Evaluation of Multiplexed CE with UV Detection for Rapid pKa Estimation of Active Pharmaceutical Ingredients

Xiaoyi Gong1,&, Margaret Figus1, Jolanta Plewa1, Dorothy A. Levorse2, Lili Zhou1, Christopher J. Welch1 1 2

Early Development Analytical Research, Merck & Co., Inc., Rahway, NJ, USA; E-Mail: [email protected] Pharmaceutical Research & Development, Merck & Co., Inc., Rahway, NJ, USA

Received: 17 March 2008 / Revised: 12 May 2008 / Accepted: 15 May 2008

Abstract The acid dissociation constant (pKa) is a key physicochemical parameter for characterizing active pharmaceutical ingredients (APIs). Early determination of pKa values is highly desirable in drug discovery, pharmaceutical process research and formulation design. To overcome the challenges of limited sample availability and potential low purity of API samples at early stages of drug development, as well as to increase sample analysis throughput, a multiplexed 96-channel capillary electrophoresis with UV detection was evaluated as a practical approach for high throughput pKa estimation of proprietary APIs in support of pharmaceutical research. Proprietary APIs with diverse structures were examined using the approach. The pKa values were successfully determined with good accuracy and precision. System robustness was demonstrated and analysis of at least eight samples can be completed within 1 h. A rapid pKa estimation procedure for marginally soluble APIs was proposed by performing single-point multiplexed CE–UV measurement without extrapolation using 10 or 20% methanol as co-solvent. Direct pKa estimation of APIs using DMSO solution samples and crude reaction samples containing a large amount of solvents and reagents and high level of impurities was also demonstrated using the multiplexed CE–UV approach.

Keywords Capillary electrophoresis UV detection Acid dissociation constant

Introduction The acid dissociation constant (pKa) is an important physicochemical parameter

for characterizing drug compounds. It has been estimated that 95% of all drug compounds are ionizable [1]. Early determination of pKa values is highly

desirable in drug discovery, pharmaceutical process research, and formulation design [2–4]. However, the challenges of determining pKa values at an early stage of drug development often include limited sample availability, low sample purity, and poor aqueous solubility of drug compounds. The increasing pace of drug discovery and development also demands new high throughput analytical techniques for rapid determination of critical physicochemical parameters including pKa values. The pKa of a compound is traditionally determined by potentiometric and spectrophotometric titration methods [5]. Potentiometric titration typically requires a relatively large amount of sample (2–20 mg) and high aqueous solubility of test compounds. For waterinsoluble compounds, the pKa values in aqueous solution can be determined by titration with organic co-solvents at different levels followed by extrapolation, but the glass electrode often needs to be calibrated using a complicated and time consuming procedure. Spectrophotometric titration requires much less sample and is suitable for compounds with sparing aqueous solubility, but it is not a general approach, as it cannot be applied to compounds lacking chromophores in proximity to the ionization centers. Both potentiometric and spectrophotometric

Original DOI: 10.1365/s10337-008-0691-6 Ó 2008 Vieweg+Teubner | GWV Fachverlage GmbH

R

O

N N

R

N

N N

O

N N

N

R

R'

Compound A

using a 96-capillary array CE instrument with UV detection for rapid pKa estimation. In addition, pKa measurements for API samples of poor aqueous solubility, crude reaction API samples, and DMSO solution samples were also performed using the multiplexed CE–UV approach.

R'

Compound B

Compound C H2N

NH2 R''

R N

X

X O

R

N

N

R'

O

Compound E

O S O

O N N

Compound G

R

R

O S N H O

Compound H

Experimental

O P OH

R' Compound D

R

Compound F

R

O

OH

R

O

OH N H

O

Compound I

R'

O

Compound J

Fig. 1. Structure of the ten Active pharmaceutical intermediates (APIs) used in the study

titration methods may yield inaccurate pKa values if test compounds are impure or have poor stability under the titration conditions. The pKa values can also be predicted by using various computational approaches and appropriate software, allowing the estimation of pKa values of hypothesized molecules that have not yet been synthesized or compounds that are hazardous or otherwise difficult for pKa measurement. The software prediction, however, is often unreliable for drug compounds with novel and complex structures [6, 7]. Capillary electrophoresis (CE) has in the last decade become an alternative method for pKa measurement [8–10]. This approach has gained increasing recent acceptance in the pharmaceutical industry owing to the distinct advantages of extremely low sample consumption, tolerance of impure or poorly soluble samples, and ease of automation. A number of efforts have recently been reported to increase the throughput of CE– pKa measurement. Pressure-assisted capillary electrophoresis (PACE) was developed for rapid pKa determination, [11] and a medium-throughput PACE pKa screening assay was reported to afford an estimated analysis throughput of

approximately 20 samples per day [12]. The throughput of PACE pKa measurement can be further improved when short capillaries were used [13]. A highthroughput pKa screening method based on PACE and mass spectrometry (MS) detection was reported to measure pKa values of more than 50 compounds in less than 3 h by pooling a number of compounds into a single sample [14]. More recently, vacuum assisted multiplexed 96channel capillary electrophoresis with UV detection [15] has been reported for high throughput pKa screening [8, 16–19]. A series of 96 commercially available water-soluble compounds were tested using a 96-capillary array CE instrument and the measured pKa values were compared and agreed well with the literature reported pKa values. The multiplexed CE–UV approach allowed the pKa estimation for between 128 and 168 compounds in an 8-h period. The multiplexed CE–UV approach was also assessed recently for a number of high-throughput pharmaceutical analysis applications including tablet assay, pKa determinations, and logP determinations [18]. In the present report, a collection of 10 proprietary APIs (Fig. 1) with one or more ionizable groups were examined

The 10 API samples used in the study (Compounds A–J) were obtained internally from Merck & Co., Inc. Methanol, DMSO, THF, pyridine, hydrochloric acid, sodium hydroxide, potassium hydroxide, potassium hydrogen phthalate, potassium chloride were purchased from Sigma-Aldrich (Milwaukee, WI, USA). The deionized (DI) water was filtered using a Mini-Q filter system (Millipore, Bedford, MA, USA). For multiplexed CE-UV pKa measurements, 24 aqueous CE running buffers covering a pH range from 1.7 to 11.2 with leveled ionic strength of I = 50 mM, and a 109 outlet reservoir buffer concentrate (100 mM sodium tetraborate) were obtained from CombiSep (Ames, IA, USA). The 109 buffer concentrate was diluted tenfold with DI water prior to use to prepare a 10 mM sodium tetraborate outlet reservoir buffer with a measured pH of 9.3. For pKa estimation of the water soluble APIs (Compounds A–F) using aqueous CE running buffers, the 24 CE running buffers were used as received unless specified otherwise, and the samples were dissolved in DI water with 0.1% DMSO at a target concentration of 0.05 mg mL-1. Samples that did not completely dissolve in DI water after sonication were filtered to remove undissolved material. For pKa estimation of the water insoluble APIs (Compounds G–J), four sets of 24-point organic cosolvent CE buffers containing 30, 40, 50, and 60% methanol (v/v) respective for each set were also purchased from CombiSep. The cosolvent CE buffers covers apparent pH ranges from 1.8 to 11.2 with leveled ionic strength of I = 50 mM. The procedure to calibrate pH values and ionic strength of the cosolvent CE buffers was described in reference [19]. For the experiments to rapidly estimate the pKa values of marginally soluble APIs by

Original

performing single-point multiplexed CE– UV measurement without extrapolation, 10 or 20% methanol (v/v) was added to the 24 aqueous CE running buffers. The pH values of the 24 running buffers were not adjusted and the aqueous pH values were used directly in these experiments for quick pKa estimation. The samples were dissolved in DI water and methanol mixtures with the same methanol compositions as the CE running buffers. A multiplexed 96channel CE system (CombiSep) with 96 bare silica capillaries (28 cm effective length, 54 cm total length, 75 lm i.d.) and a Zinc lamp (CombiSep) for UV detection at 214 nm were used for pKa measurement. Before the first run each day, the capillaries were rinsed sequentially for 5 min with 0.1 N sodium hydroxide, DI water, and 10 mM tetraborate buffer, as recommended by the vendor. Following the rinse step, the CE running buffers were pulled through the capillaries using a vacuum (-2.0 psi) for 6 min from a 96-well microtitre buffer tray. Samples were then injected hydrodynamically by applying a vacuum of -0.8 psi for 5 s from another 96-well microtitre sample tray. The buffer tray was then replaced, and a 4.5 kV voltage was applied together with a vacuum (-0.2 psi) to achieve separation. The experiments were performed at uncontrolled room temperature (*22 °C) due to lack of precise temperature control function for the multiplexed CE instrument. Between injections, the capillaries were conditioned with 10 mM tetraborate buffer for 5 min. After the separation, the pKa values were calculated using the pKa Estimator software from CombiSep. The pKa values of Compound E were also determined by potentiometric titrations with the Sirius GLpKa/D-PAS instrument using a double junction electrode. The electrode was standardized from pH 1.8 to 12.2. The KOH titrant was standardized against potassium hydrogen phthalate and was approximately 0.5 M. The sample was dissolved in about 8 mL of ionic strength adjusted (ISA) H2O (0.15 M KCl). The starting pH of the solution was adjusted with 0.5 M KOH. The solution was titrated at 25 °C in ISA H2O from pH 2 to 10. Several titrations of each compound were performed. The data was Original

analyzed using RefinementPro v.2.2 software. The pKa values of Compounds A–D and Compound F were determined by spectrophotometric titrations. The titrations were performed with the Sirius GLpKa/D-PAS using a double junction electrode. The electrode was standardized from pH 1.8 to 12.2. The KOH titrant was standardized against potassium hydrogen phthalate and was approximately 0.5 M. The sample was dissolved in MeOH to create a stock solution of *5 mg mL-1. An aliquot of the stock solution (0.05 mL) was added to the titration vial and diluted up to 10 mL ISA H2O. The starting pH of the solution was adjusted with 0.5 M HCl and the solution was titrated from about pH 2 to 11 at 25 °C. Several titrations of the compound were performed. For spectrophotometric analysis, spectra are recorded from 200 to 700 nm. The data was analyzed using RefinementPro v.2.2 software. The apparent pKa values of Compounds G–J were determined using the similar spectrophotometric titration procedure with methanol co-solvent at 50, 40, and 30 v/v%, followed by Yasuda-Shedlovsky extrapolation to determine the aqueous pKa values. The pKa values of the ten APIs were also estimated by using the commercial pKa prediction software ACD/pKa DB version 8.0 from ACD/Labs (Toronto, Canada).

Results and Discussion In a previous report, Zhou and coworkers estimated the pKa values of a series of 96 commercially available compounds using the multiplexed CEUV approach, the results generally agreeing with literature values [16]. As the compound group in that study was made-up primarily of water soluble commercial compounds, we were interested in evaluating the utility of the technique for estimating pKa values of the compounds of greatest interest in pharmaceutical research and development—the more structurally complex APIs that often have poorer water solubility or multiple ionizable groups.

Estimation of pKa values with CE relies on the dependence of electrophoretic mobility of an ionizable analyte on the extent of ionization. Comparison of the electrophoretic mobility of an analyte in a series of run buffers spanning a wide pH range clearly shows the dependence of migration as a function of pH, and can be plotted to show what is in effect a titration curve. The principal advantage of the multiplexed CE system is that it allows the entire group of run buffers at the different pH values to be run simultaneously. In the current study, the pKa values of ten proprietary APIs (Fig. 1) were estimated using the multiplexed CE–UV approach. The ten APIs have diverse and relatively complex structures with a molecular weight range of 250–760 amu, including compounds that are monobasic (Compounds A, B, G), monoacid (H–J), dibasic (Compounds C–E), and diacidic (Compound F). The aqueous solubility of the ten APIs ranged from soluble (Compound E) to marginally soluble (A–D, F) to insoluble (Compounds G–J). Compounds A–D and Compound F have marginal aqueous solubility (H < 1 mg mL-1) but were still able to be detected in aqueous solutions for pKa measurement by the multiplexed CE–UV instrument. For the water-insoluble APIs (Compounds G–J), the apparent pKa values were estimated using 30, 40, 50, and 60% methanol (v/v) as co-solvent to enhance solubility, followed by Yasuda-Shedlovsky extrapolation to determine aqueous pKa. Yasuda-Shedlovsky (Y-S) extrapolation relates the measured apparent pKa values from the methanol-aqueous solutions to the inverse of the dielectric constants (1/e) of the binary solvents through a typically linear relationship. Extrapolation to e = 78.3 (the dielectric constant of pure water) yields aqueous pKa. The Yasuda-Shedlovsky extrapolation in this study was performed using the pKa Estimator software from CombiSep, according to the procedure detailed in reference [19]. The pKa values of the ten APIs were also estimated by traditional titration methods, the industry standard for pKa measurement. Only the water soluble Compound E can be measured directly

Table 1. Comparison of pKa values estimated by titration, multiplexed CE (MCE), and ACD/pKa DB software prediction for ten proprietary APIs Compound

Acidity/basicity

Compound A Compound B

Monobase Monobase

+ +

Compound C pKa1 pKa2

Dibase

+

Compound D pKa1 pKa2

Dibase

Compound E pKa1 pKa2

Dibase

Compound F pKa1 pKa2

Diacid

Compound Compound Compound Compound

Monobase Monoacid Monoacid Monoacid

G H I J

Solubility

Titration pKa

MCE pKa

Diff. MCE

ACD pKa

Diff. ACD

8.95 4.67

8.79 4.77

-0.16 +0.10

9.36 3.68

+0.41 -0.99

2.30 4.97

2.24 5.07

-0.06 +0.10

2.05 7.11

-0.25 +2.14

4.72 9.99

4.49 10.25

-0.23 +0.26

6.56 10.35

+1.84 +0.36

4.88 7.35

4.86 7.40

-0.02 +0.05

5.80 9.16

+0.92 +1.81

2.16 7.06

2.40 6.86

+0.24 -0.20

1.64 6.34

-0.52 -0.72

6.55 5.47 6.35 4.29

6.35 5.28 6.54 4.27

-0.20 -0.19 +0.19 -0.02

5.28 4.57 5.83 4.06

-1.27 -0.90 -0.52 -0.23

+

++

+

-

++ Soluble, + marginally soluble, - insoluble

Table 2. Precision of pKa measurement by multiplexed CE Compound

Average pKa (n = 3)

Standard deviation

Compound A Compound B

8.79 4.77

0.02 0.02

Compound C pKa1 pKa2

2.24 5.07

0.03 0.02

Compound D pKa1 pKa2

4.49 10.25

0.02 0.03

by potentiometric titration in aqueous solutions for pKa estimation. Marginally water soluble Compounds A–D and Compound F were measured by spectrophotometric titration in aqueous solutions. Methanol co-solvent and extrapolation were needed for pKa measurement of water-insoluble Compounds G–J by spectrophotometric titration. The results of the study (Table 1) show generally good agreement between the pKa values of each of the APIs determined by potentiometric or spectrophotometric titration and those estimated using the multiplexed CE-UV approach, with an average absolute deviation of 0.14 pH units. This level of agreement between the two approaches is comparable to what was previously

reported in the comparison of literature reported pKa values of commercial compounds with those obtained using the multiplexed CE–UV approach [16]. Most notably, good agreement between the CE approach and potentiometric or spectrophotometric titration was observed consistently for all the APIs with a wide pKa range from 2.1 to 10.0, including dibasic and diacidic compounds and even those compounds with very poor water solubility. Software prediction of pKa values is often useful in drug discovery and development when samples of drug compounds are not available or where measurement is otherwise difficult. For the ten APIs, the predicted pKa values using the commercially available pKa

prediction software, ACD/pKa DB, are also listed and compared to the titration values in Table 1. For this group of test analytes, the absolute deviations of the predicted pKa values from the titration values ranged from 0.23 to 2.14 with an average of 0.91, which suggested that software pKa prediction yielded useful but often less reliable results comparing to the multiplexed CE–UV approach. This result was consistent with our general observation that software prediction of pKa values is better suited to simple molecular structures, and often yields less accurate results for the novel and more complex structures such as those included in our study group. It is important to note that the multiplexed CE–UV pKa measurement requires only a minimum amount of sample (50–100 lg) and offers a speedy analysis. It is therefore a preferable approach to software pKa prediction in the cases where samples exist and accurate pKa estimation is needed. The precision of the multiplexed CEUV approach for pKa measurement was investigated for the four APIs (Compounds A–D) as listed in Table 2. The precision of three consecutive runs was considered satisfactory for all the compounds with the standard deviation Original

smaller than 0.05 pH units for all the pKa values. The four APIs were also examined over a 2-day period to determine robustness of the multiplexed CE– UV approach for pKa estimation of the proprietary APIs. The between-day reproducibility was found to be sufficient with variations ranging from 0.02 to 0.21 pH units, as shown in Table 3. Although the pKa values of waterinsoluble compounds can be accurately measured by the multiplexed CE–UV approach using an organic co-solvent and extrapolation, preparing organicaqueous mixture buffer solutions and performing analysis at multiple organicaqueous compositions is somewhat time consuming. By adding 10–20% methanol as a co-solvent into buffer solutions, the solubility of many marginally watersoluble drug compounds may be enhanced enough for detection and pKa estimation by the multiplexed CE–UV instrument. The pKa values of the four APIs (Compounds A–D) measured by the multiplexed CE–UV instrument with 0, 10, and 20% methanol added in buffer solutions and sample diluents are shown in Table 4. With 10% methanol cosolvent, the deviation of the pKa values from those measured in aqueous buffer solutions ranged from 0.06 to 0.28 pH units with an average value of 0.17. With 20% methanol co-solvent, the deviation was slightly higher ranging from 0.06 to 0.51 units with an average value of 0.32. The relatively small pKa deviation observed at both methanol co-solvent levels was acceptable for many practical applications in pharmaceutical research and development that often require only rough pKa values (e.g. chemical process development, formulation design, chromatography method development). The results suggested that for some marginally water-soluble drug compounds, the pKa values in aqueous solutions may be quickly estimated in a single multiplexed CE–UV measurement by directly using 10–20% methanol co-solvent, rather than performing multiple measurements at different methanol co-solvent levels followed by extrapolation. It was also noted that in these experiments the aqueous pH values of the CE buffers were used directly for pKa estimation after adding 10 or 20% Original

Table 3. Between-Day Variation of pKa Measurement by Multiplexed CE pKa day 1a

Compound

a

pKa day 2a

Variation

Compound A Compound B

8.79 (0.24) 4.77 (0.36)

8.85 (0.46) 4.80 (0.43)

0.06 0.03

Compound C pKa1 pKa2

2.24 (1.34) 5.07 (0.41)

2.29 (5.91) 5.09 (0.79)

0.05 0.02

Compound D pKa1 pKa2

4.49 (0.51) 10.25 (0.30)

4.69 (4.5) 10.46 (0.94)

0.20 0.21

Values in parentheses are within-day RSD% of pKa values (n = 3)

Table 4. pKa measurement by multiplexed CE–UV with 10–20% methanol co-solvent Compound

Aqueous buffer

10% methanol

Deviation 10%

20% methanol

Deviation 20%

Compound A Compound B

8.79 4.77

8.73 4.49

0.06 0.28

8.47 4.26

0.32 0.51

Compound C pKa1 pKa2

2.24 5.07

1.96 4.95

0.28 0.12

2.01 4.73

0.23 0.34

Compound D pKa1 pKa2

4.49 10.25

4.33 10.15

0.16 0.10

4.43 9.79

0.06 0.46

Table 5. Direct pKa measurement of crude API samples by multiplexed CE

a b

Compound F

pKa1a n=4

Deviation pKa1

pKa2a n=4

Deviation pKa2

Pure Sample THF Solutionb 1% Pyridine, THFb 2 M Acid (HCl), THFb 2 M Base (NaOH), THFb DMSO Solutionb

2.40 2.36 2.35 2.37 2.37 2.35

0.04 0.05 0.03 0.03 0.05

6.86 6.92 6.89 7.02 6.89 6.89

0.06 0.03 0.16 0.03 0.03

(0.6) (0.2) (1.4) (0.4) (0.7) (0.9)

(0.1) (0.2) (0.5) (0.4) (0.5) (0.3)

Values in parentheses are within-day RSD% of pKa values (n = 4) Spiked with the penultimate impurity at 25% of Compound F

methanol without further pH calibration in aqueous-organic mixtures. Recalibration of pH values in aqueous-organic mixtures of the CE buffers could further improve the accuracy of the quick pKa estimation approach. Traditional, pKa measurement of APIs requires pure samples in solid form. Direct measurement using crude reaction samples is often difficult, owing to interference from impurities, reagents, and solvents with potentiometric and spectrophotometric titration assays. Meanwhile it is often desirable to know the pKa value beforehand to design an

optimal process (crystallization, extraction, chromatography, etc.) to isolate and purify a compound from crude reaction samples. It would be convenient for process chemists if pKa values could be estimated using crude reaction samples, before pure compounds are isolated and available in pure form. Furthermore, archived compounds stored in drug discovery compound repositories are often stored as DMSO solutions and may contain degradants or other impurities. As pure solid samples of each compound are not always available, it would be advantageous to use DMSO

unknown impurity was also separated from Compound F (Fig. 2) and did not interfere with the pKa measurements. It is interesting to note that even though the penultimate impurity was not well resolved from Compound F under most of the CE conditions, the co-elution did not significantly diminish the ability to accurately determine the migration times of Compound F, and to use these values to correctly estimate pKa. For the DMSO solution sample, the large amount of DMSO in the injections caused peak broadening of the DMSO peaks that are used as a neutral marker to calibrate the migration times, but the effect on the estimated pKa values and measurement precision is negligible.

Conclusions

Fig. 2. a Twenty-four electropherograms of Compound F, spiked with 25% penultimate impurity, in THF solution with 1 v/v% pyridine obtained by multiplexed CE–UV with CE running buffer pHs labeled on each electropherogram. b Electropherogram at pH 4.40 CE running condition

solution samples directly for pKa measurements of these compounds. To evaluate the feasibility of direct pKa measurement using crude reaction samples and DMSO solutions, samples of Compound F were prepared in THF and DMSO (*10 mg mL-1) and spiked with the penultimate intermediate as an impurity at 25% of target concentration. In addition, samples containing 2 M HCl, 2 M NaOH, or 1 v/v% pyridine were also prepared in THF to imitate the crude reaction samples. The prepared sample solutions were then diluted to 0.05 mg mL-1 with the sample diluent and analyzed by the

multiplexed CE-UV for pKa estimation. The estimated pKa values thus obtained were compared to those previously measured using pure samples, as listed in Table 5. Only minimum interferences of the penultimate impurity, solvents (THF, DMSO), acid (2 M HCl), base (2 M NaOH), or 1% pyridine on the correct estimation of pKa using the multiplexed CE–UV pKa were observed. As shown in Fig. 2, pyridine was well separated from Compound F under the CE conditions, and it was actually possible to determine the pKa of pyridine simultaneously using the same multiplexed CE–UV run. Another

In this report, multiplexed 96-channel CE–UV was demonstrated to be a practical approach for rapid pKa estimation of proprietary APIs with novel structures. At least eight API samples can be analyzed for pKa estimation within 1 h using this approach. The pKa values of the ten APIs estimated by the multiplexed CE–UV approach agreed well with the potentiometry values with an average absolute deviation of 0.18 pH units. The pKa measurement precision and system robustness were also examined and considered to be sufficient. A quick pKa estimation procedure for marginally soluble APIs was proposed by performing single-point multiplexed CE–UV measurement without extrapolation using 10–20% methanol as cosolvent, and the accuracy of the results were found to be acceptable for many applications. Direct pKa measurement of APIs using crude reaction samples containing large amount of solvents and reagents and high level of impurities, as well as a DMSO solution sample was also demonstrated to be feasible using the multiplexed CE–UV approach.

References 1. Wells JI (1998) In: Pharmaceutical preformulation, 1st edn. Eills Hoowood Ltd, London

Original

2. Kibbey CE, Poole SK, Robinson B, Jackson JD, Durham D (2001) J Pharm Sci 90:1164–1175. doi:10.1002/jps.1070 3. Curatolo W (1998) Pharm Sci Technol Today 1:387–393. doi:10.1016/S1461-5347 (98)00097-2 4. Kramer SD (1999) Pharm Sci Technol Today 2:373–380. doi:10.1016/S1461-5347 (99)00188-1 5. Albert A, Serjeant EP (1984) In: The determination of ionization constants: a laboratory manual. Chapman & Hall, New York 6. Perrin DD, Dempsey B, Serjeant EP (1981) In: pKa prediction for organic acids and bases. Chapman & Hall, London 7. Kogej T, Muresan S (2005) Curr Drug Discov Technol 2:221–229. doi:10.2174/ 157016305775202964

Original

8. Wan H, Ulander J (2006) Expert Opin Drug Metab Toxicol 2:139–155. doi: 10.1517/17425255.2.1.139 9. Poole SK, Patel S, Dehring K, Workman H, Poole CF (2004) J Chromatogr A 1037:445–454. doi:10.1016/j.chroma.2004. 02.087 10. Jia ZJ (2005) Curr Pharm Anal 1:41–56. doi:10.2174/1573412052953319 11. Ishihama Y, Nakamura M, Miwa T, Kajima T, Asakawa N (2002) J Pharm Sci 91:933–942. doi:10.1002/jps.10087 12. Jia ZJ, Ramstad T, Zhong M (2001) Electrophoresis 22:1112–1118. doi:10.1002/ 1522-2683()22:63.0.CO;2-1 13. Wan H, Holmen A, Nagard MB, Lindberg W (2002) J Chromatogr A 979:369– 377. doi:10.1016/S0021-9673(02)01262-1

14. Wan H, Holmen A, Wang YD, Lindberg W, Englund M, Nagard MB et al (2003) Rapid Comm Mass 17(Spec):2639–2648. doi:10.1002/rcm.1229 15. Gong X, Yeung ES (1999) Anal Chem 71:4989–4996. doi:10.1021/ac9900634 16. Pang HM, Kenseth J, Coldiron S (2004) Drug Discov Today 9:1072–1080. doi: 10.1016/S1359-6446(04)03293-3 17. Zhou CX, Jin YK, Kenseth JR, Stella M, Wehmeyer KR, Heineman WR (2005) J Pharm Sci 94:576–589. doi:10.1002/jps. 20275 18. Marsh A, Altria K (2006) Chromatographia 64:327–333. doi:10.1365/s10337-0060023-7 19. Shalaeva M, Kenseth J, Lombardo F, Bastin A (2008) J Pharm Sci (online)