Multimodal HPLC screening of polysaccharide-based chiral stationary phases

The development of single enantiomer drugs has gained considerable attention in the pharmaceutical industry. The number of new chiral drugs submitted to regulatory authorities as single enantiomers instead of racemic mixtures has tremendously increased during the past 10–15 years, and trends for future drug development are evident.¹ The growing emphasis on single enantiomer drugs has been fuelled by regulations governing chiral drugs and the recognition that enantiomers of a chiral drug substance may have dramatically different pharmacological activities and different pharmacokinetic and pharmacodynamic properties.^2–4 In addition, considerable advances in asymmetric synthesis and preparative separation of chiral molecules have enabled the production of single enantiomer drugs on a commercial scale.^5,6

Chromatographic separation of chiral drugs

As pharmaceutical companies shift towards the development of single enantiomer drugs, the need for rapid and reliable chromatographic methods to determine the enantiomeric purity (e.p.) of chiral drug substances and intermediates is growing. A whole range of analytical techniques are available for enantioselective separations of chiral drugs, including gas chromatography (GC), high-performance liquid chromatography (HPLC), packed-column sub- and supercritical fluid chromatography (SFC), capillary electrophoresis (CE) and capillary electrochromatography (CEC).^7,8 Chiral separations can be performed either directly using chiral stationary phases (CSPs) or chiral additives, or indirectly using chiral derivatizing agents to form diastereomeric pairs, which can be resolved under achiral conditions. Direct chromatographic separations are generally preferred to indirect approaches, for example, to avoid extensive sample preparation and problems as a result of racemization, kinetic resolution and enantiomeric impurities present in chiral derivatizing reagents.

Figure 1

Chiral HPLC has emerged as the method of choice for enantioselective separations of chiral drugs. HPLC is commonly employed in pharmaceutical analysis, and is well-established in drug discovery and development. It is more versatile than most other separation techniques and enables the successful resolution of a wide range of organic compounds, including most drug substances and intermediates. In addition, chiral HPLC methods may eventually be scaled up for preparative separations by simply increasing the column size and using throughput-increasing techniques such as simulated moving bed (SMB) methods.

Chiral HPLC

Generic column screening. Selection of an appropriate CSP is a bottleneck in the development of enantioselective HPLC methods. Despite the existence of a large number of CSPs — more than 100 stationary phases are commercially available, including Pirkle type, ligand-exchange, cyclodextrin, crown-ether, macrocyclic glycopeptide, protein, chiral polymers and polysaccharide-based CSPs^7,8 — it remains difficult to predict which column will provide the desired enantioseparation. Chiral recognition mechanisms underlying enantioselective separations are complex and are generally not fully understood. Chiral method development mainly relies upon trial-and-error or on the analyst's experience and is a sophisticated, time- and resource-consuming task.

Table 1 Chromatographic conditions used in the generic NP/PO screen at 20 °C.

So far, the best strategy to achieve rapid enantioselective separations for a large number of structurally diverse chiral molecules is to screen a limited number of CSPs with broad chiral recognition abilities under a variety of mobile phase conditions.^9–17 Based on the screening results the best CSP is selected and operating conditions such as mobile phase composition and temperature are adjusted to further optimize the enantioseparation.

At Johnson & Johnson Pharmaceutical Research and Development (J&J PRD), HPLC plays a predominant role in chiral chromatography. In recent years, two automated generic screening modules were introduced to rapidly develop robust and reliable chiral assays to determine the e.p. of chiral molecules. To date, more than 100 chiral drug substances and intermediates have been screened on one or both screening modules with a high level of success.

Figure 2

Screening module 1

Normal phase and polar organic mode. A first generic screening methodology was developed using four commercially available polysaccharide-based CSPs; that is, Chiralcel OD-H [cellulose tris (3,5-dimethylphenylcarbamate)]; Chiralcel OJ-H [cellulose tris (4-methylbenzoate)]; Chiralpak AD-H [amylose tris (3,5-dimethylphenylcarbamate)]; and Chiralpak AS-H [amylose tris ((S)-methylphenylcarbamate)].¹⁸ The Chiralcel and Chiralpak stationary phases (Daicel, Japan) are particularly useful as they exhibit a high and complementary chiral recognition ability towards a wide range of structurally different analytes. Moreover, they are available in bulk amounts and in a broad range of particle sizes, which assists scale up from analytical to (semi-)preparative systems. Column screening was operated in the normal phase (NP) and polar organic (PO) mode as chiral recognition of polysaccharide-based CSPs is considered to be most effective under these conditions (Table 1). Isocratic elution is applied to avoid extensive column equilibration.

As most of the chiral screening and method development work is performed in support of chemical process R&D, rapid and efficient screening is a must. Hence, solvent and column switching devices were used to automate the screening unit. Even so, to reduce total analysis time, the column switcher was rapidly replaced by a Sepmatix 8-fold parallel HPLC system (Sepiatec GmbH, Germany).^16,17 The Sepmatix HPLC system is an integrated, eight-column instrument that enables the analysis of a single sample onto eight different columns simultaneously. In addition, the screening module was extended with a Chiralpak IA and IB column. The chiral selectors in Chiralpak IA and IB are of the same nature as in Chiralpak AD-H and Chiralcel OD-H, respectively. The latter are made by physical coating of the polymer on a silica gel, whereas the chiral selectors in Chiralpak IA and IB are immobilized on the support. Hence, Chiralpak IA and IB can be operated with a greater range of solvents, and may exhibit a different enantioselectivity and efficiency than Chiralpak AD-H and Chiralcel OD-H. This then enhances the probability of success during chiral column screening. Despite the inclusion of two additional columns, the cycle time per analyte was dramatically reduced from 18.3 to 4.3 h.

To date, 150 chiral drug substances and intermediates have been analysed under NP/PO conditions on a Chiralpak AD-H, Chiralcel OD-H, Chiralpak AS-H and Chiralcel OJ-H column. About 117 target analytes were basic compounds, 27 were neutral molecules and only six compounds contained acidic functionalities. In addition, nearly 20% of the analytes had more than one chiral centre. Enantioselectivity was achieved for 85% of the molecules with at least one combination of CSP and mobile phase condition. A validated chiral HPLC method was eventually obtained for all of those analytes. In 60% of the instances, no further optimization was required. For 23 compounds (15%), little or no enantioselectivity was observed under NP/PO conditions. The analytes were subjected to the second chiral screening module, which is operated under reversed phase (RP) and polar ionic (PI) conditions (see later). Fourteen chiral compounds were successfully separated this way. The remaining nine analytes were resolved by chiral CE.¹⁹

Although for most analytes enantioselectivity was observed on more than one stationary phase and mobile phase combination, some CSPs and eluents were more successful in achieving adequate enantioseparations within a reasonable time frame (<20 min) than others (Figure 1). Approximately 48% of all chiral HPLC methods were developed and validated with a Chiralpak AD-H column. About 23% of all chiral methods were developed on a Chiralcel OJ-H column, while the Chiralcel OD-H column accounted for 19% of all successful chiral separations. Finally, 10% of all separations were achieved with a Chiralpak AS-H column.

Approximately half of the target analytes were also screened on the Chiralpak IA and IB columns. The chiral recognition ability of the Chiralcel AD-H and Chiralpak IA column were very similar. If enantioselectivity was observed on a Chiralpak AD-H, it was observed on a Chiralpak IA column in ;90% of the instances. A similar success ratio was found for the Chiralcel OD-H and Chiralpak IB columns. Even so, enantioseparations on coated and immobilized columns often differed in terms of efficiency, resolution and retention times.

PO mobile phases were generally more successful in terms of efficiency and enantioselectivity than the alkane/alcohol mixtures. At best, similar enantioselectivity and resolution were achieved with NP and PO eluents. Yet, alkane/alcohol mixtures generally resulted in much longer analysis times. Hence, future screenings could be run in the following order: alcohols>alkane/alcohol mixtures to reduce total analysis time. If a satisfactory enantioselectivity is obtained with PO mobile phases, the screening can be stopped. To improve the number of successful chiral separations further, organic solvents other than alcohols and alcohol/alkane mixtures will be evaluated, especially with the Chiralpak IA and IB columns. Similarly, other (new) columns with broad enantiorecognition mechanisms, such as the macrocyclic glycopeptides or cyclodextrines, will be assessed as a chiral selector.

Screening Module 2

Reversed phase and polar ionic mode. Following the implementation of the first module, a second generic screening unit was introduced (Table 2). Initially, the second module consisted only of a RP screen of four polysaccharide-based CSPs. The polysaccharide-based CSPs are similar to those employed in the first screening module, yet they are specifically designed for aqueous/organic mobile phases.

Table 2 Chromatographic conditions used in the generic RP/PI screen at 25 °C.

The impetus to develop a second screening module was related to the lack of retention and solubility of very polar compounds in PO solvents and alkane/alcohol mixtures respectively and the difficulty to analyse water-soluble drug products under NP conditions. RP screening enables the enantioselective separation of (very) polar analytes and facilitates the analysis of aqueous samples. Solutions based on, for example, cyclodextrines or polyethylene glycol (PEG), can directly be injected into the analytical system, avoiding the use of (extensive) sample preparation. Similarly, the e.p. of chiral drug substances in solid-state formulations can easily be determined by analysing an aqueous extract. What's more, the chiral recognition ability of polysaccharide-based CSPs in the RP mode is different, and often complementary to NP/PO conditions, enhancing the probability of success.

To separate chiral molecules under RP conditions, selection of an appropriate mobile phase is of paramount importance. As polysaccharide-based CSPs do not possess ionic sites to interact with charged molecules, analytes have to be kept neutral to avoid loss of enantioselectivity. Hence, acidic (pH 2) and basic (pH 9) aqueous buffers are used to control the pH of the mobile phase and suppress the ionization of acidic and basic analytes.

Once an appropriate aqueous buffer is selected, a racemic mixture of the target analyte is screened on all polysaccharide-based CSPs with each of four organic modifiers (acetonitrile, methanol, ethanol and 2-propanol) in a 20 min linear gradient. Screening is fully automated and takes 18 h for a single chiral molecule, including column priming, equilibrating and column rinsing. Gradient elution is applied to reduce analysis time and clean related substances from columns prior to injection of a new sample. The initial mobile phase composition is adjusted to the solvent strength of the organic modifier to achieve similar retention times with each eluent.

To date, 42 chiral drug substances and intermediates, involving 49 enantiomeric pairs, were screened on all polysaccharide-based CSPs under RP conditions. Target analytes included 38 basic, one neutral and three acidic compounds.

The outcome of the RP screening is shown in Figure 2. Enantioselectivity was observed for 41 enantiomeric pairs with at least one chiral selector and mobile phase combination. The chiral recognition ability of Chiralpak AD-RH and Chiralcel OJ-RH was generally greater than that of Chiralcel OD-RH and Chiralpak AS-RH. Together, Chiralpak AD-RH and Chiralcel OJ-RH showed enantioselectivity for 35 out of 41 enantiomeric pairs. Even so, the Chiralpak AS-RH and Chiralcel OD-RH columns were often complementary to Chiralpak AD-RH and Chiralcel OJ-RH, enhancing the number of successful enantioseparations. In addition, several compounds were better resolved on Chiralpak AS-RH or Chiralcel OD-RH, simplifying further method optimization. As for the organic modifiers, most enantioseparations were achieved with acetonitrile and methanol. At best, ethanol and 2-propanol gave similar or slightly better results.

So far, the RP screening results were solely used to evaluate the performance of the screening approach. Most of the results have not (yet) been optimized to yield a valid(ated) chiral HPLC method. Even so, the CSP and mobile phase combination that is most likely to yield a suitable method upon optimization is illustrated in Figure 3. Selection of the best screening result was based on the resolution factor, the peak shape of both enantiomers and their respective retention times.

Figure 3

Based on these findings, it is clear that none of the CSPs should be discarded from the RP screening module. A parallel screening system, however, would be very helpful to improve sample throughput and increase laboratory efficiency. To further reduce total analysis time, future screenings could be run in the order methanol>acetonitrile>ethanol>2-propanol. Hence, screening can be stopped as soon as a satisfactory enantioseparation is obtained.

Despite the high versatility of polysaccharide-based CSPs, enantiomers of several target analytes could not (sufficiently) be resolved under RP conditions. To enhance the number of successful enantioseparations, racemic mixtures of most target analytes were subsequently screened on the macrocyclic glycopeptides Chirobiotic T2 (teicoplanin) and Chirobiotic V2 (vancomycin). Macrocyclic glycopeptides exhibit a very broad range of enantioselectivity, and can be operated under various mobile phase conditions.²⁰ The chiral recognition mechanism differs from polysaccharide-based CSPs, which enhances the probability of success.

The macrocyclic glycopeptides are operated under PI mobile phase conditions (Table 2). According to the manufacturer, the PI mode improves the enantioselectivity of a broader range of chiral compounds than the RP, NP or PO mode.²¹ The interaction mechanism is predominately ionic, and method development is very simple and fast.

Figure 4

Target analytes were screened with methanol under isocratic conditions with varying ratios of acetic acid and N,N,N-triethylamine (TEA). Results of the PI screening are shown in Figure 4 for 38 chiral analytes. The chiral recognition ability of both columns was much lower than that of the polysaccharide-based CSPs, however, the PI screening was highly complementary with the RP screening. Combined with the RP screening results, chiral recognition was now observed for 44, instead of 41, of 49 enantiomeric pairs. In addition, a number of chiral analytes were better resolved on the macrocyclic glycopeptides than on the polysaccharide-based CSPs, facilitating further method development (Figure 5). Overall, Chirobiotic T2 was less successful than the vancomycin stationary phase. At best, similar results were obtained on both columns. Enantioselectivity was generally observed with all acid/base ratios, but for a number of chiral analytes enantiorecognition was only observed with one of the acid/base ratios.

Figure 5

Owing to its high complementarity, the isocratic PI screen is now systematically performed with the RP screening of polysaccharide-based CSPs. To simplify and speed-up column screening, future analysis could be run in the order Chirobiotic V2>Chirobiotic T2. Based on the current screening results, the Chirobiotic T2 could even be discarded from the screening approach (Figure 5). Columns should be run with all acid/base ratios.

From screening result to final method

Upon screening, the best CSP and mobile phase condition are selected for method optimization and validation. Selection is primarily based on the degree of enantioselectivity and chiral resolution observed for each column/mobile phase combination. A high resolution (Rs) value is generally preferred. If Rs>1.5 method optimization is not really required, unless analysis time is considered too long. If 0<Rs<1.5, operating conditions such as mobile phase composition, flow rate and temperature are adjusted to achieve baseline resolution, and eventually shorten total analysis time. In case gradient elution was applied during screening, operating conditions are often transposed to isocratic elution.

The latter instance is illustrated in Figure 6. Analyte T2650 was screened in the RP/PI mode under gradient elution. The best screening result is shown in Figure 6(a). Enantiomers were baseline separated, but eluted in the organic rinse step following the 20 min gradient. To reduce total analysis time, the assay was run isocratically with pure organic modifier. While some loss of resolution was observed, analysis time was dramatically reduced [Figure 6(b)].

Figure 6

As chiral methods are intended for quantitative (e.p.) analyses of single enantiomer drugs or intermediates, parameters such as peak shape and elution order are equally important. Methods are often required to enable the accurate determination of 0.1 area% of one enantiomer as an impurity in the other isomer. Hence, adequate peak shape is desired along with a high resolution factor. If possible, the undesired enantiomer should elute first to avoid possible interference caused by the tail of the major peak.

If the undesired enantiomer elutes after the API in all screening results, it should be adequately resolved from the major peak's tail to enable accurate determinations of the e.p. However, elution order reversal has sometimes been observed during screening, moving from one polyscaccharide-based CSP to another or even from one mobile phase condition to another.

Figure 7

Enantiomers should also be separated from any impurity present in the target analyte. If coelution or partial peak overlap is observed, further optimization is required to improve selectivity, as illustrated in the following examples.

To separate four stereoisomers of drug substance intermediate T2428 (Figure 7), a mixture containing these isomers was screened on four polysaccharide-based CSPs under RP conditions. Stereoisomers were successfully separated on a Chiralcel OJ-RH column with a water/methanol gradient. However, peak overlap was observed with stereoisomers of the precursor T2399 [Figure 7(a)]. Elution with other mobile phases such as water/acetonitrile did not improve selectivity, and separation of T2428 isomers only worsened [Figure 7(b)]. Conversely, chiral recognition of T2399 was greatly enhanced by acetonitrile, so a water/methanol/acetonitrile gradient was evaluated. Eventually, a 40 min linear gradient with a water/acetonitrile/methanol mobile phase, from 40/18/42 (v/v/v/) to 0/30/70 (v/v/v), was used to separate the stereoisomers of T2428 from T2399 isomers while achieving the desired enantioselectivity (Figure 8).

Figure 8

Similarly, the operating conditions of a chiral HPLC method were adjusted to separate the enantiomers of T1329 (Figure 9) from achiral process-related impurities, while maintaining enantioseparation. Both enantiomers were baseline resolved on a Chiralpak AD-H column operated in the PO mode. The highest chiral resolution (Rs 5 4.9) was achieved with an acetonitrile/ethanol, 1/1 (v/v) mobile phase. Analysis time was less than 8 min. When a chromatogram of a racemic sample was compared with chromatograms of the synthesis-related impurities present in representative drug substance intermediate batches, partial coelution of the R-enantiomer with one of the impurities was observed [Figure 9(a)]. According to the NP/PO screening results, enantiomers were also baseline separated with a acetonitrile/methanol, 1/1 (v/v) mobile phase, though with a lower Rs-value (Rs 5 2.9). Even so, coelution with achiral impurities was no longer an issue [Figure 9(b)].

Figure 9

Conclusion

Multimodal HPLC screening of polysaccharide-based CSPs enabled the rapid and successful enantioseparation of a large number of chiral molecules of interest to J&J PRD. To date, 150 chiral drug substances and intermediates have been screened under normal phase and polar organic conditions, while 42 chiral compounds were screened in the RP and PO mode. Successful enantioseparations were obtained for 85 and 79% of the target analytes, respectively. Based on the numerous screening results, both modules will be further rationalized to increase the number of successful chiral separations, while reducing analysis time and increasing sample throughput. Hence, promising (new) CSPs with broad and complementary chiral recognition mechanisms, such as the macrocyclic glycopeptides and cyclodextrines, will continuously be assessed. Similarly, other chromatographic techniques will be evaluated. Owing to its high speed, short analysis times, low cost, user-friendliness and limited environmental impact, packed-column SFC is of particular interest, and will become a viable alternative to chiral HPLC to separate chiral drug substances and intermediates.

References

1. H. Craner et al., Drug Discov. Today,9, 105–110 (2004).

2. www.fda.gov/cder/guidance/stereo.htm

3. A.J. Hutt, Smith and Williams' Introduction to the Principles of Drug Design and Action, 4th Edition, in J.H. Smith, Ed. (CRC Press LLC, Boca Raton, FL, USA, 2006), pp 117–183.

4. D.R. Brocks, M. Vakily and R. Mehvar, Chirality in Drug Design and Development, I.K. Reddy and R. Mehvar, Eds. (Marcel Dekker Inc., New York, NY USA, 2004), pp 191–280.

5. V. Farina, et al., Chem. Rev., 106, 2734–2793 (2006).

6. D.A. Pflum et al., OPR&D, 5, 110–115 (2001).

7. G. Gübitz and M.G. Schmid, Biopharm. Drug Dispos., 22, 291–336 (2001).

8. Y. Zhang et al., Drug Discov. Today, 10, 571–577 (2005).

9. M.E. Andersson et al., J. Chromatogr. A, 1005, 83–101 (2003).

10. C. Perrin et al., J. Chromatogr. A, 947, 69–83 (2002).

11. C. Perrin et al., J. Chromatogr. A, 966, 119–134 (2002).

12. N. Matthijs et al., J. Chromatogr. A, 1041, 119–133 (2004).

13. A.X. Wang, J.T. Lee and T.E. Beesley, LCGC N. Am., 18, 626–639 (2000).

14. M. Luz de la Puente et al., J. Chromatogr. A, 983, 101–114 (2003).

15. Y. Zhang et al., J. Chromatogr. A, 1049, 75–84 (2004).

16. P. Sajonz et al., Chirality, 18, 803–813 (2006).

17. C.J. Welch et al., J. Liq. Chromatogr. Relat. Technol., 29, 2185–2200 (2006).

18. G. Török, et al., J. Pharm. Biomed. Anal., 39, 425–430 (2005).

19. M.I. Jimidar et al., Electrophoresis, 25, 2772–2785 (2004).

20. T.E. Beesley, J.T. Lee and A.X. Wang, Chiral Separation Techniques: A Practical Approach (2nd Ed.), G. Subramanian, Ed. (Wiley-VCH Verlag GmbH, Weinheim, Germany, 2001), pp 25–54.

21. Advanced Separation Technologies Inc. Chirobiotic Handbook: A guide to Using Macrocyclic Glycopeptides Bonded Phases For Chiral LC Separations (5th Ed.) (2004).

Tom Huybrechts PhD is a scientist at the API Development–Analytical Sciences group, ChemPharm Division of Johnson & Johnson Pharmaceutical Research and Development (J&J PRD) in Beerse (Belgium). His work and research interests focus on drug chemistry and analytical method development.

Gabriella Török PhD is associate director in charge of the chemical process control group within API Development–Analytical Sciences.

Tom Vennekens is an associate scientist at the Analytical Sciences group. His current work focuses on structural elucidation by mass spectrometry. He was formerly involved in analytical method development and validation.

Rudy Sneyers is a scientist at the API Development–Analytical Sciences group. His work and research interests focus on analytical method development and validation.

Sara Vrielynck PhD is a senior scientist at the API Development–Analytical Sciences group. She is in charge of drug chemistry and analytical method development.

Ivan Somers is head of the API Development–Analytical Sciences group in Beerse (Belgium).

This article was first published in LCGC Europe,20(6), 320–335 (2007).