SMB chromatography: a fast way forward?

Simulated moving bed chromatography may provide an easier, faster and more cost-effective alternative to asymmetric synthesis for the large-scale production of chiral isomers, suggests Dr Sarah Houlton

The majority of new drugs being licensed these days are chirally pure. Indeed, the only obvious reason for a racemate to be licensed is if the two isomers are interconverted so rapidly in the body that there is little point in synthesising it chirally pure in the first place. But the large majority of drugs that need to be single enantiomers (or diastereomers) pose a challenge to the development and production chemists. It is usually relatively straightforward to synthesise molecules in chirally pure form in the laboratory, but when the synthesis needs to be scaled up, things are often not so simple.

There are essentially two strategies that can be applied to creating single isomers: synthesising them in a chirally pure way, or making a racemate and finding a method of separating the two. The former would be the ideal situation, but it is not always possible to find an economically viable route, as the chemistry may be too complex, expensive or dangerous, or the asymmetric synthesis gives too low an ee. Conversely, the chemistry involved in making a racemate is usually cheaper and easier. Often the best strategy, therefore, is to synthesise the drug or intermediate in racemic form and separate the isomers, especially if the racemate is cheap and the separation is an easy one. In addition, it is usually simpler and quicker to develop a racemic synthesis, and chromatography may be the answer if the development process is to be fast-tracked into the production of a chiral drug.

Classical resolution, as devised by Pasteur to separate tartrate isomers in the 19th century, is still sometimes used industrially. Numerous more modern techniques have been devised since then, notably various different forms of chromatography, which can often be the cheapest and fastest way of producing chirally pure products. The basic principle involves the mixture being adsorbed onto a suitable stationary phase, and the components being moved along it at different rates as a mobile phase is flushed through.

Preparatory scale chromatography is essentially the same as the column chromatography performed in the laboratory. It is necessarily a batch process, and is run in a similar fashion to an analytical hplc machine, though obviously on a much larger scale. Eluent is pumped through columns maybe 10 or 15cm in diameter at a rate of several hundred ml/min.

Simulated moving bed chromatography, or SMB, is becoming more popular for large-scale chiral separations, because it is a continuous process. First developed 40 years ago, SMB is a continuous countercurrent separation process that involves a multi-column system.

In a true moving bed system, the countercurrent is achieved by the solid phase moving down the column by gravity. When the solid phase reaches the bottom, all the compound has been eluted from it, and it is then recycled to the top of the column. The mobile phase moves in the opposite direction — upwards — being injected at the bottom. The mixture to be separated is injected in the middle, and the flow rate set so that the desired extract moves towards the bottom of the column, and the raffinate towards the top.

However, the fact that the solid phase needs to be recycled to the top of the column makes the system difficult to operate in practice. A simulated moving bed system is much simpler to use, as the adsorbent remains in a fixed position, as in standard column chromatography. It consists of several fixed columns joined head to tail, and the countercurrent effect is created by gradually moving the inlet and outlet points clockwise around the system. This has the effect of simulating a solid flow in the opposite direction. Flow rates are set so that a good separation of raffinate and desired extract are achieved, and the whole system is computer-controlled to optimise the inputs and outputs.

SMB has been used at a hundreds of tonnes scale for the purification of p-xylene, the separation of fructose and other sugars in water, and the isolation of lysine from fermentation, as well as a variety of other low cost, high volume products. More recently, though, it has been applied to the much higher value separations involved in chiral pharmaceuticals. It has several advantages over classical separations: it uses less solvent; and isolation of the product is easier as all that needs to be done is to evaporate the solvent, rather than the protracted filtration and extraction that classical resolution engenders.

This shift towards use in the more complex separations involved in chiral APIs has become possible because of a number of improvements in the technology. Not only is the equipment being made more reliable, but also, importantly, improved stationary phases are now available in commercial quantities from companies like Daicel.

Dr Tom Archibald, vp of technology development at NextPharma Technologies, explained at the Chiral Technology conference at this year's CPhI exhibition that SMB systems have now been in operation for some time, and have proven to be robust, with no change in product quality or decrease in productivity. 'SMB is an inexpensive and practical way to manufacture commercial quantities of drug intermediates and APIs,' he said.

drawbacks

The biggest drawback of using a physical separation is that the maximum possible yield that can be achieved is 50%, as the other 50% is the wrong isomer. This can often be recycled by a racemisation procedure, and fed back into the separation.

But where in the synthetic procedure is it best to incorporate the separation step? In principle, SMB can be applied at any step along the way, from starting material through to the final API. Factors that should be considered when developing the route include the number of reactions, reaction yields and the cost of reagents.

Assuming that SMB adds less cost than a chemical reaction and is cheaper than a classical resolution, as is usually the case, Archibald claims that it is generally cheaper to perform the separation earlier on in the synthetic route. It means less is spent on reagents at later stages, as the unwanted half has already been discarded, and the reaction volumes will be half the size.

The important question that needs to be answered when deciding where in the process to use chromatography is, 'How much does the final product cost?' Archibald claims that if you put the separation step in at the right place then, despite the 50% maximum yield, it can be very cost-effective.

Another advantage that SMB chromatography has over other chiral syntheses is that it does not require any specific functionality to be present in the molecule. 'Unlike catalysis and other synthetic procedures,' says Archibald, 'it can be done with hydrocarbons. You do not need functional groups, and you can do it whenever you want – wherever it fits best into the process. As you create value in the synthetic chain, chromatography means you are throwing away 50% of that value. You need to look at the bigger picture, and choose the best option.' The lower the yields of later steps, the more important it is to separate early on.

A good example of a synthesis that includes a SMB separation is a published route to an intermediate for Pfizer's antidepressant sertraline (Scheme 1). Performing the SMB separation first was indeed found to be most cost-effective. The reductive amination has a high yield, is selective for cis products, and only half the material has to be used in the reaction. In addition, the unwanted tetralone isomer can be racemised and reused, whereas there is no simple racemisation for the reduced product. This synthesis has been developed into a high throughput method, yielding 370kg product per kilo of chiral stationary phase a year.

One reason for performing the separation later on in the process is if significant racemisation is seen in later steps of the synthesis. 'You have to look at the whole process as there can be problems later,' says Archibald. 'But equally, another argument for separating early is that there is less functionality in the product that can be affected by the separation.'

A number of factors must be considered for a successful SMB separation to be developed. The mixture to be separated must:

Ideally, the separation should be quick and work at high concentrations.

In general, says Archibald, productivity greater than 1kg product per kilo of chiral stationary phase a day will give favourable economics.

It is often possible to increase the yield by racemising the unwanted isomer and performing the separation again. But, even if a simple racemisation exists, it can often be cheaper to throw it away and start again. There are several potential practical limitations:

Choosing the right chiral stationary phase (CSP) is essential. Ideally, it must have a high and extended chiral recognition ability, with a high loading capacity, and be both chemically and physically stable. It must be available in quantities of at least half a tonne, without being excessively expensive, and tolerate a wide range of mobile phases. Eric Francotte of Novartis in Basel, Switzerland, speaking at the Chiral Technology conference, said that it is not possible to use just any chiral stationary phase for preparative separations.

There are essentially two strategies that can be applied to finding the best separation: selecting or adapting the CSP to suit the racemate, or adapting the solute to fit the CSP. The former can involve a random screening of CSPs, choosing it more specifically based either on experience or via a database, or designing a specific CSP by structural correlation, imprinting or molecular modelling (which has, up to now, not been particularly useful). A combinatorial approach is being used increasingly often. Similarly, the combinatorial approach can be applied to the second strategy, the alternatives being derivatising the racemate or using a precursor of the desired compound instead.

Francotte says that the available phases, Chiralcel OS, Chiralpak AD, Chiralpak AS and Chiralcel OJ, shown in Figure 1, should cover 85% of all chiral separations, based on a random screening of 500 racemic compounds. He believes that polysaccharide stationary phases are particularly successful because they have so many chiral receptors along the chain.

However, he explained, their major limitation is that they are highly soluble in many organic solvents, which reduces stability, and cuts the possibilities of influencing selectivity by varying the mobile phase. This can be overcome if the polysaccharides are immobilised onto silica

As Dr Richard G. Einig, director of the quality unit at Johnson Matthey company Pharm-Eco explained, including chromatographic separations within the route to a pharmaceutical active adds further regulatory requirements. The 2000 ICH draft consensus guidelines on GMPs for APIs and quality contain several sections that must be applied to production-scale syntheses of APIs that include chromatographic procedures.

Of particular relevance are the sections on equipment design, selection of media, use and reuse of media and solvents, personnel training, process documentation and the validation of equipment, and the process and cleaning procedures. 'Chromatography can significantly shorten the time to market for new APIs, and it can be done in full compliance with regulatory guidelines,' says Einig. 'Purity and yields can be higher than can usually be achieved by traditional processes.' But, he adds, care must be taken to ensure all regulatory concerns are considered in full.

SMB chromatography is routinely performed in the laboratory, and work in recent years has proved its feasibility in production. UCB in Belgium is able to resolve 12t p.a. of racemate in a pilot unit, and the process, using 50kg of CSP, was proved to be economically competitive in tests over two years. It is also now being used successfully in production. Aerojet Fine Chemicals of Sacramento, US is building an SMB unit to process more than 500t p.a. of racemate under contract for UCB. And Lundbeck of Denmark produces several hundred tons p.a. of its SSRI antidepressant S-citalopram.

challenges

Several challenges remain, however. Being able to increase loading capacity would lead to higher productivity. And a greater number of chiral stationary phases would enable more difficult separations to be performed, with the best CSP option being identified through a combinatorial process.

Work is under way into the separation of much more complex systems, such as purifying products from fermentation broths or solid phase syntheses. Systems are being developed with bigger columns that allow larger production scale separations to be performed, as is on-line racemisation to increase productivity.

Separating racemates into their enantiomers is never going to be the answer for every chiral synthesis problem. But for those situations where a good separation can be achieved, and racemate synthesis is much cheaper or more efficient than an asymmetric synthesis procedure, then it can provide an excellent route to making production scale quantities of single isomer APIs or intermediates.