Practical solvent system selection for counter-current separation of pharmaceutical compounds

Practical solvent system selection for counter-current separation of pharmaceutical compounds

Journal of Chromatography A, 1207 (2008) 190–192 Contents lists available at ScienceDirect Journal of Chromatography A journal homepage: www.elsevie...

248KB Sizes 0 Downloads 4 Views

Journal of Chromatography A, 1207 (2008) 190–192

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Short communication

Practical solvent system selection for counter-current separation of pharmaceutical compounds Stéphane Dubant ∗ , Ben Mathews, Paul Higginson, Robert Crook, Martin Snowden, John Mitchell Chemical R&D, Pfizer Global R&D, Sandwich, Kent CT13 9NJ, UK

a r t i c l e

i n f o

Article history: Received 24 June 2008 Received in revised form 20 August 2008 Accepted 25 August 2008 Available online 5 September 2008 Keywords: Counter-current chromatography Solvent system selection Pharmaceutical Statistical approach

a b s t r a c t Counter-current chromatography (CCC) is a technique that shows a lot of potential for large scale purification. Its usefulness in a “research and development” pharmaceutical environment has been investigated, and the conclusions are shown in this article. The use of CCC requires the development of an appropriate solvent system (a parameter of critical importance), a process which can be tedious. This article presents a novel strategy, combining a statistical approach and fast HPLC to generate a three-dimensional partition coefficient map and rapidly predict an optimal solvent system. This screen is performed in half a day and involves 9 experiments per solvent mixture. Test separations were performed using that screen to ensure the validity of the method. © 2008 Elsevier B.V. All rights reserved.

1. Introduction Counter-current chromatography, first introduced in 1970 by Ito and Bowman [1], is an effective separation technique based on the differences of partitioning of the compounds in a sample between two non-miscible solvent systems. The separation times were at first rather long [1]. In the early 1980s, high-speed CCC (HSCCC) enhanced the performance of the technique [2]. This was achieved through a better mixing of the solvent mixture and greater retention of the stationary phase. This enhancement allowed separations to be performed in much shorter time. Since then, this technique has been widely used for the separation of natural products [3–5]. The utility of this technique in the pharmaceutical field has also been demonstrated, mainly for the purification of active compounds from traditional Chinese medicines [6], but to our knowledge no work has been published on the purification of pharmaceutical synthetic mixtures. Although not as intuitive to use as most solid-phase liquid chromatography techniques [e.g. highperformance liquid chromatography (HPLC), normal-phase liquid chromatography or supercritical fluid chromatography (SFC)] at the moment or as efficient (in regard of theoretical plate count) as HPLC on of the most commonly used preparative chromatographic technique, CCC presents some advantages. It is a versatile technique since any solvents that are required to obtain the desired selectivity can be used as long as the chosen solvent system is biphasic. Its

∗ Corresponding author. E-mail address: [email protected] (S. Dubant). 0021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2008.08.113

running cost is lower than the other chromatographic techniques in regard of solvent use and the absence of an expensive solid stationary phase. Also, the instrument is not as intolerant to particulates in the sample as for example a HPLC system is. The choice of the solvent system is a critical step in the method development of a CCC separation as it is the correct choice of the two solvent mixtures that will ensure the successful separation of the compounds in the mixture. Therefore, several articles have been published on different ways to efficiently obtain the most suitable solvent system for a specific separation. The partitioning coefficient parameter (K) is of paramount importance as the value of this parameter is going to determine the retention time for a compound in the column. CCC can even be used to predict the partition coefficient of a compound in a solvent mixture [7]. Ito gave a comprehensive summary of the process involved to determine appropriate conditions for a compound [8]. Hilal and Karickhoff did some work to predict the partitioning of the compound (therefore a solvent system can also be predicted) using computational chemistry [9]. It can be a very powerful tool but requires prior knowledge of the structure (and properties) of your target. Garrard suggested the use of automation to screen a table of 28 solvent mixtures and then select the best option out of those for a CCC run [10]. To determine the partition coefficient of 28 mixtures can be long, especially if automation is not available, and still not provide a readily suitable solvent system. Friesen and Pauli developed the G.U.E.S.S. method, a practical approach to the problem by comparing the behaviour of the target compound under a set of specific conditions with the behaviour of some reference products [11] on a CCC instrument. To use that method, a pure reference sample of your target

S. Dubant et al. / J. Chromatogr. A 1207 (2008) 190–192

191

Table 1 List of the mixture tested routinely in the CCC screen to find separation conditions Sample

A B C D E F G H I

Aqueous phase

Organic phase

Water (%)

Methanol (%)

Heptane (%)

EtOAc (%)

25 25 50 50 50 50 75 75 75

75 75 50 50 50 50 25 25 25

50 75 25 50 50 75 25 50 75

50 25 75 50 50 25 75 50 25

D and E are identical experiments to confirm the reproducibility of the measurements and give confidence in the statistical calculations.

compound is needed to be able to assess its behaviour in the different conditions. These methods are not suitable for use in the high throughput environment of the pharmaceutical industry. A fit for purpose, fast and reliable method development was still needed for the development of CCC in a pharmaceutical environment. This article will deal with the usefulness of CCC in the pharmaceutical industry and the work done to make this technique a fast, accessible, efficient and reliable tool to routinely purify a wide variety of pharmaceutical compounds by developing an efficient way to screen for the most appropriate solvent mixture to provide the selectivity required to achieve a successful separation. 2. Experimental—solvent system choice Two different CCC instruments were used in this study: a Quattro CCC from AECS Quickprep (Bridgend, UK) with a 30 mL column and a 250 mL column volume and a MIDI CCC from Dynamic Extraction (Slough, UK) with a 37 mL and a 950 mL column volume. The detector used on both instruments was an ELSD 2000 evaporative light scattering detection (ELSD) system from Alltech, Carnforth, UK. Typical experimental parameters are as follows: Quattro CCC: 850 rpm, 30 ◦ C, lower phase stationary, flowrate 20 mL/min on the 250 mL column, 80% stationary phase retention (Sf) MIDI CCC: 1400 rpm, 30 ◦ C, lower phase stationary, flowrate 1 mL/min on the small column and 25.5 mL/min on the large column, 75% stationary phase retention (Sf) ELSD 2000: a split was used to have a constant flow of 0.5 mL/min in the detector with a temperature of 40 ◦ C and a flow of 2.4 L/min of nitrogen The use of CCC as a standard separation technique in the laboratory requires a reasonably fast and reliable method development procedure. The use of CCC in the laboratory showed that most of the compounds encountered in a pharmaceutical environment are suitable to be separated using an n-heptane/ethyl acetate/methanol/water mixture (e.g. Arizona type solvent mixtures). Therefore, this work was focused on method development for this solvent family. A design of experiment (DoE) set of experiments has been established and tested to rapidly provide suitable solvent conditions for the desired separation (see Table 1 for screening conditions). It involves the determination of 9 partition coefficients per solvent system family. A liquid handler (Gilson 234 Autoinjector) was used to dispense the appropriate amount of solvent in each vial (2 mL in total), already containing about 1 mg of the sample to screen. The method used for the partition coefficient

Fig. 1. Map generated for the prediction of conditions for chloropropamide. The dots represent the actual values measured during the screen. The percentage of solvents presented on the axis are the solvent ratios used to mix the solvent system and do not represent the actual composition of the equilibrated solvent systems. The specific equation producing that map is K = 0.0346678 – 11.9342 × water + 4.95155 × heptane + 73.5977 × water2 − 97.3039 × water × heptane + 36.3119 × heptane2 and R2 =97.2%.

measurement is the flask-shake method with analysis of both layers using fast gradient HPLC. This DoE is 32 factorial design with one corner left out due to experimental constraints (one of the corner values, n-heptane/ethyl acetate/methanol/water 25/75/75/25 is not usable as it would give just one phase). The 9 partition coefficients gathered from this screen allow the mapping of the entire practical polarity range of the solvent family tested (see an example in Fig. 1). The limits imposed to the amount of a specific solvent in the quaternary mixture were determined to ensure the presence of two phases, a good phase separation and settling time. This three-dimensional (3D) partition coefficient map and the use of a statistical software allows the prediction of a solvent system that will give the desired partition coefficient for the target compound. The software will generate an equation to calculate the 3D partition coefficient map. A generic equation is given below: K = A+b × (ratio water) + c × (ratio heptane) + d × (ratio water)2 + e × (ratio water) × (ratio heptane) + f × (ratio heptane)2 with A, b, c, d, e and f values that are specific to one compound and determined thanks to the K values obtained from the screen. From that equation the R2 value is obtained which describes (in %) how well the model equation fits the K values that have been measured during the screen. A value of R2 ≥ 90% is recommended to have a reliable prediction.

Fig. 2. CCC chromatograms of a synthetic pharmaceutical sample purified by CCC (left: 200 mg in 2 mL, 37 mL coil, 1400 rpm, 1 mL/min; Sf 77%, normal phase mode; right: 5.2 g in 50 mL, 950 mL coil, 1400 rpm, 25.5 mL/min, Sf 75%, normal phase mode). The method was developed using the screening technique, tested on a small scale and directly scaled up. The target is the last eluting compound. The differences observed on the chromatograms between the analytical and preparative runs are believed to be related to detector saturation.

192

S. Dubant et al. / J. Chromatogr. A 1207 (2008) 190–192

Table 2 Verification of the accuracy of the computer prediction for CCC solvent systems; column volume is 250 mL, target K = 1.0, normal phase mode, 50 mg injected in 2 mL of mobile phase Compound

Reserpin Ninhydrin Chloropropamide Dipropyl phtalate Methyl prednisolone Cortisone Lidocaine

Predicted solvent mixture Water (%)

MeOH (%)

Heptane (%)

Ethyl acetate (%)

40 65 55 25 70 70 45

60 35 45 75 30 30 55

55 30 45 90 40 40 55

45 70 55 10 60 60 45

A partition coefficient of 1.0 is usually chosen if the selectivity is good enough, which enables a rapid separation. In the eventuality of the compound being too polar or too apolar to be separated by an Arizona-type solvent system, the same screening principle can be applied to any appropriate solvent system family as long as a biphasic mixture is retained with the new choice of solvents [e.g. n-heptane/dichloromethane (DCM)/methanol/water]. If chromatography is used to monitor the partition coefficient evolution through the different mixtures, data is obtained for every single compound in the mixture and conditions can then be predicted for any of them without extra experimentation. The maps obtained for the target compound and key impurities can be overlaid to visualize the regions of greatest selectivity. Several model compounds have been tested using this method to ensure its robustness. The surface response in an Arizona-type solvent system was mapped for those different compounds and a solvent mixture that would give a partition coefficient of 1 was predicted for each of them. The experimental retention volume of a specific compound in a CCC run was then determined using the corresponding solvent systems. For each example, the screen for conditions and the test-run on a CCC were performed within a halfday.

Retention volume measured (mL)

K (exp)

257 264 210 243 253 278 260

1.03 1.06 0.84 0.97 1.01 1.11 1.04

The above method development technique was then used to purify several compounds from mixtures coming from synthetic pharmaceutical reactions. The method development was carried out and transferred to either a Quattro CCC (AECS Quickprep) or MIDI (Dynamic Extraction) CCC instrument. An example is given in Fig. 2. The method developed allowed a relatively short run-time of 45 min and efficient separation of the target compound from impurities. The target, a pharmaceutical compound, was recovered with 88% theoretical recovery and 98% purity (HPLC–UV) in both runs. 4. Conclusion A new screening method has been developed to select a solvent mixture for a classical CCC run. This method has been tested with model compounds to ensure the solvent system prediction is fit for purpose and then applied to real pharmaceutical samples. It proved to be both fast and reliable so long as the crude sample can be rapidly analysed by HPLC. This approach enabled the routine use of CCC as a separation tool for purifications ranging from 20 mg to a few hundred grams on laboratory scale. References

3. Results and discussion The experimental values obtained for the partition coefficients varied between 0.84 and 1.11. This variation is considered fit for purpose for most of the separations, as long as the required selectivity is achieved. The prediction proved to be good enough for use in each case, giving confidence in the reliability of the prediction as can be seen in Table 2. This method can be used to reliably and rapidly predict quaternary solvent system using any solvent for a specific separation within a few hours.

[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]

Y. Ito, R.L. Bowman, Science 167 (1970) 281. Y. Ito, J. Chromatogr. 214 (1981) 122. R. Liu, X. Chu, A. Sun, L. Kong, J. Chromatogr. A 1074 (2005) 139. Y. Wei, T. Zhang, G. Xu, Y. Ito, J. Chromatogr. A 929 (2001) 169. G. Leitao, P.A. de Souza, A. Moraes, L. Brown, J. Liq. Chromatogr. Rel. Technol. 28 (2005) 2053. A. Marston, K. Hostettmann, GIT Labor Fachz. 11 (7–8) (2007) 34. A. Berthod, S. Carda-Broch, J. Chromatogr. A 1037 (2004) 3. Y. Ito, J. Chromatogr. A 1065 (2005) 145. S.H. Hilal, S.W. Karickhoff, QSAR Comb. Sci. 23 (2004) 709. I.J. Garrard, J. Liq. Chromatogr. Rel. Technol. 28 (2005) 1923. J.B. Friesen, G.F. Pauli, J. Liq. Chromatogr. Rel. Technol. 28 (2005) 2777.