A

twmogtpo©

KS

1

y[napos

0d

Journal of Chromatography A, 1163 (2007) 228–236

Analytical and semipreparative enantioseparation of 9-hydroxyrisperidone,the main metabolite of risperidone, using high-performance liquid

chromatography and capillary electrophoresisValidation and determination of enantiomeric purity

Cecile Danel a, Christine Barthelemy b, Dalila Azarzar a, Hugues Robert b,Jean-Paul Bonte a, Pascal Odou b, Claude Vaccher a,∗

a Laboratoire de Chimie Analytique EA 4034, Faculte des Sciences Pharmaceutiques et Biologiques, Universite de Lille 2,B.P. 83, 3 rue du Pr. Laguesse, 59006 Lille Cedex, France

b Laboratoire de Biopharmacie et Pharmacie Clinique, EA 4034, Faculte des Sciences Pharmaceutiques et Biologiques,Universite de Lille 2, B.P. 83, 3 rue du Pr. Laguesse, 59006 Lille Cedex, France

Received 12 March 2007; received in revised form 6 June 2007; accepted 13 June 2007Available online 22 June 2007

bstract

The HPLC semipreparative enantioseparation of 9-hydroxyrisperidone (9-OHRisp) was studied by optimizing various experimental conditions:he nature of the chiral stationary phase (CSP), mobile phase composition, temperature and analyte loading. This semipreparative enantioseparationas successfully completed using the polysaccharide Chiralcel OJ chiral stationary phase and a n-hexane/ethanol/methanol (50/35/15, v/v/v) ternaryobile phase. To assess the enantiomeric purity of both isolated isomers, three analytical methods using UV detection were developed and validated:

ne CE method using dual cyclodextrin mode and two HPLC methods using either the Chiralcel OJ CSP in normal-phase mode or the �-acidlycoprotein (�-AGP) CSP in reversed-phase mode. The three methods make it possible to obtain excellent enantioseparations (Rs > 3) with analysisimes lower than 15 min, and the calculated limits of detection allow for the determination of minor enantiomeric impurities (0.1%). Enantiomeric

urity obtained for dextrorotatory and levorotatory enantiomers was superior to 99.9% and equal to 98.9%, respectively, which proved the successf the semipreparative enantioseparation. A brief comparison of the performances of the analytical methods completes this work.

2007 Elsevier B.V. All rights reserved.

eywords: Risperidone; 9-Hydroxyrisperidone; Chiral liquid chromatography; Chiral capillary electrophoresis; Polysaccharides; �-Acid glycoprotein; Cyclodextrins;

as[p(c[

emipreparative enantioseparation; Enantiomeric purity

. Introduction

Risperidone (Risp), 3-{2-[4-(6-fluoro-1,2-benzisoxazol-3-l)-1-piperidin]ethyl}-6,7,8,9-tetrahydro-2-methyl-4H-pyrido-1,2-�]pyrimidine-4-one (Risperdal, Janssen-Cilag Inter-ational, Issy-les-Moulineaux, France), is a benzisoxazoletypical antipsychotic agent approved for the treatment of

sychosis (including schizophrenia) [1] and also some formsf bipolar disorder [2], psychotic depression [3] and Touretteyndrome [4]. It is a selective monoaminergic antagonist with

∗ Corresponding author.E-mail address: claude.vaccher@univ-lille2.fr (C. Vaccher).

cf(sEPo

021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2007.06.023

strong affinity for serotonin type 2 (5-HT2) receptors and alightly weaker affinity for dopamine type 2 (D2) receptors5]. Risp is metabolized in the liver by cytochrome P450 [6] toroduce mainly the active 9-hydroxyrisperidone (9-OHRisp)Fig. 1): “active moiety” during risperidone treatment isonsidered to be the sum of Risp and 9-OHRisp concentrations7]. 9-Hydroxylation results in the formation of a stereogenicenter yielding two enantiomers. The marketing of its racemicorm is expected under the International Non-proprietary NameINN) paliperidone. In May 2006, Janssen-Cilag International

ubmitted a Marketing Authorization Application to theuropean Medicines Evaluations Agency for Paliperidonerolonged Release Tablets, an oral medication for the treatmentf schizophrenia to be taken once daily [8].

C. Danel et al. / J. Chromatogr. A 1163 (2007) 228–236 229

sperid

cAf9(9Csdfcbnio9atmbmup[

aepsbbvar

fctwO(or(

H

riroappwdtt

2

2

w

wopD�S(LfFpF

2

Hs1CpMCc

Fig. 1. Chemical structures of (a) ri

It is widely recognized that stereochemistry plays a cru-ial role in the interaction of drugs with biological targets.s an enantioselective metabolism has already been reported

or the 9-OHRisp by Yasui-Furukori et al. [9], the (+)--hydroxylation seems to be the major metabolic pathwaymainly by Cytochrome P450 2D6 (CYP2D6)) whereas (−)--hydroxylation might be achieved to a slight extent (mainly byYP3A4), the preparative enantioseparation of the 9-OHRisp

eems an essential step in this drug development. The onlyescribed preparation of 9-OHRisp enantiomers described soar is the one reported in the US patent [10] by its inventors byrystallizing diastereoisomeric salts with (+)-3,4-dihydro-1H-2-enzopyran-2-carbonyl chloride in dichloromethane. However,o information is given about the analytical method makingt possible to determine the enantiomeric purity. To the bestf our knowledge, only two analytical enantioseparations of-OHRisp have been reported so far, both using HPLC andchiral stationary phase (CSP) �-acid glycoprotein (�-AGP):

he method developed by Yasui-Furukori et al. [9,11] with aobile phase consisted of 85% (v/v) of potassium phosphate

uffer 50 mM pH 6.5 and 15% (v/v) of methanol and theethod described by the user’s guide of the Chiral-AGP col-

mn with a mobile phase consisted of 98 % (v/v) of sodiumhosphate buffer 10 mM pH 6.0 and 2% (v/v) of 2-propanol12].

HPLC using CSPs has become the most effective methodvailable for isolating both enantiomers and determining theirnantiomeric purities. Among the large class of CSPs, theolysaccharides represent one of the most successful chiralelectors: it appears that up to 90% of racemic compounds cane resolved with not more than four different polysaccharide-ased phases [13]. Moreover, they can be used with a largeariety of mobile phases (in normal and reversed-phase modes)nd have considerable loading capacities to perform preparativeesolutions.

On the analytical scale, the other most widely used techniqueor enantioseparation is capillary electrophoresis (CE) addingyclodextrins (CD) as chiral selectors in the background elec-rolyte (BGE). Previously, a capillary electrophoretic methodas successfully developed for the enantioseparation of the 9-HRisp in the dual CD mode using neutral and anionic CDs

hydroxypropylated-�-CD and sulfated-�-CD); the optimizingf the method was based on a central composite design and

esponse surface methodology and resulted in high resolution3.13) and in a short run time (about 13 min) [14].

The aims of this study were firstly to optimize an analyticalPLC method so as to obtain the semipreparative enantiosepa-

c1pO

one and (b) 9-hydroxyrisperidone.

ation of 9-OHRisp: the various steps in developing the methodnvolved optimizing the experimental parameters (polysaccha-ide CSPs, mobile phases and temperature) and making anverloading study. Secondly, the analytical HPLC enantiosep-ration of 9-OHRisp using Chiralcel OJ CSP in the normalhase mode and the �-AGP glycoprotein CSP in the reversedhase mode as well as our previously described CE method [14]ere validated (in terms of repeatability, linearity and limits ofetection (LOD) and quantification (LOQ)) to quantify the enan-iomeric purity of the isolated enantiomers and in a final stagehey were compared.

. Materials and methods

.1. Chemicals

Risperidone and its main metabolite 9-hydroxyrisperidoneere a gift from Janssen-Cilag.Ethanol, 1-propanol, 2-propanol, methanol and n-hexane

ere HPLC grades from Merck (Fontenay-sous-bois, France)r Carlo Erba (Val de Reuil, France). Sodium hydroxide andhosphoric acid were purchased from Baker (Paris, France).iethylamine, triethylamine, dimethylamine and the sulfated--CD (with MS 6–11 per CD molecule) were purchased fromigma-Aldrich (Saint Quentin Fallavier, France). HP-�-CDswith MS 5-7 per CD molecule) were a gift from the Roquetteaboratories (Lestrem, France). Deionized water was obtained

rom a Milli-Q system (Millipore, Saint Quentin-en-Yvelines,rance). Polyethylene oxide (PEO) (0.4%, Mw = 300,000) wasurchased from Beckman (Beckman Coulter France, Villepinte,rance).

.2. Instrumentation

Chromatographic analyses were performed on a Waters 600PLC system equipped with a Waters 996 photodiode array

pectrophotometer, a 7125 Rheodyne injector (with 20, 50,00 or 200 �L loop), and a Waters in-line degasser apparatus.hromatographic data were collected and processed on a com-uter running with Empower software from Waters (Milford,A, USA). Chiral analyses were performed on polysaccharidehiralpak AD amylose column (tris-3,5-dimethylphenyl-arbamate; 250 × 4.6 mm I.D.; 10 �m), Chiralpak AS amylose

olumn (tris-(S)-1-phenylethylcarbamate; 250 × 4.6 mm I.D.;0 �m), Chiralcel OD-H cellulose column (tris-3,5-dimethyl-henylcarbamate; 250 × 4.6 mm I.D.; 5 �m) and ChiralcelJ cellulose column (tris-4-methylbenzoate; 250 × 4.6 mm

2 atogr

IoC

a0mppsfAa1vdtacoU1

cacfwfT2n(1wwspaBBpSpt

3

3t

3

ipra

ticO

2((talminaUhisa(aOc

1iett1organic modifier. The use of 2-propanol instead of 1-propanolresults in a significant increase in 9-OHRisp retention and enan-tioselectivity: for n-hexane/alcohol (70/30, v/v) eluents, k1 and α

Table 1Analytical chromatographic parameters: retention factor (k1), enantioselectivityfactor (�) and enantioseparation (Rs) of 9-OHRisp on amylose or cellulose CSPswith n-hexane/alcohol (70/30, v/v) eluents

k1 α Rs

Chiralpak ADn-Hexane/ethanol 7.02 1.12 <0.5n-Hexane/1-propanol 2.08 1.00 n.r.n-Hexane/2-propanol 1.76 1.00 n.r.

Chiralpak ASn-Hexane/ethanol 0.75 1.00 n.r.n-Hexane/1-propanol 0.78 1.00 n.r.n-Hexane/2-propanol 1.46 1.00 n.r.

Chiralcel OD-Hn-Hexane/ethanol 1.03 1.15 <0.5n-Hexane/1-propanol 1.14 1.00 n.r.n-Hexane/2-propanol 1.85 1.00 n.r.

Chiralcel OJ

30 C. Danel et al. / J. Chrom

.D.; 10 �m) (Daicel Chemical Industries, Baker France) orn a �-AGP column (150 × 4 mm I.D.; 5 �m) (ChromTech,ongleton, UK).

Compounds were chromatographed by dissolving them in thelcohol of the corresponding mobile phase to a concentration of.50 mM (otherwise specified) and passed through a 0.45 �membrane filter prior to column loading. Chromatography was

erformed at 30 ◦C (otherwise, specified) with a constant mobilehase flow of 0.8 mL min−1. The column void time (t0) was con-idered to be equal to the peak of the solvent front and was takenrom each particular run. It was about 4.00 min for ChiralpakD, 4.08 min for Chiralpak AS, 3.90 min for Chiralcel OD-H

nd 4.05 min for Chiralcel OJ (0.8 mL min−1), and it was about.95 min for the �-AGP column. Retention times were meanalues of duplicate determinations. Risp and 9-OHRisp wereetected at 205, 238 and 280 nm. The semipreparative enan-ioseparation of 9-OHRisp was performed on the Chiralcel OJnalytical column. Optical rotations of methanol solutions (con-entration of 2 mg mL−1) using the Na D line (589 nm) werebtained using a Perkin-Elmer 241 polarimeter (Boston, MA,SA). The volume of the cell and optical path were 1 mL and0 cm, respectively. Measurements were performed at 20 ◦C.

CE experiments were performed on a Beckman P/ACE MDQapillary electrophoresis system, including an on-column diode-rray UV-detector. The whole system was driven by a personalomputer with a 32 Karat software (Beckman Coulter) packageor system control, data collection and analysis. It was equippedith a 50.2 cm (40 cm effective length) × 50 �m I.D. untreated

used-silica capillary (Composite Metal Services, Ilkley, UK).he capillary was mounted on a cartridge and thermostated at98 ± 0.1 K, and the applied voltage was 20 kV. A hydrody-amic injection was made with a 5 s injection time at 1.0 psiotherwise specified). Risp and 9-OHRisp were detected at90, 234 and 270 nm. New capillaries were flushed for 20 minith 0.1 M sodium hydroxide (NaOH) (P = 20 psi) and 5 minith water (P = 20 psi). Each day the capillary was flushed

uccessively with NaOH (5 min, 20 psi), water (1 min, 20 psi),olyethylene oxide (PEO) (1 min, 20 psi), water (1 min, 20 psi)nd then with background electrolyte (BGE) (3 min, 20 psi).etween each run, it was treated with water (1 min, 20 psi) andGE (3 min, 20 psi). Phosphate buffers were prepared from ahosphoric acid solution adjusted to pH 2.5 by addition of TEA.ample solutions at 100 mg L−1 (otherwise specified) in 2.5 mMhosphate buffer (pH 2.5) were obtained from methanolic solu-ions at 1 g L−1.

. Results and discussion

.1. Optimization of an HPLC analytical method andransposition to the semipreparative mode

.1.1. Analytical optimization using polysaccharide CSPsOptimal experimental conditions were chosen after screen-

ng various parameters: nature of the CSP, nature of the mobilehase and temperature. The choice of the optimal conditionsesults from a compromise between higher resolutions and lowernalysis times. Since they have proved their adequacy in enan-

nnn

C

. A 1163 (2007) 228–236

ioresolving many chiral compounds in the normal phase mode,nvestigation was made into the ability of the four amylose andellulose CSPs – Chiralpak AS and AD and Chiralcel OD-H andJ – to resolve the enantiomers of 9-OHRisp.First, for each CSP, n-hexane/alcohol (ethanol, 1-propanol or

-propanol) mobile phases were tested in the range 80/20–50/50v/v). Table 1 lists the best results obtained when using 70/30%v/v) mobile phases. Both amylose CSPs AD and AS provedheir inability to enantioresolve 9-OHRisp since no enantiosep-ration was obtained with the Chiralpak AS and only a veryow one with the Chiralpak AD using the n-hexane/ethanol

obile phases. Even if the results obtained with OD-H CSP arensufficient, the low 9-OHRisp–CSP interactions observed with-hexane/ethanol eluents result in enantioselectivity (k2 and α

re 1.17 and 1.15, respectively, for the 70/30 (v/v) mobile phase).sing OJ CSP, enantioselectivity is obtained whatever the alco-ol used. The use of OJ CSP instead of OD-H leads to a hugencrease in retention of almost 200%, associated with a con-iderable improvement in enantioselectivity. For example, withn-hexane/ethanol (70/30, v/v) mobile phase, the parameters

k1, k2, α) are (1.03, 1.17, 1.15) and (2.86, 4.71, 1.64) on ODnd OJ, respectively. Based on these previous results, ChiralcelJ was selected for the following optimizing of mobile phase

omposition (nature and percentage of the alcohol).The change of the mobile phase modifier from ethanol to

-propanol, which leads to a decrease of the solvent polar-ty, results in an increase in retention (in accordance with thenhancement of the hydrogen bonding between the analyte andhe CSP) which is associated with a decrease in enantioselec-ivity: for n-hexane/alcohol (70/30, v/v) eluents, k1 increases4% and α decreases 9% using 1-propanol instead of ethanol as

-Hexane/ethanol 2.86 1.64 1.10-Hexane/1-propanol 3.27 1.50 0.80-Hexane/2-propanol 5.10 2.33 1.15

onditions: flow rate, 0.8 mL min−1; temperature, 30 ◦C; detection, λ = 205 nm.

atogr. A 1163 (2007) 228–236 231

itrsoBsti

weic41do[miosta5c

oatCos

rac

rtdmt

FHn

eAtrtpp(aoF

p

TCw

M

nnnnnnn

C

C. Danel et al. / J. Chrom

ncrease of 55% using ramified instead of linear propanol. Sincehe polarity of both alcohols is similar, the observed changes inetention and stereoselectivity are more probably due to theirteric difference and the alteration in the geometry and/or sizef the chiral cavities that can be induced by bulky alcohols [15].ecause ethanol gives satisfactory enantioselectivities in the

hortest analysing times, higher throughput of the semiprepara-ive method can be expected with this organic modifier, and sot was selected.

The amount of ethanol in the n-hexane/alcohol mobile phaseas studied in the range 20–50% (v/v). When the percentage of

thanol in the eluent increases, a decrease in the retention factorss observed while enantioselectivity remains almost constant: thehromatographic parameters (k1, k2,α) obtained with 20%, 30%,0% and 50% (v/v) of ethanol are (5.22, 8.93, 1.71), (2.84, 4.69,.65), (1.49, 2.77, 1.86) and (0.82, 1.52, 1.85), respectively. Theecrease of the retention can be related to the increasing abilityf the more polar eluent to displace the analyte from the CSP16] (through the formation of hydrogen bonds between the polarodifier and the CSP) and the higher solubility of the analyte

n the eluent. Since the enantioselectivity seems independentf the mobile phase polarity, the enantioselective mechanismhould be governed by at least one other type of interaction (otherhan hydrogen bondings like �–� and dipole–dipole interactionsnd steric fit in the cavities) [15]. A final amount of alcohol of0% (v/v) was finally selected to reduce analysing times withoutompromising enantioselectivity.

However, this usual optimizing of the method (optimizationf the CSP and nature and concentration of the alcohol in thepolar mobile phase) gives insufficient results if we considerhe enantioseparations obtained. Indeed, even if the selected OJSP and the n-hexane/ethanol (50/50, v/v) mobile phase offerptimal analysis time and enantioselectivity, they do not give aatisfactory enantioresolution due to high peak tailing (Fig. 2a).

The second step of the optimization was then to enhance theesolution by improving the peak shape. The mobile phase to bedjust was chosen in accordance with our previous results andonsisted of a n-hexane/ethanol (50/50, v/v) mixture.

Since amines are often added to the mobile phases toeduce peak tailing by masking the residual silanol groups of

he polysaccharide CSPs [17], diethylamine, triethylamine orimethylamine were added in the n-hexane/ethanol (50/50, v/v)obile phase in various concentrations. The maximal concen-

ration was restricted to 0.3% (v/v) to minimize any destructive

aup(

able 2hromatographic parameters: retention factor (k1), enantioselectivity factor (�), enaith n-hexane/alcohol—50/50 (v/v) eluents: addition of amines or use of ternary n-h

obile phase (v/v) k1 α

-Hexane/ethanol (50/50) 0.83 1.8-Hexane/ethanol (TEA 0.1%) (50/50) 1.11 2.1-Hexane/ethanol (TEA 0.2%) (50/50) 1.11 2.1-Hexane/ethanol (TEA 0.3%) (50/50) 1.12 2.1-Hexane/ethanol/methanol (50/45/5) 0.83 2.0-Hexane/ethanol/methanol (50/35/15) 0.79 1.9-Hexane/ethanol/methanol (50/25/25) 0.75 1.8

onditions: flow rate, 0.8 mL min−1; temperature, 30 ◦C; detection, λ = 205 nm.

ig. 2. Chromatograms obtained for 9-OHRisp with Chiralcel OJ. (a) n-exane/ethanol (50/50, v/v), (b) n-hexane/ethanol (TEA 0.2%) (50/50, v/v), (c)-hexane/ethanol/methanol (50/35/15, v/v/v) (other conditions as in Table 1).

ffects on the CSP. The results obtained are reported in Table 2.mine addition involves an increase in retention and enan-

ioselectivity and a great improvement in both efficiency andesolution. However, it appears that whatever the nature andhe concentration of the added amine, all the chromatographicarameters remain almost constant. For example, for mobilehase without amine, with 0.2% of DEA and with 0.2% of TEA,k1, α, Rs, N1) are (0.83, 1.83, <1.5, 880), (1.02, 2.16, 4.92, 1998)nd (1.11, 2.15, 5.02, 1928), respectively. The chromatogramsbtained without or with 0.2% of TEA additive are displayed inig. 2(a) and (b).

Since adding methanol in n-hexane/ethanol binary mobilehases could provide an enhancement of the enantioresolution

nd peak efficiency [18,19], the enantioseparations of 9-OHRispsing ternary n-hexane/ethanol/methanol mobile phases wereerformed for various amounts of methanol in n-hexane/alcohol50/50, v/v) eluent after verifying the miscibility of the ternary

ntioseparation (Rs) and efficiencies (N1 and N2) of 9-OHRisp on Chiralcel OJexane/ethanol/methanol mobile phases

Rs N1 N2

3 1.20 880 7706 4.94 1833 15775 5.02 1928 16129 5.10 1879 15774 4.21 1902 16715 4.21 2221 19588 3.96 2265 2083

2 atogr. A 1163 (2007) 228–236

maaow(

etuarl(alt

stm

ppatefi(

3C

tThhTpreptwe

3tocpwsssa+

Fig. 3. Chromatogram of the semipreparative enantioresolution of 9-OHRisp:Pc0

3p

tmov

3

uttpawewm

qsTepsf(boit

Otc

32 C. Danel et al. / J. Chrom

ixtures obtained. A great improvement in enantioresolutionnd efficiency was observed when methanol was added (Fig. 2and c). However, although an increase in the amount of methanolffers greater efficiency and shorter retention times, this choiceas limited by a decrease in enantioselectivity and resolution

Table 2).The addition of amine or methanol in the n-hexane/ethanol

luent was shown to enhance the efficiency and enantiosepara-ion of 9OHRisp on Chiralcel OJ. However, several reasons leds to prefer the addition of methanol to that of amine. Firstly, theddition of amine may have destructive effects on polysaccha-ides CSPs; secondly, the lower viscosity of methanol providesower column pressure which is very high for n-hexane/alcohol50/50, v/v) eluents; lastly, as illustrated in Fig. 2, the highbsorbance of the amine additive implies a considerably higherimit of detection which is a major drawback for analytical quan-itative analysis.

Finally, since our strategy was focused on developing aemipreparative method which can offer high enantiosepara-ion with short run times, the 50/35/15 (v/v/v) n-hexane/ethanol/

ethanol mobile phase was chosen as a good compromise.As for temperature, the enantioseparation of 9-OHRisp was

erformed in the 15–45 ◦C range. It was found that higher tem-eratures led to a decrease in retention factors, enantioselectivitynd enantioseparation. 20 ◦C was therefore the chosen tempera-ure, since the retention times remain satisfactory and the highernantioseparation makes higher overloadings possible. In thesenal experimental conditions, the chromatographic parametersk2, α, Rs, N2) are (2.42, 2.15, 4.42, 1347).

.1.2. Semipreparative enantioseparation on analyticalhiralcel OJ CSP

To enhance the throughput of the preparative enantiosepara-ion, a study must be made of loading capacity with the racemate.he choice of the loading results from a compromise betweenigher throughput and higher enantiomeric purity because theigher the amount of analyte injected, the lower the resolution.he influence of overloading was first examined for higher sam-le concentrations in the 0.5–3 mM range: enantioseparationemains excellent in this range (these concentrations cannot benhanced due to the low solubility of 9-OHRisp in the mobilehase). Secondly, the injected volumes (at maximal concentra-ion) were also within the 20–400 �L range: these overloadingsere limited by the touching band phenomenon and the loss of

nantioseparation.The final concentration and injection volume selected were

mM and 400 �L, respectively, which corresponds to an injec-ion of 0.52 mg. The semipreparative enantioseparation of 20 mgf 9-OHRisp was obtained through successive injections. Thehromatogram of the semipreparative enantioseparation is dis-layed in Fig. 3.The collected fractions from each enantiomerere pooled and evaporated under vacuum. The yields of the

emipreparative separation were 83 and 78% for the first and

econd isolated enantiomer of 9-OHRisp, respectively. Theirpecific rotations were measured in methanol (C = 2 mg mL−1)nd were, for the first and second isolated enantiomer [α]20

D =24 and −23, respectively.

ot(w

SC, Chiralcel OJ; eluent, n-hexane/ethanol/methanol (50/35/15, v/v/v); sampleoncentration, 3 mM; injected volume, 400 �L; temperature, 20 ◦C; flow rate,.8 mL min−1.

.2. CE and HPLC methods to determine enantiomericurity

To quantify the enantiomeric purity of the isolated enan-iomers, powerful analytical methods are needed. One CE

ethod using the dual cyclodextrin mode and two HPLC meth-ds using Chiralcel OJ or �-AGP CSPs were developed andalidated prior to determining enantiomeric purity.

.2.1. CE method: dual cyclodextrin modeA CE method had previously been developed successfully

sing a central composite design for the enantioseparation ofhe 9-OHRisp in the dual CD mode. In the final optimal condi-ions, the background electrolyte was composed of an 80 mMhosphate buffer pH 2.5, hydroxypropylated-�-CD (37 mM)nd sulfated-�-CD (3.7%, w/v); the voltage and temperatureere 20 kV and 25 ◦C, respectively. In these conditions, both

nantiomers were resolved in an analysis time of about 13 minith a Rs of 3.13 [14] which is satisfactory for envisaging deter-inations of enantiomeric purity.The repeatability of the method in terms of qualitative and

uantitative analysis was investigated by injecting the standardolutions of the racemic 9-OHRisp (100 mg L−1) eight times.o take into account the difference in residence time of eachnantiomer in the detector, the corrected areas (i.e., peak areaer migration time) instead of the peak areas alone were con-idered. Relative standard deviations (RSDs) were calculatedor migration times and the corrected areas of both enantiomersTable 3). The repeatability of the corrected areas was improvedy considering each enantiomer as the internal standard of thether: the obtained RSD was then lower than 2% (0.81%) whichs acceptable according to the common acceptance criteria ofhe ICH [20].

Linearity was studied using five solutions of racemic 9-HRisp covering the ranges 80–120 mg L−1 (80–120% of the

arget concentration) and 1–5 mg L−1 (1–5% of the target con-entration), these two ranges corresponding to the concentration

f the main and minor enantiomer, respectively. Triplicate injec-ions of each solution were applied, and the calibration curvescorrected areas versus concentration) showed good linearityith determination coefficients r2 superior to 0.990 (Table 4).

C. Danel et al. / J. Chromatogr. A 1163 (2007) 228–236 233

Table 3Repeatability of the migration or retention times and the areas for the CE and HPLC methods

CE method HPLC methods

(+)-9-OHRisp (−)-9-OHRisp �-AGP CSP OJ CSP

(+)-9-OHRisp (−)-9-OHRisp (+)-9-OHRisp (−)-9-OHRisp

R 0.86 0.98 0.10 0.19R 0.98 1.21 0.51 0.33

Mete2rtcwaifemarndbO0t

TS

LISCC

C

LISCC

C

C

C

SD of migration or retention time (%) 1.27 1.30SD of the (corrected) area (%) 1.18 1.33

oreover, the slopes and intercept to zero obtained for bothnantiomers were compared with each other, and the calculatedvalues proved that they were not significantly different for eithernantiomers since the t values were inferior to 2.056 (t(0.05,6)). Lastly, the slopes and intercepts to zero obtained in bothanges concentration were also compared: whereas the intercepto zero obtained in the two ranges did not appear to be signifi-antly different, the slopes obtained were significantly differentith t values higher than 2.056. The concentrations of the main

nd minor enantiomers determined through calibration curvesn the 80–120% and 1–5% ranges, respectively, and not directlyrom area percentages had then to be taken into account whenvaluating enantiomeric purities. The LOD and LOQ were deter-ined by serial dilution so as to obtain signal to noise ratios of 3

nd 10, respectively and were 75 and 250 ng mL−1 (Table 5),espectively, at the wavelength which gives the optimal sig-al/noise ratio (234 nm). Since no enantiomeric impurity wasetected for (+)-9-OHRisp, the LOD and LOQ were validated

y spiking the (+)-9-OHRisp solution at 50 mg L−1 with (−)-9-HRisp at 75 ng mL−1 and 250 ng mL−1, which corresponds to.15% and 0.5% of the target concentration (100% correspondso 50 mg L−1 for each enantiomer) (Fig. 4).

Fig. 4. Electropherograms of (+)-9OHRisp spiked with 0.15% (a) and 0.50%(b) of (−)-9OHRisp using the validated CE method.

able 4tudy of the linearity in the ranges 80–120% and 1–5% for the CE and HPLC methods: calibration curves and statistical parameters

CE method HPLC methods

(+)-9-OHRisp (−)-9-OHRisp �-AGP CSP OJ CSP

(+)-9-OHRisp (−)-9-OHRisp (+)-9-OHRisp (−)-9-OHRisp

inearity: range 80–120%ntercept to zero −287 −462 −4683 −10423 1.36 × 105 1.19 × 105

lope 100 101 1.34 × 109 1.31 × 109 2.17 × 107 2.20 × 107

oefficient of determination (r2) 0.992 0.992 0.996 0.997 0.998 0.998omparison of the intercepts (tcalculated); t(0.05; 26) = 2.056

0.496 0.625 0.068

omparison of the slopes (tcalculated); t(0.05; 26) = 2.056

0.257 0.626 0.448

inearity: range 1–5%ntercept to zero 9 1 278 −82 −1.09 × 104 −1.26 × 104

lope 92 91 1.19 × 109 1.24 × 109 2.14 × 107 2.13 × 107

oefficient of determination (r2) 0.996 0.996 0.991 0.992 0.999 0.999omparison of the intercepts (tcalculated); t(0.05; 26) = 2.056

1.080 1.338 0.331

omparison of the slopes (tcalculated); t(0.05; 26) = 2.056

0.240 1.373 0.349

omparison of the intercepts in thetwo ranges (t calculated)a

1.181a 1.839b 0.783a 1.486b 0.833a 0.331b

omparison of the slopes in the tworanges (t calculated)a

2.992a 3.017b 1.934a 2.753b 1.022a 0.349b

a (+)-9-OHRisp in the range 80–120% and (−)-9-OHRisp in the range 1–5%.b (−)-9-OHRisp in the range 80–120% and (+)-9-OHRisp in the range 1–5%.

234 C. Danel et al. / J. Chromatogr

Tabl

e5

Lim

itsof

dete

ctio

n(L

OD

)an

dqu

antifi

catio

n(L

OQ

)fo

rth

eC

Ean

dH

PLC

met

hods

CE

met

hod

(λ=

234

nm)

HPL

Cm

etho

ds

(+)-

9-O

HR

ispa

(−)-

9-O

HR

isp

�-A

GP

CSP

(λ=

238

nm)

OJ

CSP

(λ=

280

nm)

(+)-

9-O

HR

isp

(−)-

9-O

HR

ispa

(+)-

9-O

HR

ispa

(−)-

9-O

HR

isp

Lim

itof

dete

ctio

n(L

OD

)75

ngm

L−1

(50

ngm

L−1

)75

ngm

L−1

(50

ngm

L−1

)36

ngm

L−1

(8ng

mL

−1)

30ng

mL

−1(6

ngm

L−1

)36

ngm

L−1

(8ng

mL

−1)

55ng

mL

−1(1

1ng

mL

−1)

Lim

itof

quan

tifica

tion

(LO

Q)

250

ngm

L−1

(167

ngm

L−1

)25

0ng

mL

−1(1

67ng

mL

−1)

119

ngm

L−1

(26

ngm

L−1

)10

0ng

mL

−1(2

0ng

mL

−1)

122

ngm

L−1

(26

ngm

L−1

)18

4ng

mL

−1(3

8ng

mL

−1)

Inpa

rent

hesi

s,th

eL

OD

and

LO

Qob

tain

edby

inje

ctin

ggr

eate

rvo

lum

esa

mpl

e:(2

psi,

5s)

inst

ead

of(1

psi,

5s)

inC

Ean

d10

0�

Lin

stea

dof

20�

Lin

HPL

C.

aFi

rstd

etec

ted

enan

tiom

er.

ltertr

3O

(ert5a

tr9obp9mp(wrtt1ho

TCef

M

p

p

p

nt

. A 1163 (2007) 228–236

Enantiomeric purities were determined by injecting the iso-ated enantiomer using this validated method and were superioro 99.9% and 98.9% for the (+)-9-OHRisp (first detectednantiomer) and (−)-9-OHRisp (second detected enantiomer),espectively. When injecting (2 psi, 5 s) instead of (1 psi, 5 s),he LOD and LOQ can be reduced to 50 and 167 ng mL−1,espectively.

.2.2. HPLC methods: α-acid glycoprotein and ChiralcelJ CSPsEnantioseparation of 9-OHRisp with �-acid glycoprotein

�-AGP) CSP has already been investigated [9,11], and wenhanced the described enantioseparation (only the baselineesolution had been obtained) by studying the influence ofhe aqueous mobile phase composition with various pH of the0 mM phosphate buffer and various amounts of organic solventt 25 ◦C (Table 6).

Firstly, the influence of the pH mobile phase on the enan-ioseparation was studied in the range 4.0–6.2. The pKa of theisperidone are 3.1 and 8.1 [21]: the monoprotonation of the-OHRisp may be supposed to be in the 4.0–6.2 pH range. More-ver, the �-AGP CSP isoelectric point is 2.7 in the phosphateuffer [22]: this CSP becomes less positively charged for higherH. Thus, its ability to establish ion-pairing with the cationic-OHRisp is enhanced for higher pH. These considerationsay explain the increase in retention. The chromatographic

arameters (k2, α, Rs) obtained with the buffer/methanol90/10, v/v) mobile phase for a pH 4.0, pH 4.9 and pH 6.2ere (0.25, 1, n.r.), (1.75, 1.30, 1.52) and (6.78, 1.41, 3.17),

espectively. The pH 6.2 was selected to favor higher enan-ioresolution. Since the nature of the organic modifier is known

o greatly influence the enantioseparation, the use of methanol,-propanol, 2-propanol and acetonitrile was investigated. Theighest retention time, enantioselectivity and resolution werebtained when acetonitrile, the weakest hydrogen bonding

able 6hromatographic parameters: retention factor (k1), enantioselectivity factor (α),nantioseparation (Rs) and efficiencies (N1 and N2) of 9-OHRisp on �-AGP CSPor various mobile phases

obile phase (v/v) k1 α Rs N1 N2

H 4.0Buffer/methanol (99/1) 0.66 1.51 1.16 673 503Buffer/methanol (95/5) 0.43 1.41 0.96 1384 893Buffer/methanol (90/10) 0.25 1.00 n.r. 412 –

H 4.9Buffer/methanol (95/5) 1.78 1.49 2.40 1449 963Buffer/methanol (90/10) 1.35 1.30 1.52 1404 1455

H 6.2Buffer/methanol (90/10) 4.81 1.41 3.17 1330 1084Buffer/methanol (87/13) 4.63 1.41 3.06 1437 1198Buffer/methanol (85/15) 3.49 1.28 1.80 1442 1231Buffer/1-propanol (90/10) 1.68 1.00 n.r. 1738 –Buffer/2-propanol (90/10) 2.18 1.05 <0.5 n.d. n.d.Buffer/acetonitrile (90/10) 5.65 1.59 3.50 1341 985

.r.: no resolution; n.d.: not determined. Conditions: flow rate, 0.8 mL min−1;emperature, 30 ◦C; detection, λ = 205 nm.

atogr. A 1163 (2007) 228–236 235

so113wswlr

fiet(aia(

ctt1mα

2e

aeru

ttsbftocstcoodtdatotwcT

Fig. 5. Chromatograms of (+)-9OHRisp spiked with 0.05% (a) and 0.15% (b)o(

mtatarvectat(

ltfst

C. Danel et al. / J. Chrom

olvent, was used: the chromatographic parameters (k2, α, Rs)btained for buffer (pH 6.2)/organic modifier (90/10, v/v) with-propanol, 2-propanol, methanol and acetonitrile were (1.68,.00, n.r.), (2.29, 1.05, <0.5), (6.78, 1.41, 3.17) and (8.98, 1.59,.50). Since the resolution obtained with methanol is sufficientith shorter analysis times than acetonitrile, it was finally

elected. Lastly, the amount of methanol in the mobile phaseas studied: it was shown that higher amounts of methanol

ead to severe decreases in retention, enantioselectivity andesolution.

A buffer pH 6.2/methanol (87/13, v/v) mobile phase wasnally chosen since it allows the resolution of the 9-OHRispnantiomers (Rs = 3.06) with an analysis time of almost 17 min:he chromatographic parameters (k2, α, Rs, N2) obtained were6.53, 1.41, 3.06, 1198). Moreover, the specificity of the methods regards Risp was established by injecting a sample contain-ng both Risp and 9-OHRisp: a high resolution between Rispnd (−)-9OHRisp (first detected enantiomer) was determinedRs = 2.95).

The earlier developed semipreparative method on the Chiral-el OJ CSP required minor adjustments to be more effective inhe analytical scale: an higher temperature of 30 ◦C was foundo provide a more rapid analytical method (analysis time of3 min) while retaining sufficient enantioseparation: the chro-atographic parameters of the analytical enantioseparation (k2,, Rs, N2) obtained with this method were (1.65, 1.94, 3.87,087). Risp is co-eluted with the (+)-9-OHRisp (first detectednantiomer).

Prior to quantifiying the enantiomeric purities, both HPLCnalytical methods were validated in terms of repeatability, lin-arity, LOD and LOQ. The target concentrations (100%) ofacemic 9-OHRisp were 0.50 and 0.30 mM for the methodssing the OJ and �-AGP CSPs, respectively.

Standard solutions of racemic 9-OHRisp were injected eightimes to assess the intraday repeatability of both HPLC methods:he RSD, summarized in Table 3, were considered acceptableince they were lower than 2%. Linear ranges were validatedy injecting, in triplicate, five solutions covering each rangerom 80 to 120% and from 1 to 5% of the target concen-ration. Parameters of the calibration curves and the resultsf the various statistical tests are reported in Table 4. Theorrelation of the peak area versus analyte concentrationshowed excellent linearity over the ranges studied. Whereashe HPLC method using Chiralcel OJ CSP offers direct cal-ulation of enantiomeric purity from respective area percentagef both enantiomers (because the slopes and intercepts to zerobtained in the two concentration ranges are not significantlyifferent), the HPLC method using the �-AGP CSP, requireshat the concentrations of the main and minor enantiomersetermined through the calibration curves in the 80–120%nd 1–5% ranges, should be taken into account respectivelyo assess enantiomeric purity. With �-AGP CSP, the elutionrder determined by injecting the previously isolated enan-

iomers concorded with other publications [9]: (−)-9-OHRispas detected before the (+)-9-OHRisp (reverse elution order

ompared to that observed when using Chiralcel OJ CSP).he LOD and LOQ were assessed at 280 or 238 nm for the

tf

1

f (−)-9OHRisp using the validated HPLC method with the Chiralcel OJ CSP280 nm).

ethods using Chiralcel OJ or �-AGP CSPs, respectively. Forhe HPLC method using Chiralcel OJ CSP, the LODs are 36nd 55 ng mL−1 for (+)-9-OHRisp and (−)-9-OHRisp, respec-ively, and for the HPLC method using the �-AGP CSP, theyre 30 and 36 ng mL−1 for (−)-9-OHRisp and (+)-9-OHRisp,espectively (Table 5). These LOD and LOQ values werealidated by spiking the 100% (+)-9-OHRisp solution (purenantiomer) with (−)-9-OHRisp at 55 and 184 ng mL−1, whichorresponds to 0.05% and 0.16% of the target concentration forhe HPLC method using Chiralcel OJ CSP (Fig. 5) and at 30nd 100 ng mL−1 which corresponds to 0.05% and 0.16% of thearget concentration for the HPLC method using �-AGP CSPFig. 6).

Enantiomeric purities were determined by injecting the iso-ated enantiomers using both validated methods and were foundo be superior to 99.9% for (+)-9-OHRisp and equal to 98.9%or (−)-9-OHRisp. The enantiomeric impurity quantified in theecond isolated enantiomer (−)-9-OHRisp may be attributedo the tailing of the first peak obtained for such loading or to

he manual work which could induced errors in the collectedractions.

The LOD and LOQ may be greatly reduced by injections00 �L instead of 20 �L (values are displayed in Table 5).

236 C. Danel et al. / J. Chromatogr

F(

4

Op2s9cwtper

cbi

tbopCas

A

o9

R

[

[

[

[[

[

[[

[[[

ig. 6. Chromatograms of (+)-9OHRisp spiked with 0.05% (a) and 0.15% (b) of−)-9OHRisp using the validated HPLC method with the �-AGP CSP (238 nm).

. Conclusion

In conclusion, the semipreparative enantioseparation of 9-HRisp was successfully completed by HPLC using theolysaccharide Chiralcel OJ CSP and a ternary mobile phase:0 mg of the racemic 9-OHRisp was enantioresolved by succes-ive injections. Three analytical methods of enantioseparation of-OHRisp were developed and validated to obtain the quantifi-ation of the enantiomeric purity: the three analytical methods,hich display similar performances in terms of enantiosepara-

ion (Rs between 3.1 and 3.9) and analysis times (almost 14 min),rovide identical enantiomeric purities: higher than 99.9% andqual to 98.9% for (+)-9OHRisp and (−)-9OHRisp enantiomers,espectively.

Although their performances seem similar, these methods areomplementary. Firstly, the reversal elution order observed inoth HPLC methods may be useful for detecting and quantify-ng the minor enantiomer: since the major peak tends to tail,

[

[

. A 1163 (2007) 228–236

he quantification of the second eluted minor enantiomer maye erroneous and a reversal elution order may be required. Sec-ndly, the CE and the HPLC methods using �-AGP CSP areerformed in aqueous media while the HPLC method usinghiralcel OJ CSP is in organic media. Finally, the CE methodnd the HPLC method using the �-AGP CSP have proved theirpecificity towards risperidone.

cknowledgments

The authors are grateful to Roquette Laboratories for the giftf HP-�-CD and to Janssen-Cilag for the gift of risperidone and-hydroxyrisperidone.

eferences

[1] A. Claus, J. Bollen, H. De Cuyper, M. Eneman, M. Malfroid, J. Peuskens,S. Heylen, Acta Psychiatr. Scand. 85 (1992) 295.

[2] S.N. Ghaemi, G.S. Sachs, C.F. Baldassano, C. Truman, Can. J. Psychiat.42 (1997) 196.

[3] N.J. Rubin, J.M. Arceneaux, Acta Psychiatr. Scand. 104 (2001) 402.[4] L. Scahill, J.F. Leckman, R.T. Chultz, L. Katsovich, B.S. Peterson, Neu-

rology 60 (2003) 1130.[5] A. Schotte, P.F.M. Janssen, W. Gommeren, W.H.M.L. Luyten, P. Van-

Gompel, A.S. Lesage, K. DeLoore, J.E. Leysen, Psychopharmacology 124(1996) 57.

[6] J.A. Bork, T. Rogers, P.J. Wedlund, J. De Leon, J. Clin. Psychiat. 60 (1999)469.

[7] G. Mannens, M.L. Huang, W. Meuldermans, J. Hendrickx, R. Woesten-borghs, J. Heykants, Drug Metab. Dispos. 21 (1993) 1134.

[8] M. Vermeir, S. Boom, I. Naessens, K. Talluri, M. Eerdekens, Eur. Neu-ropsychopharm. 15 (2005) S648.

[9] N. Yasui-Furukori, M. Hidestrand, E. Spina, G. Facciola, M.G. Scordo, G.Tybring, Drug Metab. Dispos. 29 (2001) 1263.

10] C.G.M., Janssen, A.G. Knaeps, L.E.J., Ludo, J. Vandenberk, US Pat.5,158,952 (1992).

11] N. Yasui-Furukori, K. Mihara, T. Kondo, T. Kubota, T. Iga, Y. Takarada,R. De Vries, S. Kaneko, T. Tateishi, J. Clin. Pharmacol. 43 (2003) 122.

12] User’s guide of the Chiral-AGP column, Application Note 17, ChromTech,Congleton, UK, 1998; http://www.chromtech.se.

13] E.R. Francotte, J. Chomatogr. A 906 (2001) 379.14] C. Danel, P. Chaminade, P. Odou, C. Barthelemy, D. Azarzar, J.P. Bonte,

C. Vaccher, Electrophoresis 28 (2007) in press.15] T. O’Brien, L. Crocker, R. Thompson, P.H. Toma, D.A. Conlon, B. Feibush,

C. Moeder, G. Bicker, N. Grinberg, Anal. Chem. 69 (1997) 1999.16] A. Kunath, F. Theil, J. Wagner, J. Chromatogr. A 684 (1994) 162.17] S.C. Sharma, M.B. Evans, S.J. Evans, J. Pharm. Biomed. Anal. 13 (1995)

129.18] A. Kunath, F. Theil, K. Jahnisch, J. Chromatrogr. A 728 (1996) 249.19] X. Luo, Z. Zhai, X. Wu, Y. Shi, L. Chen, Y. Li, J. Sep. Sci. 29 (2006) 164.20] Proceedings of the International Conference on Harmonization (ICH) of

Technical Requirements for Registration of Pharmaceuticals or HumanUse, IFPMA, Geneva, 1996.

21] M.I. El-Barghouthi, N.A. Masoud, J.K. Al-Kafawein, M.B. Zughul, A.A.Badwan, J. Incl. Phenom. Macro. 53 (2005) 15.

22] J. Hermansson, Trends Anal. Chem. 8 (1989) 251.