half of the investment has to be disposed of or re-racemized. Intermediate intervention—either chiral separations or asymmetric steps—means more target product in the final feed stream, decreasing both waste and the required column dimensions.

Separate early in product development

"If you can pull it off," says Firoz D. Antia, PhD, "the trick is to use SMB early in development, to get material for preclinical and clinical testing. That gives the chemists time to develop more elegant asymmetric chiral syntheses, if they are practical—though SMB does have the potential for being an economical manufacturing route."

Antia is executive director of chemical and analytical development at Palatin Technologies (Cranbury, NJ, www.palatin.com), a small biotech and pharmaceutical company—and a long-time observer of and advocate for multicolumn chromatographic methods.

"You can concentrate a lot of effort in tweaking the process to try to get those last few percent, which can take a lot of time and money. You can save a lot of time and money, especially in the early stages, if you just do a chiral separation," says Cox. Finding a synthetic route can take years, whereas, "Within a few days, you can develop a chromatographic process that will give you material for testing."

In most cases, the product's developers have already done the basic work necessary to support a chromatographic process separation: Developers have almost always put together an HPLC analytical method proving product identity and purity. Olivier Dapremont, AFC's director of chromatographic separations, points out that these analytical methods are, in fact, well-worked-out separations that could potentially be scaled up to process scale. "They already have a very efficient method for purifying the product," Dapremont says, "but they don't necessarily realize that."

Still Climbing the Adoption Curve

Antia notes two common sources of resistance to incorporating SMB into a commercial process. "People avoid chromatographic steps because of the misperception that they are expensive and consume large volumes of solvent," Antia says. There is also an esthetic objection. "To the synthetic chemist, the engineering approach may smack of lack of elegance. They would much rather develop an asymmetric synthesis. But that takes time and effort that you don't always have."

It took years for pharmaceutical manufacturers to adopt its first commercial SMB process.

UCB Pharma in Belgium became the first company to use SMB to produce a pharmaceutical intermediate, in 1997. FDA inspected its first commercial-scale SMB installation, a Honeywell intermediates process in Arklow, Ireland, in 2000. Today, there are at least seven commercial-scale multicolumn continuous chromatography processes in operation (Table II).

Table II: Multicolumn Continuous Chromatography Processes

As Dapremont sees it, the pharma industry is still in the early stages of understanding and adopting SMB as a production technique, though activity has accelerated over the past three or four years.

"People in production like to use things they've used for a long time, but the momentum is coming and more large-pharma companies are looking at this process."

Most potential users know something about the options for continuous chromatography, says Paradis. Awareness is about fifty percent, but adoption is still close to the base of the growth S-curve. Education is still required to help potential users understand where making the commitment might pay off—especially in the early stages.

Highly efficient solvent recovery, documented at up to 99.8%, is a hallmark feature of closed-loop continuous chromatography systems.

"Some of the preparations we've worked with already had synthetic routes, and the developers were prepared to go with synthetic routes," says Paradis. "But they had yield and solvent issues. When they look at all of that, the separation route is often competitive right down to final production. We've had clients who had achieved synthetic routes, but chose a separation based process because it was more cost effective."

Beyond chirality

Dapremont emphasizes that multicolumn continuous chromatography has applications beyond chiral separations. The plant's 6 X 75-mm SMB system is dedicated to the company's cytotoxics production, manufacturing chemotherapeutics. It is used to purify paclitaxel (Taxol) from partially purified extracts of Canadian yew for Bioxel Pharma (Sainte-Foy, Québec, Canada, www.bioxelpharma.com).

These are complex mixtures of closely related compounds, mixtures that change according to the season in which the plant material is harvested. Dapremont calls this a typical case for chromatography, which can be optimized to deliver any two out of the list of three desirable performance characteristics: purity, yield, and speed. In this case, speed is less important than purity and yield.

Here, process chromatographers modify mobile- and stationary-phase conditions so that the complex separation starts to look like a binary separation, with the paclitaxel (along with a few easily removed impurities) in one peak and all of the other components clustered into another peak.

Fast chromatography for generics method development?

SMB can also help separate pure monomeric product from di- and tri-mers. With other approaches, Dapremont says, "You can lose 20 percent of your product to remove one percent of impurity." With this kind of process separation, the starting material is not the 50/50 mixture common in chiral separations, but rather 95% or more drug compound. With some sacrifice in speed, chromatographic processing removes the unwanted fractions while retaining almost all of the desired species.

Affinity and ion-exchange methods, already running in the food processing industry, may come to biopharmaceutical processing next: as Novasep COO Rachel DeLuca puts it, "The next step is biopharma. It is accepted for small molecules, so now we attack the biopharma market."

References

1. P. Van Arnum, "Single-Enantiomer Drugs Drive Advances in Asymmetric Synthesis," Pharm. Technol. 29 (4), 58–66 (2006).

2. B. Koppenhoefer et al., "CHIRBASE, A Graphical Molecular Database on the Separation of Enantiomers by Liquid-, Supercritical Fluid-, and Gas Chromatography." Chirality 5 (4), 213–219 (1993).

3. O. Ludemann-Hombourger, R.M. Nicoud, and M. Bailly, "The 'Varicol' Process: A New Multicolumn Continuous Chromatographic Process," Separat. Sci. Technol. 35 (12), 1829–1862 (2000).

4. D. McCormick, "Bioseparations Look Ahead to the Past," Pharm.Technol. 29 (7), 36–44 (2005).

5. M. Hamende, "Case study in production-scale multicolumn continuous chromatography," in Preparative Enantioselective Chromatography, G. Cox, Ed., (Blackwell Publishing, Ames, IA, 2005) pp. 253–276.

6. M. McCoy, "SMB Emerges as Chiral Technique," Chem. Eng. News 78 (25), 17–19 (2000).

7. Ibid.

8. Company Data

9. G. J. Quallich, "Development of the commercial process for Zoloft/sertraline," Chirality 17, S120–126 (2005).

10. M. McCoy, "Divide and Conquer," Chem. Eng. News 82 (41), (2004).

11. The Presidential Green Chemistry Challenge Awards Program, Summary of 2005 Award Entries and Recipients, US Environmental Protection Agency, EPA744-R-05-002, June 2005, www.epa.gov/greenchemistry.