Trial Separations

A significant number of major drug products are chiral, and many more such compounds are being brought forward in R&D. Chromatographic separations are the exception rather than the rule in producing chiral pharmaceuticals, with chiral pool, asymmetric synthesis, crystallization, and other resolution methods seeing greater use. But chromatography is coming to the forefront at those times during drug development when it's critical to have separate enantiomers.

Regulatory concerns, spurred in part by the technological advances that enable separation, have necessitated the isolation and testing of the enantiomers of potential chiral drugs. A single enantiomer may, in fact, be found to be the most effective and safest drug candidate. To zero in on the answers quickly, many pharmaceutical company researchers have turned to chiral chromatography to rapidly produce high-purity enantiomers from racemic mixtures.

Most recently, interest has expanded in using supercritical fluid chromatography (SFC) with chiral stationary phases (CSPs). According to suppliers and end users, most major drug firms have owned SFC equipment for some time and now are talking a bit more about applications. SFC has also moved from the analytical scale, simply for identifying enantiomers, to the semipreparative or preparative scales to produce between a few milligrams and hundreds of grams of material.

"From a very practical perspective, SFC is bringing a new level to drug discovery and development," explains Yingru Zhang, senior researcher in discovery analytical sciences at Bristol-Myers Squibb's Pharmaceutical Research Institute. "The ability to obtain enantiomers for a wide variety of chiral molecules quickly for pharmacological and toxicological testing is extremely important.

"What's crucial is getting the material, getting the tests done, and moving on, even at a relatively higher cost, because what matters is the time and the purity of the enantiomers," Zhang adds. Meanwhile, as work proceeds, the decision can be made whether to invest the time and money in pursuing an asymmetric synthesis or other routes to a desired enantiomer, or to separate the higher value products from possibly less expensive racemates.

SFC does not necessarily provide an advantage in enantioselectivity compared with other liquid chromatography (LC) methods, points out Karen W. Phinney, research chemist at the National Institute of Standards & Technology, in a recent review article (Anal. Bioanal. Chem. 2005, 382, 639).

But the reduced viscosities and increased diffusivities of supercritical fluids, compared with typical organic liquids, allow for higher flow rates, rapid column equilibration, and faster resolution. As a result, SFC using columns half the size or smaller than in LC is often many times more productive. This reduces solvent use and waste, speeds up method development, and makes large-scale separations more feasible.

A major benefit of SFC is its simple mobile-phase composition and the chance to replace solvents such as hexane with inexpensive, readily available, less flammable, and low-toxicity mobile phases, Phinney tells C&EN. The nomenclature is often loosely applied, but what is generally called SFC employs carbon dioxide above or near its critical temperature of 31 °C and pressure of 73 bar, combined with an organic modifier, such as methanol or ethanol.

At concentrations typically much less than 50% by volume, the modifier increases the solvent strength of the system and elutes most compounds. Modifier choice can alter retention times, selectivity, and resolution, Phinney explains. Small amounts of polar substances are sometimes incorporated as additives to help elution. Columns can contain one of many possible CSPs to achieve the desired enantioselectivity. All these factors can be tweaked to find the best separation conditions.

"IT WASN'T until about a decade ago that commercial SFC equipment similar enough in operation to LC instruments became available and made the transition easy," Phinney says. The equipment could even use many existing CSPs. With SFC, after reducing the pressure and vaporizing the eluent, compounds are collected in small volumes of solvent (the modifier). This is unlike LC, where very large amounts of solvent must be removed and then disposed of or recycled. "So SFC for chiral separations really started to take off," she adds. "It was just the perfect application and what the chiral community had been waiting for."

Many instrument manufacturers have entered and exited the market over the past 20 years, says Terry A. Berger, who has long championed using SFC for pharmaceutical applications. In 2000, Mettler Toledo bought Berger Instruments, a company formed in 1995 when Berger and partners bought out the SFC business he helped build at Hewlett-Packard.

In February, Mettler introduced the Berger SFC Multigram III, a preparative-scale system targeting higher sample throughput. Using 2- to 5-cm-diameter columns and flow rates of 200 mL per minute, the system is five to 10 times faster than comparable high-performance LC (HPLC) setups, Mettler says.

Other producers of preparative-scale systems include Thar Technologies and Novasep. In January, Thar launched its SuperDiscovery FC system, which combines analytical and semipreparative-scale features. The company offers three other preparative-scale systems; the largest is capable of purifying up to 1.5 kg of material.

Chiral chromatography, especially SFC, meets the demands being put upon it by drug researchers, Zhang says. "Not too long ago, chiral separation was arguably the most difficult of chromatography challenges. But thanks to the dramatic advances in CSPs, almost all chiral molecules can be resolved."

CSPs--supplied by companies including Chiral Technologies, Eka Chemicals, and Regis Technologies--are based on materials that induce specific stereoselective interactions and preferentially bind different enantiomers. The most commonly used materials include Pirkle-type, derivatized polysaccharide, macrocyclic (such as cyclodextrins, glycopeptides, and crown ethers), ligand-exchange, protein, and other polymer-based CSPs.

According to an analysis by Eric R. Francotte of Novartis, an estimated 1,300 CSPs have been prepared and more than 200 are being sold. After reviewing about 1,000 racemic separations, he also found, and reported at the Chiral Europe 2004 meeting, that about 90% of the mixtures could be separated by four CSPs: the cellulose derivatives Chiralcel OD and Chiralcel OJ and the amylose derivatives Chiralpak AD and Chiralpak AS--all made by Chiral Technologies' parent, Daicel Chemical Industries.

Earlier this year, Chiral Technologies launched Chiralpak IA and Chiralpak IB, which extend the coated polysaccharide technology by immobilizing the chiral selector on the silica support. "One of the drivers was to have something that was going to be more robust," says Rodger W. Stringham, director of technology at Chiral Technologies. "The other is that now you can use a wider range of solvents, which gives separations that couldn't be done before."

The company just recently licensed another CSP technology based on quinine and quinidine derivatives from the lab of Wolfgang Lindner, professor of chemistry at the University of Vienna. "These derivatives possess a tertiary amine in a binding cleft along with additional hydrogen-bonding sites," Stringham explains. These features will make possible high-resolution separations of molecules with carboxylic, phosphonic, phosphoric, or sulfonic acid groups.

Unfortunately, chiral HPLC or SFC separations are not predictable. Even structurally related compounds can require completely different CSP and solvent combinations. Choosing the proper combination of CSP and SFC solvent modifier, for example, is largely a hit-or-miss proposition. Researchers resort to past experience and automated screening.

Fortuitously, SFC's rapid flow rate and fast equilibration make it extremely suitable for such approaches. "Fast screening is key to any method development for purification," says Craig White, senior researcher in analytical technologies at Eli Lilly & Co. "It is very much trial and error, but a high success rate can be achieved if the right choice of column and solvents is tried." Not surprisingly, many researchers start with the four columns from Chiral Technologies.

White's lab has developed a rapid SFC screen using short 10-cm-long columns, high flow rates, and fast gradients (J. Chromatogr. A 2005, 1074, 163). An initial assessment using a methanol modifier and the four columns is run serially, followed by screens using ethanol and isopropanol. One racemate can be analyzed this way in 80 minutes, making it possible to screen several samples overnight and then scale up the results the next day to preparative scale.

SPEED WAS the main driver behind Lilly's investment in SFC. "The added benefits were reduced solvent handling issues and, therefore, the increased capacity to support purifications larger than 5 g, and significant operational cost savings," White says. "SFC has been so successful, it is now used as the primary method for chiral analysis and purification within our lab." Over 12 months, White estimates saving more than $70,000, largely in solvent use, by switching from HPLC to SFC.

Previously, any larger scale purifications by HPLC had to be sent for health and safety reasons to specialist labs within the company, which are equipped to handle large solvent volumes. White's lab now has the flexibility to support SFC separations as large as 60 g, when separation and solubility are not limiting, and provide fast sample turnaround for medicinal chemistry colleagues. He hopes to move to even larger SFC separations next year to further support lead optimization.

Meanwhile, Zhang has put together a multicolumn approach that tests five columns in parallel (J. Chromatogr. A 2004, 1049, 75.) His system also uses online analytics; after independent UV analysis of each channel to determine resolution of the enantiomers, the five channels are pooled and a single circular dichroism detector tracks the elution order of the enantiomers. Initially developed for HPLC, it also works with SFC, Zhang notes.

Sanofi-Aventis researchers and a collaborator at the Free University of Brussels created a screen using the four popular columns and just two modifiers (J. Chromatogr. A 2005, 1088, 67). The starting conditions were selected based on a statistical analysis of 10 years of experience, and an overall 95% success rate in separating hundreds of chiral molecules generated in early drug discovery programs. A feature of their automated system, which switches serially between the columns and modifiers, is that it can be stopped as soon as a separation is achieved.

Among their goals was developing a fast SFC screen that would lead to generic high-throughput and high-performance separations. To test this, they screened a set of 40 marketed chiral drugs to evaluate the system's general applicability and achieved a similar high success rate of 98%. "The SFC screen is currently an integral part of our analytical support," they write, "and is considered the first try for chiral separations of new compounds."

Once the right separation conditions have been determined, SFC is fairly straightforward, says Christina Kraml, research scientist in discovery analytical science at Wyeth Research. She has conducted several comparisons with HPLC, measuring productivity, solvent use, and throughput. Kraml has been able to separate 65 g of material with just 6% modifier, for example, achieving what she calls "dramatically low solvent consumption."

Separating tens or hundreds of grams is done through repeated injections of material dissolved in a solvent, typically the modifier. And the material's solubility is a significant factor for getting concentrated solutions for injection. Injecting a few grams at a time to test throughput, "the reproducibility in SFC separation is amazing from one injection to another," Kraml says. "And you want that so you can walk away or let it run overnight and do the job."

SFC has limitations. Not all separations are possible, and thus most researchers see it as complementary to HPLC. Many researchers still are unfamiliar with the technique, and some consider the hardware or related engineering complex. Improved instrumentation, though, has reduced most skepticism surrounding the technique, one researcher comments. This is especially true concerning its use as a purification tool and related recovery issues caused by aerosol formation, since acceptable recovery levels of 85% are achievable.

TO IMPROVE separation, Kraml and coworkers have looked at derivatizing chiral primary and secondary amines, common intermediates and final products in enantioselective syntheses. Although many others have tried derivatization to improve selectivity, Kraml says carbobenzyloxy (cbz) derivatives haven't really been fully explored for this purpose. Cbz derivatives often appear in synthetic routes, she points out, since the cbz moiety is an easily attached and removed protecting group.

In some cases, the amine can be a terrible product to try to separate, Kraml explains, "but if we actually target it in the synthesis and add a cbz group, suddenly the solubility is good, and you can get a really nice separation."

The Wyeth team surveyed 11 different chiral amines, using five different CSPs and a variety of mobile phases. In all, 55 methods were developed for each racemate. In the end, cbz-derivatized amines were consistently best resolved by HPLC or SFC.

In testing the potential to scale up with SFC, the researchers found they could produce 1 g per minute of a very soluble amine with 98% recovery and at least 99.4% enantiopurity. They also found that all the derivatized amines were separated by Chiralpak AD or AD-H columns.

"The point was not that we found something that improves separation, but that if you can improve the separation, you can make a really big difference in throughput," she says. "So making the derivative is worth the trouble." Such details become important, she adds, now that the technology allows for larger scale separations and demand within R&D is increasing.

IN ANOTHER approach to separating amines, Stringham saw an improvement with a strong acid additive (J. Chromatogr. A 2005, 1070, 163). Separating basic compounds by SFC is often unsuccessful, possibly because the analytes interact with the mobile and stationary phases. Just 0.1% ethanesulfonic acid in 20% ethanol formed ion pairs with the amines that were stable to SFC separation. Separations were obtained for 36 out of 45 diverse compounds, many of which were not previously separable by SFC.

Like Kraml, Christopher J. Welch, who heads the analysis and preparative separations group within process research at Merck, advocates careful consideration of where separations might best be applied. Strategies for the efficient use of preparative-scale chromatography within synthetic routes are still evolving, according to Welch and coworkers (Chirality 2004, 16, 609). But they have shown how SFC led to a key enantiomerically pure 3-aryl--lactone intermediate to support rapid route exploration and optimization.

Welch and other coworkers also used SFC to facilitate early synthetic studies of an HIV protease inhibitor (Org. Process Res. Dev. 2004, 8, 186). During the past few years, the Merck lab has routinely used semipreparative SFC for purifying up to about 20 g. In the HIV project, after initially using SFC, the group scaled up, using a 30-cm-diameter column HPLC system, to produce 5.6 kg of a key intermediate.

Since then, the lab has acquired larger SFC equipment and has used it for what Welch believes are the first kilogram-scale SFC separations in the pharmaceutical industry (LC-GC 2005, 23, 16). Preparative-scale SFC helps support and expedite preclinical and early development work, he says, and is where the most dramatic cost and time savings come into play.

Scaling up SFC is rapid and straightforward, generally requiring only a simple extension of the results obtained in small-scale screens. "The labor input for developing a chromatographic method is very small, so the cost of defining the chromatographic option is very low compared with the cost of developing a synthetic route," he says. "That's really a core issue and one of the key drivers behind the increasing use of chromatography in industry today."

Overall, Welch believes, it's a matter of striking a balance between synthetic quality and speed--that is, between producing material quickly for rapid evaluation and developing syntheses that may be suitable for very large scale production.

Currently, SFC equipment costs more than comparable HPLC systems. One factor, Berger says, is that SFC instrumentation has only recently moved beyond its first generation, whereas major instrument companies are competing against each other in larger markets with more mature LC technologies. Despite having to compete against technology from these companies, "SFC will drive HPLC out of the semiprep and prep business," he believes.

Lower operating costs for SFC compared with HPLC quickly make up for the cost difference, users claim. "Pharmaceutical companies spend about $25 million per year on chromatography hardware for early drug discovery," Berger says. "If it were all HPLC, it then would cost them between $750 million and $1 billion to operate," and about $300 million if all SFC.

Seeing that "there's more money to be made in providing a service than in selling hardware," Berger has started up AccelaPure, based in Newark, Del., to offer contract SFC separations and consulting. The company opened its doors in March and has been hiring scientists, building labs, and buying and modifying equipment for increased throughput. It has already done some work for customers.

AccelaPure is competing with many technology and custom chemical providers that also offer separations services using HPLC or SFC. For very large scale separations, even some commercial production, the method of choice has been simulated moving-bed chromatography (SMB), a multicolumn, continuous method that makes more efficient use of solvents than does single-column HPLC.

Production scale might be the next stage in SFC's evolution, although the equipment doesn't yet exist and the economics relative to other purification or resolution methods are uncertain, Berger cautions. Some users believe, however, that it's only a matter of time and that the "greenness" of SFC and environmental concerns in manufacturing will spur further adoption. Separately, Novasep and collaborators (Ind. Eng. Chem. Res. 2001, 40, 4603) and researchers at the University of Hamburg (J. Chromatogr. A 1999, 865, 175) have built prototype systems combining SFC and SMB that some users consider promising.

Although fast to develop, chromatographic methods tend to be "the method of last resort, and part of the reason is it's been hard to make it cheap enough," Berger says. "There's always been that bar to jump over, but the bar has gotten lowered in large part because of the significantly lower operating costs for SFC."