a rapid cze method for the analysis of benzodiazepines in spiked beverages

Rebecca WebbPhilip DobleMichael Dawson

Centre for Forensic Science,Department of Chemistry,Materials and Forensic Science,University of Technology,Sydney, NSW, Australia

Received February 14, 2007Revised May 30, 2007Accepted May 31, 2007

Research Article

A rapid CZE method for the analysis ofbenzodiazepines in spiked beverages

A rapid CZE method was developed for the simultaneous determination of nine benzodia-zepines in spiked beverages (nitrazepam oxazepam, alprazolam, flunitrazepam, temaze-pam, diazepam, 7-aminoflunitrazepam, 7-aminonitrazepam and 7-aminoclonazepam).The method employed a double-coated capillary coated with poly(diallyldimethylammo-nium chloride) and then dextran sulphate. The BGE conditions were 100 mM ammoniumphosphate buffer, pH 2.5, which gave baseline resolution between consecutive peaks and arun time of less than 6.5 min. This method offers improvements in both resolution and runtime, compared to those attained under analogous conditions with an uncoated capillary.The validated method was successfully applied to beverages that had been spiked withbenzodiazepines at concentrations simulating prescription tablets. No sample pretreatmentwas required to quantify five benzodiazepines in Coca-Cola, orange juice, beer, bourbon andBacardi. The exception was white wine, where the complex sample matrix did not enablethe accurate quantification of nitrazepam.

Keywords:

Benzodiazepines / CZE / Doubly coated capillary / Drink spiking / Forensic scienceDOI 10.1002/elps.200700110

Electrophoresis 2007, 28, 3553–3565 3553

1 Introduction

Benzodiazepines are widely used in the clinical treatment ofanxiety, insomnia and panic disorder [1]. In Australia, recentpublic concern surrounding the issue of drink spiking hashighlighted the abuse potential of benzodiazepines, makingtheir analysis of interest from a forensic and toxicologicalstandpoint. The term ‘drink spiking’ refers to drugs oralcohol being added to a drink without the consent of theperson consuming it [2]. The spiking substance, commonlybenzodiazepines such as flunitrazepam (FLU), alprazolam(ALP), diazepam (DIA), temazepam (TEM) and oxazepam(OXA) [3, 4], can be added to both alcoholic and non-alco-holic drinks. The effects can range from vomiting to loss ofconsciousness, and can also include poor coordination andbalance, slurred speech, muscle spasms, respiratory diffi-

culties and loss of control. Often the victim will have lastinganterograde amnesia for events that occur under the influ-ence of the drug [4].

For an incident to be defined as drink spiking, it need notinvolve further criminal victimisation, although such of-fences, including drug facilitated sexual assault and robbery,can occur [5]. A recent report by the Australian Institute ofCriminology (AIC) [5] estimated that there were between3000 and 4000 suspected incidents of drink spiking acrossAustralia in the 2002/2003 financial year, with approximatelyone-third of incidents estimated to have been associated withsexual assault. Prior to the AICs report, there was no separateoffence category in any Australian jurisdiction for the act ofspiking someone’s drink. However, the AICs report hasprompted the New South Wales (NSW) Government to cre-ate a new summary offence of drink spiking, which willattract a maximum penalty of 2 years in jail.

While HPLC remains the most common technique forthe analysis of benzodiazepines [6–10], several CE methodshave recently been developed. Still considered an emergingtechnique in forensic toxicology, this alternative to HPLCoffers advantages such as low solvent and sample require-ments, high efficiency and rapid analysis times. Migrationtimes of solutes in CE are dependent upon the charge/sizeratio of the solute and the level of EOF in the capillary [11],which is largely dependent on pH. The EOF decreasesrapidly below pH 6 and, at pH 2–3, it is almost completelyeliminated, resulting in increased analysis times for cationic

Correspondence: Dr. Philip Doble, Centre for Forensic Science,Faculty of Science, University of Technology, Sydney (UTS), POBox 123, Broadway NSW, AustraliaE-mail: [email protected]: 161-2-9514-1460

Abbreviations: ALP, alprazolam; DIA, diazepam; DS, dextran sul-phate; FLU, flunitrazepam; 7-NH2-CLO, 7-aminoclonazepam; 7-

NH2-FLU, 7-aminoflunitrazepam; 7-NH2-NIT, 7-aminonitrazepam;NIT, nitrazepam; OXA, oxazepam; PDDAC, poly(diallyldimethyl-ammonium chloride); TEM, temazepam

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

3554 R. Webb et al. Electrophoresis 2007, 28, 3553–3565

compounds [12]. Considering this, and given the low pKa’s ofbenzodiazepines, which makes them difficult to ionise andtherefore difficult to analyse by CZE, the majority of benzo-diazepine analyses are performed with micellar EKC(MEKC) [13–20]. CEC on octadecyl stationary phases [21] andon novel stationary phases based on modified silyl silica gel[22] have also been developed. Only a handful of methodshave employed CZE for the analysis of benzodiazepines [18,19, 22–25]. Despite this, CZE offers potential advantages overMEKC, including better efficiency and precision, improvedresolution, BGE simplicity and the possibility of couplingwith MS.

The coating of capillaries in CE is often performed to sta-bilise the EOF and eliminate associated peak tailing, poorefficiency [26] and peak area variability [11], by reducingadsorption of buffer or sample components to the capillarysurface. It is especially relevant for the analysis of basic ana-lytes, which are positively charged and therefore attracted tothe negatively charged capillary wall. Capillary coating mayalso be used to assist in changing the direction and/or speed ofthe EOF. Coatings can be classified as either static, wherebythe coating material is permanently attached to the capillarywall by way of covalent bonds, or dynamic, whereby the capil-lary is rinsed with a solution containing the coating reagent,which becomes adsorbed onto the capillary wall throughelectrostatic interactions with the capillary silanol groups.Static coatings render the capillary wall uncharged, whichcauses elimination of the EOF and can result in lengthy anal-ysis times. In addition, the coating procedures are oftenlengthy and require multistep chemical reactions, which canproduce a large variance between capillaries. In contrast topermanent coatings, dynamic coatings are relatively simple toprepare. The main drawback is that dynamically coated capil-laries require occasional regeneration, and addition of thecoating agent into the BGE may be necessary [27]. As analternative to the dynamic capillaries coated with a monolayerof a positively charged polymer, which causes reversal of theEOF and the slow migration of cations, a second anionicpolymer layer can be electrostatically adsorbed onto the firstcationic polymer layer to create a double-coated capillary. Thisresults in a pH-independent EOF and allows for very rapidcationic separations. Numerous double-coated capillarieshave been prepared with the aid of different polymers,including the poly(diallyldimethylammonium chloride)/dex-tran sulphate (PDDAC/DS)-coated capillary utilised byZakaria et al. [28] for the analysis of alkaloids. This PDDAC/DS coating method was also successfully applied by Kelly et al.[29] for the chiral separation of methadone and its two majormetabolites. A number of other double-coated capillaries havebeen prepared from polymers such as PDDAC/poly(sodiumundecylenic sulphate) [30], Polybrene/poly(vinylsulphonate)[31], Polybrene/DS [32] and the commercially available CEo-fix® or CElixir® buffer systems [33–37].

The lengthy analysis times which arise from the sub-sequent reduction or elimination of the EOF at low pH haslimited the number of studies employing CZE for the anal-

ysis of benzodiazepines. A coated CZE method could allowthe EOF to be introduced and permit a fast cationic separa-tion, thereby overcoming the problem of long run timesassociated with analysis at low pH. To date, the only methodpublished for the analysis of benzodiazepines using a dou-ble-coated capillary was has been by Vanhoenacker et al.[33], who employed the commercially available CEofix coat-ing system for the analysis of six benzodiazepines by CE-DAD, CE-MS and CE-MS-MS in 100 mM formic acid atpH 2.4.

While there have been a number of studies focusing onthe analysis of biological samples after the administration ofspiked beverages, many of these are screening tests thatemploy colour tests on drink coasters [38] and other similarconsumer products for the detection of benzodiazepines andother frequently encountered drugs, such as g-hydroxy-butyric acid (GHB) and its precursor g-butyrolactone (GBL).Only three studies have focused on the analysis of spikedbeverages by confirmatory methods. Bishop et al. [13] appliedan MEKC method with an analysis time of approximately14 min for the analysis of benzodiazepines and GHB in anumber of beverages. The main disadvantage of this methodwas that it required an ethyl acetate liquid–liquid extraction(LLE). A similar method was employed by Chen and Hu [39]who extracted benzodiazepines from several beverages usingchloroform or dichloromethane prior to analysis by directelectrospray probe/MS. Rao et al. [40] employed an HPLCmethod for the determination of ALP, DIA and chloralhydrate in the alcoholic beverage toddy. The analysis timewas 15 min and the only sample pretreatment required wasfiltration through 0.45 mm Nylon membranes.

In this investigation, we report a fast CZE-DAD separa-tion for the analysis of six parent benzodiazepines plusthree metabolites, through the use of a dynamic doubly-coated capillary coated with a polycation of PDDAC and apolyanion of DS. The three metabolites were added todemonstrate the potential of the method to be applied tobiological matrices such as blood or urine. The method wasapplied to the analysis of beverages spiked with DIA, nitra-zepam (NIT), ALP, FLU, TEM and OXA, with no samplepretreatment required. A variety of drinks frequently con-sumed at bars and parties, including Coca-Cola™, orangejuice, beer, white wine, bourbon and Bacardi™, were selectedfor the analysis.

2 Materials and methods

2.1 Instrumentation

Separations were performed on an Agilent ChemStationCapillary Electrophoresis System (Agilent Technologies,USA), equipped with DAD. Uncoated 50 mm ID fused-silica capillaries of 69 cm total length (60 cm effectivelength) were purchased from Polymicro Technologies(Phoenix, USA). Detection windows were made by burning


Electrophoresis 2007, 28, 3553–3565 CE and CEC 3555

Figure 1. Influence of (i) phos-phate buffer concentration and(ii) pH on average product reso-lution for coated and uncoatedcapillaries.

the outer polyimide coating to expose the bare capillary.Aqueous standards of each analyte were prepared by dilutionof stock solutions with Milli-Q water. Duplicate injections foreach standard were performed hydrodynamically at50.0 mbar for 2 s. Separations were carried out at 257C withan applied voltage of 128 kV, and detection at 200 nm. At thebeginning of each day, the capillary was flushed with 0.1 NNaOH for 5 min, Milli-Q water for 5 min and running bufferfor 10 min. Prior to each injection, the capillary was flushedwith the running buffer for 3 min. The position of the EOFwas verified for each condition by injection of an acetone so-lution (10% acetone in running buffer).

Coated capillaries were prepared by a method descri-bed previously [28, 29]. To summarise, the capillary wasflushed with 1 N NaOH for 30 min, followed by Milli-Qwater for 15 min, and was then allowed to stand for30 min. A solution of 1% v/v PDDAC was flushed through

the capillary for 15 min, and then left to stand for 15 min.A two-minute flush with Milli-Q water preceded a 15 minflush with 1.5% w/v DS solution. After allowing the cap-illary to stand for 30 min, Milli-Q water was flushedthrough for 5 min.

2.2 Reagents

All reagents were of analytical grade unless stated otherwise.NIT, OXA, ALP, FLU, TEM and DIA were obtained fromSigma–Aldrich (Sydney, Australia). 7-Aminoflunitrazepam(7-NH2-FLU), 7-aminonitrazepam (7-NH2-NIT) and 7-ami-noclonazepam (7-NH2-CLO) were from Novachem Pty Ltd(Melbourne, Australia). All buffers were prepared in Milli-Qgrade (18 MO/cm) water by dilution of stock solutions ofphosphoric acid (H3PO4) (APS Chemicals, Sydney, Aus-tralia). Ammonia solution (NH4OH, APS Chemicals) was



Figure 2. Influence of (i) phos-phate buffer concentration and(ii) pH on run time for coatedand uncoated capillaries.

used for pH adjustment. 1 N NaOH (HPCE grade) wasobtained from Agilent Technologies (Sydney, Australia).PDDAC (20 wt% in water) and DS were obtained fromSigma–Aldrich. Prior to use, all buffers were degassed bysonication and filtered using 0.45 mm Nylon syringe filters(Bonnett Equipment, Sydney, Australia).

2.3 Calculations

Migration times were determined by the time correspondingto the peak apex of each analyte. The resolution betweenconsecutive analyte peaks was calculated as the difference inmigration time divided by half the sum of the peak widths.The product resolution (PRs) was calculated by the multi-plication of each of the individual resolution values.

2.4 Method development

In CE separations, the pH of the BGE must be selected care-fully so as to ensure ionisation of the species. If the pH is toohigh, the basic analytes may be insufficiently ionised toachieve complete separation. If the pH of the BGE is too low,the EOF is minimised, resulting in long analysis times andlow S/Ns. The ionic strength of the BGE can also influencethe analysis time. High ionic strength BGEs decrease thezeta potential, and hence the EOF is also decreased. With thisin mind, and considering the pKas of the benzodiazepines, aBGE of ammonium phosphate was investigated in the rangeof pH 1.9–2.5 (constant concentration 100 mM) and 25–100 mM (constant pH 2.5). Experiments were performed onboth a PDDAC/DS-coated capillary and a bare uncoated cap-illary to assess the effectiveness of the coated capillary in



Figure 3. Example electro-pherogram of a standard mix-ture of nine benzodiazepines ona coated capillary (7-NH2-NIT60 mg/mL, 7-NH2-FLU 60 mg/mL,7-NH2-CLO 60 mg/mL, DIA120 mg/mL, NIT 120 mg/mL, ALP225 mg/mL, FLU 120 mg/mL TEM120 mg/mL, OXA 120 mg/mL)–50 mm ID PDDAC/DS-coatedcapillary, total length 69 cm,effective length 60 cm, 100 mMphosphate buffer, pH 2.5,applied voltage 1 28 kV, tem-perature 257C, injection 2 s at50 mbar, 200 nm detectionwavelength. (1) 7-NH2-NIT, (2) 7-NH2-FLU, (3) 7-NH2-CLO, (4) DIA,(5) NIT, (6) ALP, (7) FLU, (8) OXA,(9) TEM, (10) EOF.

Figure 4. Example electro-pherograms of (i) spiked Bacardiand (ii) blank Bacardi (DIA 50 mg/mL, NIT 25 mg/mL, ALP 10 mg/mL, FLU 25 mg/mL TEM 150 mg/mL, OXA 150 mg/mL)–50 mm IDPDDAC/DS-coated capillary,total length 69 cm, effectivelength 60 cm, 100 mM phos-phate buffer, pH 2.5, appliedvoltage 1 28 kV, temperature257C, injection 2 s at 50 mbar,200 nm detection wavelength.(1) DIA, (2) NIT, (3) FLU, (4) OXA,(5) TEM.



Figure 5. Example electro-pherograms of (i) spiked bour-bon and (ii) blank bourbon –refer to Fig. 4 for conditions.

separating all nine analytes. Separations were assessedaccording to product resolution.

2.5 Method validation

The method was validated for aqueous standards and cali-bration curves were obtained by analysing working standardsolutions, diluted to obtain final concentrations of 17.0, 22.7,45.5, 68.2, 90.9 and 182 mg/mL for DIA, NIT, FLU, TEM,OXA, 7-NH2-NIT, 7-NH2-CLO and 7-NH2-FLU, and 51.1,68.2, 136.4, 204.5 and 273 mg/mL for ALP. Calibrationstandards were analysed each day, and standard curves wereconstructed using linear regression.

Accuracy and precision were calculated at high and lowconcentration for each drug, with five replicates performed at

each concentration. Accuracy was expressed as the calculatedconcentration as a percentage of nominal concentration.Precision (% CV) was determined to be the SD in peak areadivided by the average of the three replicates, expressed as apercentage. The LOD was defined as an S/N of 3:1 and theLOQ as an S/N of 10:1.

Recovery was calculated for each analyte as the averagepeak area of the replicate spiked samples as a percentage ofthe average peak area in aqueous standards.

2.6 Sample preparation

Six beverages, including Bacardi rum, Victorian Bitter™ beer,Jacob’s Creek™ Semillon Sauvignon Blanc white wine, Coca-cola, Jim Beam™ bourbon and Just Juice™ orange juice, were



Figure 6. Example electro-pherograms of (i) spiked Coca-cola and (ii) blank Coca-cola –refer to Fig. 4 for conditions.

selected for analysis. Beverages were spiked with each drugat a concentration designed to simulate a prescription tablet.Information regarding tablet strengths was obtained fromthe manufacturers’ websites, and the following tabletstrengths were utilised: DIA 10 mg, NIT 5 mg, ALP 2 mg,FLU 5 mg, TEM 30 mg and OXA 30 mg. A standard drinkwas assumed to be 200 mL. Thus, the following final con-centrations were obtained: DIA 50 mg/mL, NIT 25 mg/mL,ALP 10 mg/mL, FLU 25 mg/mL, TEM 150 mg/mL and OXA150 mg/mL. Metabolites were excluded from the analysis.Beverages were analysed in replicate without dilution bydirect injection. No sample pretreatment was required priorto analysis, however the particulate matter in orange juicewas allowed to settle before it was decanted into the samplevial.

3 Results and discussion

3.1 Method development

Given the acidic range in which the experiments were per-formed, the positively charged basic analytes migrated aheadof the EOF under the application of a positive potential. Aswould be expected at low pH, 7-NH2-NIT, which is proto-natable on the –NH2 group (pKa 4.6) and the C=N group(pKa 2.5), migrates through the capillary first. It is closely fol-lowed by 7-NH2-FLU and 7-NH2-CLO, which are also proto-natable at these two positions. The remainder of the analyteswere separated in line with their pKa values of the mono-protonated species; that is DIA (pKa 3.3), NIT (pKa 3.2), ALP(pKa 2.4), FLU (pKa 1.8), OXA (pKa 1.7) then TEM (pKa 1.6).



Figure 7. Example electro-pherograms of (i) spiked wineand (ii) blank wine – refer toFig. 4 for conditions.

Compared with an uncoated bare fused-silica capillary,separations performed on the coated capillary were superiorin terms of resolution. While comparable resolutions wereobtained in 25 and 50 mM phosphate buffer, the coated cap-illary gave a significantly better separation at 75 and 100 mM(Fig. 1i). And although slightly better separations wereobtained on an uncoated capillary below pH 2.4, there was adramatic increase in resolution at pH 2.5 for the coated cap-illary (Fig. 1ii). Both these observations are attributable to thefact that the coated capillary allowed baseline separation to beobtained between 7-NH2-FLU and 7-NH2-CLO. This was notachievable on an uncoated capillary under any of the condi-tions investigated.

In addition to improved resolution, the coated capillarygave a faster run time than the bare fused-silica capillaryunder all BGE concentrations and pHs investigated. The runtime on the uncoated capillary was affected by the BGE con-centration, showing a steady increase as the BGE concentra-

tion was increased (Fig. 2i). In contrast, the run time on theuncoated capillary remained relatively constant at around6 min. Varying pH had little effect on the run times of eitherthe coated or uncoated capillaries; however, the coated capil-lary improved the run time by about 1 min, or around 17%(Fig. 2ii). Whilst not a dramatic improvement, it becomesmore significant during routine analysis when there is a highsample throughput. The study by Vanhoenacker et al. [33]showed a more dramatic reduction in migration times when acoated capillary was employed for the analysis of benzodiaze-pines versus an uncoated capillary. However, different analyteswere selected for analysis including lorazepam, which had asignificantly longer migration time and showed the greatestimprovement when analysed on the coated capillary.

Considering all the conditions evaluated on both thecoated and uncoated capillaries (BGE concentration 25–100 mM, pH 1.9–2.5), the best separation (Fig. 3) was foundto occur in 100 mM phosphate buffer at pH 2.5 on the



Figure 8. Example electro-pherograms of (i) spiked beerand (ii) blank beer – refer to Fig. 4for conditions.

PDDAC/DS-coated capillary. Under these conditions, base-line separation of all nine analytes was possible in a run timeof less than 6.5 min.

This method compares favourably with the other CZEmethods for benzodiazepine analysis. The CZE methoddeveloped by McGrath et al. [18] for the analysis of chlordi-azepoxide, DIA, flurazepam and NIT had a total analysistime of 8 min for the four analytes when employing a20 mM citric acid 1 15% MeOH BGE. McClean et al. [25]described a CZE method in which 15 benzodiazepines wereanalysed in 20 mM citric acid 1 15% MeOH at pH 2.5.Under these conditions, only 12 analytes could be resolvedwithin 20 min using DAD, and many of these did not havebaseline resolution. Clearly, the current method, which isable to baseline resolve nine benzodiazepines, offers con-siderable advantages in terms of analysis time and resolutionwhen compared with previous CZE methods.

3.2 Stability of the EOF

In contrast to some previous findings [28, 41], the EOF ofthe double-coated PDDAC/DS capillary was not reproduci-ble and tended to decrease as more injections were per-formed. This finding was in agreement with Kelly et al.[29], who also found a large amount of variation in theEOF. The likely reason for this seems to be the gradualremoval of DS from the surface of the capillary wall by therunning buffer during analyses. Kelly et al. suggested thatrunning buffer containing a small amount of DS mayreduce variation in the EOF by preventing the removal ofthe DS coating during analysis. This approach was inves-tigated in the range of 0–2% DS in 100 mM phosphatebuffer at pH 2.5. EOF values were determined by an injec-tion of acetone as a neutral marker, with ten replicatesperformed at each concentration. The % CV of the migra-



Figure 9. Example electro-pherograms of (i) spiked orangejuice and (ii) blank orange juice –refer to Fig. 4 for conditions.

tion time was calculated, and a minimum of 0.029% wasobtained at 1.5% DS. At concentrations of DS above 1.5%,the % CV began to increase.

These conditions were then applied to the separation of astandard mixture of the nine benzodiazepines. 1.5% DS wasadded to the running buffer (100 mM phosphate buffer,pH 2.5) and ten replicate injections were performed. Unfor-tunately, the addition of DS caused distorted peak shapes andcomigration of most of the analytes with the EOF. This islikely due to the positively charged benzodiazepines inter-acting with the negatively charged DS to form neutral spe-cies which migrate with the EOF. As such, the best approachfor maintaining a reproducible EOF when analysing posi-tively charged species on the PDDAC/DS-coated capillaryseems to be periodic flushing with DS. The frequency atwhich this should be performed can be assessed by regularinjections of acetone to measure any deterioration in the

EOF. In this work, flushing with DS was required approxi-mately every 20 injections.

3.3 Method validation

Calibration curves were linear in the specified concentrationranges and correlation coefficients (r2) ranging from 0.9891to 0.9976, and LODs from 2.7 to 41.5 mg/mL were estab-lished (Table 1). The linear ranges were within the analysedconcentration of the analytes, rather than the true linearrange. Peak area repeatability was in the order of 0.9–11%RSD and the migration time repeatability was 0.3–0.8%RSD, with accuracies between 95 and 123% (Table 2). Thehigher LOD and LOQ for ALP is likely to be due to the poorpeak shape associated with this analyte. This may be aresult of ring opening, which is known to occur in acidicmedium [42].



Table 1. Calibration and LOD data

Analyte Correlationcoefficient (r2)

Range(mg/mL)

LOD(mg/mL)

LOQ(mg/mL)

7-NH2-NIT 0.9972 17.0–90.9 4.0 13.37-NH2-FLU 0.9976 17.0–90.9 3.5 11.77-NH2-CLO 0.9933 17.0–90.9 2.7 9.0DIA 0.9976 17.0–90.9 3.0 10.1NIT 0.9960 17.0–90.9 3.4 11.4ALP 0.9891 51.1–272.7 41.5 138.2FLU 0.9949 17.0–90.9 6.6 21.9OXA 0.9946 17.0–90.9 4.2 13.9TEM 0.9938 17.0–90.9 4.8 16.1

Table 2. Accuracy and precision data

Analyte Concen-tration(mg/mL)

Accuracy(%)

RSD peakareas (%)

RSDmigrationtimes (%)

7-NH2-NIT 90.9 99 0.9 0.517.0 97 2.6 0.5

7-NH2-FLU 90.9 101 1.0 0.517.0 110 3.2 0.4

7-NH2-CLO 90.9 100 0.8 0.617.0 111 6.1 0.4

DIA 90.9 101 0.9 0.617.0 112 5.6 0.4

NIT 90.9 100 2.8 0.617.0 108 1.3 0.3

ALP 272.7 95 3.1 0.851.1 110 4.2 0.3

FLU 90.9 101 3.0 0.817.0 123 11.0 0.4

OXA 90.9 97 3.6 0.817.0 115 6.7 0.5

TEM 90.9 103 2.0 0.817.0 119 9.6 0.5

3.4 Method application

Example electropherograms of each beverage, both spikedand blank, are illustrated in Fig. 4–9. In the case of Bacardi(Fig. 4ii), the sample matrix is very clean and there are nointerfering peaks when compared to the spiked sample(Fig. 4i). This is also the case for bourbon (Fig. 5) andCoca-cola (Fig. 6), which only exhibit one peak corre-sponding to the EOF. In these beverages, there are no dif-ficulties with peak assignment and each analyte can bequantified. However, the sample matrices for the particulartypes of wine (Fig. 7ii), beer (Fig. 8ii) and orange juice(Fig. 9ii) analysed are more complicated. In the case ofwine, there is a small peak (tm = 4.5 min) in the blanksample, which is also the migration time of NIT in the

spiked sample (Fig. 7i). As a result, NIT cannot be accu-rately quantified in this sample. While the beer and orangejuice also have complex electropherograms, there is nopeak overlap between any of the matrix components andthe analyte peaks in either of these cases, thus permittingthe quantification of each analyte.

The recovery data for each analyte is reported in Table 3.While the CZE-DAD method was capable of quantifyingDIA, NIT, FLU, TEM and OXA, it lacked the sensitivityrequired to detect ALP at a concentration designed to simu-late a standard drink spiked with a prescription tablet of thisparticular drug. Recoveries of 79.4–135.8% were obtained for

Table 3. Recovery of each drug from spiked beverages

Beverage Com-pound

Concen-trationadded(mg/mL)

Concen-trationfound(mg/mL)

Recovery

Bacardi DIA 50 46.9 93.8NIT 25 25.4 101.6ALP 10 BLODa) n/aFLU 25 34.0 135.8TEM 150 119.2 79.4OXA 150 158.6 105.8

Wine DIA 50 39.7 79.4NIT 25 20.7 82.7ALP 10 BLODa) n/aFLU 25 30.0 120.2TEM 150 130.7 87.1OXA 150 152.3 101.5

Bourbon DIA 50 28.6 57.2NIT 25 17.9 71.8ALP 10 BLODa) n/aFLU 25 13.3 53.3TEM 150 74.7 49.8OXA 150 88.7 59.1

Beer DIA 50 41.9 83.9NIT 25 19.0 75.9ALP 10 BLODa) n/aFLU 25 27.6 110.4TEM 150 131.0 87.3OXA 150 149.4 99.6

Orange juice DIA 50 35.4 70.7NIT 25 12.8 51.3ALP 10 BLODa) n/aFLU 25 21.9 87.6TEM 150 115.3 76.9OXA 150 145.2 96.8

Coca-cola DIA 50 34.0 68.1NIT 25 16.3 65.2ALP 10 BLODa) n/aFLU 25 25.4 101.7TEM 150 130.5 87.0OXA 150 143.8 95.9

a) Below LOD.



Bacardi, 79.4–120% for white wine, 49.8–71.8% for bourbon,75.9–110.4% for beer, 51.3–96.8% for orange juice and 65.2–101.7% for Coca-cola.

Compared to the other published methods for theanalysis of spiked beverages [13, 39, 40], the PDDAC/DS-coated method offers considerably shorter run times.Another advantage of this method is that it does notrequire an extraction step. Direct injection was sufficient todetect and quantify five of the six benzodiazepines ana-lysed, however the sensitivity was not adequate for thedetection of ALP at 10 mg/mL. This method is also appli-cable for the analysis of three benzodiazepine metabolites,which may render it appropriate for the analysis of biolog-ical samples.

4 Concluding remarks

A rapid CZE method for the simultaneous determinationof nine benzodiazepines was developed using a doubly-coated PDDAC/DS capillary. The selected BGE conditionswere 100 mM ammonium phosphate buffer, pH 2.5, whichgave baseline resolution between consecutive peaks and arun time of less than 6.5 min. This method offers signifi-cant improvements both in terms of resolution and runtime when compared with CZE analyses of benzodiaze-pines on bare fused-silica capillaries. Peak area repeat-ability was in the order of 0.9–11% RSD and the migrationtime repeatability was 0.3–0.8% RSD. The validated meth-od was successfully applied to beverages that had beenspiked with benzodiazepines at concentrations simulatingthe addition of the dose of benzodiazepine found in a sin-gle prescription tablets. No sample pretreatment wasrequired to quantify five of the six benzodiazepines in thebeverages investigated, with the exception of wine, whosecomplex sample matrix did not enable the accurate quan-tification of NIT.

5 References

[1] Barker, M. J., Jackson M., Greenwood, K. M., Aust. Psychol.2003, 38, 202–213.

[2] Drink spiking – For workers, Australian Drug Foundation,Drug Info Clearinghouse ,http://druginfo.adf.org.au/arti-cle.asp?id=4911., Australian Drug Foundation, 2002.

[3] Negrusz, A., Gaensslen, R. E., Anal. Bioanal. Chem. 2003, 376,1192–1197.

[4] Schwartz, R. H., Milteer, R., LeBeau, M. A., South Med. J.2000, 93, 558–561.

[5] Taylor, N., Prichard, J., Charlton, K., National Project on DrinkSpiking: Investigating the Nature and Extent of Drink Spikingin Australia, Australian Institute of Criminology, Canberra2004 November.

[6] Zilli, M. A., Nisi, G., J. Chromatogr. 1986, 378, 492–497.

[7] Puopolo, P. R., Pothier, M. E., Volpicelli, S. A., Flood, J. A.,Clin. Chem. 1991, 37, 701–706.

[8] Adams, H. A., Weber, B., Bachmann, M. B., Guerin, M.,Hempelmann, G., Anaesthesist 1992, 4, 619–624.

[9] Casas, M., Berrueta, L. A., Gallo, B., Vicente, F., Chromato-graphia 1992, 34, 79–82.

[10] Ferrara, S. D., Tedeschi, L., Frison, G., Castagna, F., J. Anal.Toxicol. 1992, 16, 217–222.

[11] Altria, K. D., J. Pharm. Biomed. Anal. 2003, 31, 447–453.

[12] Bendahl, L., Hansen, S. H., Gammelgaard, B., Electrophore-sis 2001, 22, 2565–2573.

[13] Bishop, S. C., Lerch, M., McCord, B. R., Forensic Sci. Int.2004, 141, 7–15.

[14] Suzuki, Y., Arakawa, H., Maeda, M., Biomed. Chromatogr.2004, 18, 150–154.

[15] Bechet, I., Fillet, M., Hubert, P., Crommen, J., Electrophoresis1994, 15, 1316–1321.

[16] Renou-Gonnord, M. F., David, K., J. Chromatogr. A 1996,735, 249–261.

[17] Tagliaro, F., Smith, F. P., Turrina, S., Equisetto, V., Marigo, M.,J. Chromatogr. A 1996, 735, 227–235.

[18] McGrath, G., McClean, S., O’Kane, E., Smyth, W. F., Tagliaro,F., J. Chromatogr. A 1996, 735, 237–247.

[19] Nevado, J. J. B., Penalvo, G. C., Calderon, M. J. P., J. Chro-matogr. B 2002, 773, 151–158.

[20] Steinmann, L., Thormann, W., J. Capil. Electrophor. 1995, 2,81–88.

[21] Kapnissi, C. P., Warner, I. M., J. Chromatogr. Sci. 2004, 42,238.

[22] Ohyama, K., Wada, M., Ohba, Y., Fujishita, O., Nakashima,K., Kuroda, N., Biomed. Chromatogr. 2004, 18, 396.

[23] McClean, S., O’Kane, E., Smyth, W. F., Electrophoresis 2000,21, 1381–1389.

[24] Peyrin, E., Guillaume, Y. C., J. Chromatogr. A 1999, 849, 563–573.

[25] McClean, S., O’Kane, E., Hillis, J., Smyth, W. F., J. Chroma-togr. A 1999, 838, 273–291.

[26] Shah, R. S., Wang, Q., Lee, M. L., J. Chromatogr. A 2002, 952,267–274.

[27] Horváth, J., Dolník, V., Electrophoresis 2001, 22, 644–655.

[28] Zakaria, P., Macka, M., Haddad, P. R., J. Chromatogr. A 2003,985, 493–501.

[29] Kelly, T., Doble, P., Dawson, M., Electrophoresis 2003, 24,2106–2110.

[30] Kamande, M. W., Kapnissi, C. P., Zhu, X. F., Akbay, C., War-ner, I. M., Electrophoresis 2003, 24, 945.

[31] Catai, J. R., Somsen, G. W., de Jong, G. J., Electrophoresis2004, 25, 817–824.

[32] Katayama, H., Ishihama, Y., Asakawa, N., Anal. Chem. 1998,70, 2254.

[33] Vanhoenacker, G., de l’Escaille, F., De Keukeleire, D., Sandra,P., J. Pharm. Biomed. Anal. 2004, 34, 595–606.



[34] Vanhoenacker, G., de l’Escaille, F., De Keukeleire, D., Sandra,P., J. Chromatogr. B 2004, 799, 323–330.

[35] Lurie, I. S., Hays, P. A., Parker, K., Electrophoresis 2004, 25,1580.

[36] Lurie, I. S., Panicker, S., Hays, P. A., Garcia, A. D., Geer, B. L.,J. Chromatogr. A 2003, 984, 109–120.

[37] Lurie, I. S., Hays, P. A., Garcia, A. E., Panicker, S., J. Chroma-togr. A 2004, 1034, 227–235.

[38] Meyers, J. E., Almirall, J. R., J. Anal. Toxicol. 2004, 28, 685–688.

[39] Chen., Y., Hu, A., Forensic Sci. Int. 1999, 103, 79.

[40] Rao, R. N., Parimala, P., Khalid, S., Alvi, S. N., Anal. Sci. 2004,20, 383–386.

[41] Zakaria, P., Macka, M., Haddad, P. R., Electrophoresis 2002,23, 1844–1852.

[42] Rieck, W., Platt, D., J. Chromatogr. B 1992, 578, 259–263.


a rapid cze method for the analysis of benzodiazepines in spiked beverages

Documents