A

Eupcfi1a2hp©

Kt

1

mdaaEtlia

i

0d

Analytica Chimica Acta 588 (2007) 20–25

Rapid multi-residue method for the quantitative determination andconfirmation of glucocorticosteroids in bovine milk using liquid

chromatography–electrospray ionization–tandem mass spectrometry�

Mark McDonald a,∗, Kristina Granelli b, Pernilla Sjoberg b

a Central Meat Control Laboratory, Department of Agriculture, Backweston Campus, Young’s Cross, Celbridge, Co. Kildare, Republic of Irelandb Chemistry Division 1, National Food Administration, Box 622, SE-75126 Uppsala, Sweden

Received 16 June 2006; received in revised form 19 December 2006; accepted 29 January 2007Available online 7 February 2007

bstract

Dexamethasone, betamethasone and prednisolone are synthetic glucocorticosteroids authorised for therapeutic use in bovine animals within theuropean Union. Dexamethasone and betamethasone are used mainly for the treatment of metabolic and inflammatory diseases. Prednisolone issed to treat bovine mastitis. Maximum residue limits (MRLs) of 0.3 �g kg−1 for both dexamethasone and betamethasone and 6.0 �g kg−1 forrednisolone in bovine milk have been established. 6�-Methylprednisolone and flumethasone are not authorised for use in bovine animals and areompletely banned in bovine milk. The proposed method is based on deprotenisation of milk using 20% (w/v) trichloroacetic acid. Samples areltered using glass microfibre filters and subject to clean-up using OASIS HLB solid phase extraction. Separation was achieved on a Hypercarb00 mm × 2.1 mm × 5 �m column. Mobile phase was: 90/10 acetonitrile/0.1% formic acid in water; flow rate was 600 �L min−1. The methodllowed the rapid identification and confirmation of the five glucocorticosteroids according to the criteria laid down in Commission Decision002/657/EC. Matrix calibration curves for all compounds were linear in the interval 0.0 MRL to 2.0 MRL with a correlation coefficient (r2)

igher than 0.96. Relative recoveries ranged from 97% for betamethasone to 111% for prednisolone. Precision at the MRL ranged from 3.8% forrednisolone to 13.8% for betamethasone. Decision limits, CC�, and detection capability, CC� have been calculated for all compounds.

2007 Elsevier B.V. All rights reserved.

eywords: Synthetic glucocorticosteroids; Dexamethasone; Betamethasone; Prednisolone; 6�-Methylprednisolone; Flumethasone; Milk; Liquid chromatography

klbadtccb

andem mass spectrometry; Validation

. Introduction

Synthetic glucocorticoids are widely used in veterinaryedicine, particularly for the treatment of anti-inflammatory

iseases, shock and circulatory collapse and acetonemia. Dex-methasone, betamethasone and prednisolone are specificallyuthorised for therapeutic use in bovine animals within theuropean Union. Under Community Regulation 2377/90 [1],

he European Medicines Agency (EMEA) set maximum residue

imits of 0.3, 0.3 and 6.0 �g kg−1, respectively for these residuesn bovine milk [2]. Residues of these catabolic componentslong with 6�-methyl prednisolone and flumethasone are also

� This paper was intended for inclusion in the Drug Residue Analysis SI buts now too late∗ Corresponding author. Tel.: +353 1 6157358; fax: +353 1 615 7359.

E-mail address: mark.mcdonald@agriculture.gov.ie (M. McDonald).

ai

otb[t

003-2670/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.aca.2007.01.075

nown for their growth promoting properties [3]. They increaseive weight gain even at low concentrations. This practise isanned by European legislation. From both an animal welfarend consumer protection perspective, there is an urgent need toevelop comprehensive control measures to monitor glucocor-icosteroids at low levels. A major analytical challenge for suchontrol is the development of rapid analytical methodology thatannot only detect such compounds at 0.5 MRL as now requiredy legislation, but also the chromatographic separation of dex-methasone and betamethasone which has proven to be difficultn the past.

Initial methods for the detection of such residues were basedn liquid chromatography (LC) with ultra violet (UV) detec-

ion. Such methods lacked specificity and were soon replacedy gas chromatography (GC) linked to mass spectrometry (MS)4,5]. The need for derivatisation prior to GC–MS analysis madehis methodology very time consuming. Improvements in liquid

a Chi

cmtm

tptatapcbbb

opanmt0secLmd2o

2

2

e(tApspe

6odOf(

2

staAMTcmw

2

wtaQWI(u96pemaisMva

TM

A

DBPD6F

M. McDonald et al. / Analytic

hromatography–tandem mass spectrometry (LC–MS-MS) noweans that the detection of such residues using triple quadrupole

andem mass spectrometers, particularly in multiple reactiononitoring mode (MRM), is possible.Antignac et al. [6] demonstrated the ability of LC–MS-MS

o detect glucocorticosteroids in hair/urine/tissue samples usingositive electrospray. Van den Hauwe et al. [7] have descriedhe detection of these compounds in similar matrices using neg-tive electrospray. This was possible due to the ionisation ofhese compounds with formic acid, producing [M + formate]−dducts which were then used as the precursor ion in multi-le reaction monitoring. Methods using atmospheric pressurehemical ionisation (APCI) and ion trap LC–MS-MS have alsoeen published in the literature [8–19]. Cherlet et al. [20] haveeen reported a method for the detection of dexamethasone inovine milk by LC–APCI-MS-MS.

No multiresidue method for the detection and confirmationf glucocorticosteroids in bovine milk has been published. Thisaper presents a rapid method for the detection, quantificationnd confirmation of dexamethasone, betamethasone, pred-isolone, flumethasone and 6�-methylprednisolone in bovineilk. Sample preparation is based on protein precipitation using

richloroacetic acid. A required performance level (RPL) of.3 �g kg−1 was set for 6�-methylprednisolone and flumetha-one. Clean-up was carried out using Oasis HLB solid phasextraction. Rapid separation is achieved using a Hypercarbolumn. Mass spectrometric detection is performed usingC–ESI-MS-MS in negative mode with multiple reactiononitoring. Analytical limits are determined and validation

ata is presented in accordance with Commission Decision002/657/EC [21]. The method may be used for the detectionf all 5 residues at 0.5 MRL/RPL.

. Experimental

.1. Chemicals

Chemicals were all of analytical-reagent grade unless oth-rwise stated. HPLC-grade acetonitrile was from Lab ScanDublin, Ireland). Gradient grade methanol, n-hexane andrichloroacetic acid were from Merck (Darmstadt, Germany).

20% trichloroacetic acid (TCA) solution (w/v) was pre-

ared by dissolving 200 g TCA in 1000 mL of water and aolution of 20 mM NaOH (Eka Nobel, Suite, Sweden) wasrepared by dissolving 0.8 g in 1000 mL of water. Dexam-thasone (DEX), betamethasone (BETA), flumethasone (FLU),

h0o1

able 1RM transitions, optimised cone voltages and collision energies

nalyte Cone (V) Transition 1 Co

examethasone 30 437.2 > 361.3 16etamethasone 30 437.2 > 361.3 18rednisolone 30 405.2 > 295.1 30eltafludrocortisone 30 423.0 > 347.1 15�-Methylprednisolone 30 419.2 > 343.3 15lumethasone 30 455.1 > 379.3 30

mica Acta 588 (2007) 20–25 21

�-methylprednisolone (MPRE) and prednisolone (PRE) werebtained from Sigma–Aldrich (Stockholm, Sweden). Deltaflu-rocortisone (DFUD) was from Steraloids (Newport, USA).ASIS HLB 3cc, 60 mg columns were supplied by Waters (Mil-

ord, USA). Whatman GF/B filters were supplied by WhatmanGmbH, Dassel, Germany).

.2. Standard solutions

Individual stock standard solutions at 1 mg mL−1 of eachtandard were made. From these stocks, 100 �g mL−1 solu-ions were prepared for PRE and DFUD (internal standard)nd 10 �g mL−1 solutions for FLU, MPRE, DEX and BETA.

mixed working standard was then prepared by diluting FLU,PRE, DEX and BETA to 50 ng mL−1 and PRE to 1 �g mL−1.

he individual working standard solution of DFUD had aoncentration of 1 �g mL−1. All solutions were prepared inethanol. The standards are stable for a minimum of 1 yearhen stored in the dark at −20 ◦C.

.3. Equipment

The water used was purified by a Milli-Q plus 185ater system, Millpore. Multifuge 35R and Biofuge 13 cen-

rifuges were from Hereaus. Whirlimixer was from Fisonsnd shaker from Edmund Buchler GmbH. A Micromassuattro Ultima mass spectrometer (Waters) connected to aaters Alliance 2695 Liquid Chromatograph system was used.

nstrument control was carried out using Masslynx softwareVersion 4.0). Separation was achieved using a hypercarb col-mn (100 �m × 2.1 �m × 5 �m), mobile phase was isocratic:0/10 acetonitrile/0.1% formic acid in water: flow rate was00 �L min−1. Column temperature was ambient room tem-erature. The mass spectrometer was operated in negativelectrospray ionisation (ESI−) mode. During method develop-ent standards were infused directly into mass spectrometer atconcentration of 1 �g mL−1 and a flow rate of 10 �L min−1

n electrospray negative mode. The [M + formate ion]− waselected as the precursor ion for all transitions. The two strongest

RM transition signals were selected for each compound. Coneoltage was 30 V and collision energies were optimised andre given in Table 1. Hexapole 1 was optimised to 30 V and

exapole 2 was optimised to 0.1 V, with the aperture set to.2 V and the multiplier to 650 V. Nitrogen was used as des-lvation gas at a flow of 500 L h−1 and cone gas at a flow of00 L h −1.

llision energy (eV) Transition 2 Collision energy (eV)

437.2 > 345.3 25437.2 > 345.3 25405.2 > 280.1 36423.0 > 293.2 30419.2 > 309.2 30455.1 > 325.3 15

2 a Chi

2

iamstacld

2

Aos

2

2

tiu(titsttst

2

3tuNtertcso

2

p2b

wery, 20 different milk samples, chosen at random, were spikedwith DEX, BETA, FLU, MPRE at 0.3 �g kg−1, and PRE at6.0 �g kg−1 and then analysed. The analysis was carried out

2 M. McDonald et al. / Analytic

.4. Calibration curves

Matrix matched calibration curves were prepared by spik-ng milk samples with glucocorticosteroids at 0.0, 0.5, 1.0, 1.5nd 2.0 MRL/RPL, note: for FLU and MPRE, a required perfor-ance level of 0.3 �g kg−1 was set. A fixed amount of internal

tandard, DFUD was added to all samples which were then quan-ified using internal relative response. Relative response of thenalyte was used and generated a linear regression calibrationurve using quanlynx software. The calibration curves were allinear in the 0.0 to 2.0 MRL/RPL range with all coefficients ofetermination (r2) > 0.96.

.5. Milk samples

Batches of milk samples were collected by the National Fooddministration from Kungsangens Forskingscentrum as partf the Swedish National Residue Control Plan. Samples weretored in frozen condition (−85 ◦C) until analysis.

.6. Sample preparation.

.6.1. ExtractionAfter thawing, 20 g of sample is weighed into a 50 mL cen-

rifuge tube and 240 �L of DFUD internal standard (L �g mL−1)s added to all samples. Samples are spiked with varying vol-mes of mixed working standard, FLU, MPRE, DEX and BETA50 ng mL−1) and PRE (1 �g mL−1) to the required concen-ration. Samples were vortexed for 30 s and allowed to standn the dark at 4 ◦C for 15 min. Two milliliters of 20% (w/v)richloroacetic acid is added. Samples were vortexed for 30 s,haken for 15 min (Edmund Buchler) and allowed to stand inhe dark at 4 ◦C for 30 min. Samples were subsequently cen-rifuged at 4000 × g for 10 min at 10 ◦C. Gravity filtering of theubsequent supernatant was carried out using Whatman GF/Bype glass microfibre filters.

.6.2. Clean-upOASIS HLB 3cc, 60 mg columns were conditioned with

mL of methanol and 3 mL of water as recommended byhe manufacturer. The sample solution was added to the col-mn. The columns were washed with 3 mL 40/60 (v/v) 20 mMaOH/methanol, 3 mL of water and then 3 mL of n-hexane. Elu-

ion was carried out with 3 mL of methanol. The eluant wasvaporated to dryness at 40 ◦C under a stream of nitrogen. Theesidue was redissolved in 100 �L of mobile phase and vor-exed for 30 s. The resulting aliquot was transferred to a microentrifuge tube and centrifuged at 16000 × g for 5 min. The finalolution was further transferred to an LC vial and 20 �L injectednto the LC–MS-MS system.

.7. Validation of the method

CC� and CC� were determined by spiking 20 different sam-les at the MRL/RPL on three different days (day 1, n = 8; day, n = 7; day 3, n = 5. For repeatability and reproducibility, oneatch of milk was analysed (n = 6) on three occasions over 5

Fa

mica Acta 588 (2007) 20–25

eeks, spiked at 0.5, 1.0 and 1.5 MRL/RPL. For relative recov-

ig. 1. Chromatographic separation of all six glucocorticosteroids in milk spikedt MRL/RPL level. Flow rate 600 �L min−1.

a Chi

oruostcda

3

3

aiwrabflIt1bTfMr

ir

tbl

tmPwrt

3

p2mserpts

M. McDonald et al. / Analytic

ver a 4-week period (day 1, n = 8; day 2, n = 7; day 3, n = 5). Allesults were calculated against matrix matched standard curvessing internal standard. Absolute recovery was determined onne day by comparing absolute peak area response for six blankamples spiked at the MRL/RPL before sample preparation tohe same six blank samples spiked after sample preparation andomparing peak areas. For this experiment, no internal stan-ard was used, and quantitation was carried out using externalbsolute response.

. Results and discussion

.1. Separation conditions

Optimum separation of the five glucocorticosteroids waschieved on a 100 mm × 2.1 mm Hypercarb column (5 �m) withsocratic mobile phase 90/10 acetonitrile/0.1% formic acid inater, see Fig. 1. Previous papers [7,13] have published a flow

ate of 0.22 mL min−1, a formic acid concentration of 0.3% andtotal run time of 30 min. Mobile phase flow rate was varied

etween 250 and 600 �L min−1. The effect of this change inow rate on chromatography and sensitivity is shown in Fig. 2.ncreasing the flow rate from 250 to 600 �L min−1 decreasedhe retention time of the last eluting compound, BETA, from1.18 to 4.47 min giving an overall run time of 6 min withaseline separation between DEX and BETA still observed.

he TIC (total ion count) almost doubled when increasing

rom 250 to 600 �L min−1 giving higher sensitivity at the lowRPL levels required. Further increasing the flow rate decreased

esponse. Optimisation experiments showed that no increase

ne1a

Fig. 2. Dexamethasone and Betamethasone retent

mica Acta 588 (2007) 20–25 23

n source/desolvation temperature or desolvation gas flow wasequired at elevated flow rates.

Experiments showed that by reducing the formic acid concen-ration from 0.3 to 0.1%, a two-fold increase in sensitivity coulde achieved again giving higher sensitivity at the low MRPLevels required.

In all cases, four identification points, one parent and tworansitions were monitored. Ion ratios all complied with the per-

itted tolerances outlined in Commission Decision 2002/657.RE had a relative ion ratio of 77% between transition 1 and 2,hilst FLU, DFUD, MPRE, DEX and BETA had relation ion

atios of 12, 22, 24, 8 and 7%, respectively. All relative retentionimes were within the ±2.5% tolerance permitted.

.2. Recoveries and trueness

Relative recovery was calculated by spiking individual sam-les at the MRL/RPL on three different days (day 1, n = 8; day, n = 7; day 3, n = 5). Results were calculated against matrixatched standard curves using internal standard added before

ample preparation and are presented in Table 2. Relative recov-ries ranged from 97% for BETA to 111% for PRE. Absoluteecoveries were determined on one day by comparing absoluteeak area response for six individual blank samples spiked athe MRL/RPL before sample preparation to the same six blankamples spiked after sample preparation. In this case, no inter-

al standard was used, and quantitation was carried out usingxternal absolute response. Absolute recoveries ranged from5.1% for MPRE to 27.2% for PRE. The high relative recoveriesnd use of internal standard ensure that the method is suitable

ion times at 0.25, 0.40 and 0.60 mL min−1.

24 M. McDonald et al. / Analytica Chimica Acta 588 (2007) 20–25

Table 2Relative recovery, trueness and absolute recovery data for all five glucocorticosteroids at the MRL/RPL

Analyte Level (�g kg−1) Relative recovery (%) Trueness (%) Absolute recovery (%)

Dexamethasone 0.3 106.0 6 18.3Betamethasone 0.3 97.0 −3 15.8Prednisolone 6.0 111.3 11.3 20.26�-Methylprednisolone 0.3 108.3 8.3 15.1Flumethasone 0.3 108.7 8.7 27.2

Table 3Precision data at 0.5, 1.0 and 1.5 MRL/RPL

Analyte Precision (CV%), week 1 (n = 6) Precision (CV%), week 2 (n = 6) Precision (CV%), week 3 (n = 6)

0.5 MRL 1.0 MRL 1.5 MRL 0.5 MRL 1.0 MRL 1.5 MRL 0.5 MRL 1.0 MRL 1.5 MRL

Dexamethasone 5.6 6.8 4.6 5.5 4.8 4.0 8.1 6.6 8.2Betamethasone 4.7 8.6 7.3 14.6 11.9 21.0 9.8 13.8 18.0P .96 .9F .2

fr

3

RdwwComplv

3b(dCaCa

TC

A

DBP6F

M8

4

MtfcsDrcA

A

Fa

redinsolone 7.4 3.9 3.2 12�-Methylprednisolone 7.3 8.6 5.1 9lumethasone 7.3 7.9 2.9 4

or the determination of glucocorticoids in milk at the levelequired.

.3. Precision, CCα and CCβ

Precision was calculated as CV% at the 0.5, 1.0 and 1.5 MRL/PL level using the same sample and determined on the sameay. Results are shown in Table 3. Overall, good precisionas observed for DEX, MPRE, PRE and FLU at all levels,ith no CV% > 10% except for PRE at 0.5 MRL on week 2.V% for BETA was >10% on 5 occasions and >20% once,n week 2 at 1.5 MRL. This higher CV% observed for BETAay be due to the lower ion count observed for this com-

ound in comparison to the other analytes, with a correspondingower signal to noise ratio giving rise to a higher relativeariance.

CC� and CC� were calculated in accordance with Sections.1.2.5 and 3.1.2.6 of Commission Decision 2003/657/EC [21]y fortifying samples at the MRL/RPL on three different daysday 1, n = 8; day 2 n = 7; day 3 n = 5) and calculating stan-ard deviations. Values are given in Table 4. For all compounds,

−1 −1

C� is <0.4 �g kg , with FLU the lowest at 0.367 �g kgnd BETA the highest at 0.393 �g kg−1. All compounds hadC� < 0.5 �g kg−1 with FLU again the lowest at 0.408 �g kg−1

nd BETA the highest at 0.496 �g kg−1. PRE with a higher

able 4C� and CC� values for glucocorticosteroid compounds in milk

nalyte MRL/MRPL(�g kg−1)

CC�

(�g kg−1)CC�

(�g kg−1)

examethasone 0.3 0.38 0.45etamethasone 0.3 0.39 0.50rednisolone 6.0 7.67 8.67�-Methylprednisolone 0.3 0.38 0.43lumethasone 0.3 0.37 0.41

SS

R

3.8 5.7 6.8 4.1 7.24.1 3.9 9.4 7.3 9.44.6 5.0 3.9 2.8 5.0

RL at 6.0 �g kg−1 had a CC� of 6.673 �g kg−1 and CC� of.672 �g kg−1.

. Conclusions

Baseline separation of DEX and BETA along with PRE,PRE and FLU can be achieved with a total LC–MS-MS run

ime of 6 min. This rapid elution, along with low levels oformic acid in the mobile phase, increases sensitivity for allompounds. The same parent could be used for both MRM tran-itions monitored. Validated in accordance with Commissionecision 2003/657/EC, this method is currently in used for the

apid multi-residue confirmation and quantification of gluco-orticosteroids in bovine milk at the Swedish National Fooddministration.

cknowledgements

The Author would like to thank the Irish Departments ofinance and Agriculture for both funding and support. We wouldlso like to thank all those involved in the Irish–Swedish Civilervice exchange programme, especially the Host Institute, thewedish National Food Administration, Uppsala.

eferences

[1] EEC Council Regulation No. 2377/90/EC, Off. J. Eur. Commun. (1990)L224.

[2] EEC Council Regulation No. 2593/99/EC, Off. J. Eur. Commun. (1999)L315.

[3] Ph. Gaignage, G. Lognay, D. Bosson, D. Vertongen, Ph. Dreze, M.Marlier, M. Severn, Eur. J. Drug. Metabol. Pharmackinet. 16 (1991)

219.

[4] K. De Wasch, H. de Brabender, I. Vandereyt, P. Batjoens, K. Vanoosthuyze,C. Van Peteghem, Analyst 119 (1994) 2557.

[5] O.H. Hidalgo, M.J. Lopez, E.A. Carazo, M. San Andres Larrea, B.A. Thea,Reuvers, J. Chromatogr. B 788 (2003) 137–146.

a Chi

[

[[

[

[

[

[[

[

M. McDonald et al. / Analytic

[6] J.-P. Antignac, F. Monteau, J. Negriolli, F. Andre, B. LeBizec, Chro-matographia (1/2) (2004) 59.

[7] O. Van den Hauwe, F. Dumoulin, C. Elliott, C. Van Peteghem, J. Chro-matogr. B 817 (2005) 215–223.

[8] M.J. O’Keeffe, S. Martin, L. Regan, Anal. Chim. Acta 483 (2003) 341–350.[9] S.R. Savu, L. Silvestro, A. Haag, F. Sorgel, J. Mass Spectrom. 31 (1996)

1351–1363.10] J.-P. Antignac, B. LeBizec, F. Monteau, F. Poulain, F. Andre, J. Chromatogr.

B 757 (2001) 11–19.

11] P.W. Tang, W.C. Law, T.S.M. Wan, J. Chromatogr. B 757 (2001) 229–244.12] A.A.M. Stolker, P.L.W.J. Schwillens, L.A. Van Ginkel, U.A.Th. Brinkman,

J. Chromatogr. A 893 (2000) 55–67.13] O. Van den Hauwe, F. Dumoulin, J.P. Antignac, M.P. Bouche, C. Elliott,

C. Van Peteghem, Anal. Chim. Acta 473 (2002) 127.

[

[[

mica Acta 588 (2007) 20–25 25

14] G.N.W. Leung, E.W. Chung, E.M. Ho, W.H. Kwok, D.K.K. Leung, F.P.W.Tang, T.S.M. Wan, N.H. Yu, J. Chromatogr. B 825 (2005) 47–56.

15] F. Guan, C.E. Uboh, L.R. Soma, Y. Luo, J. Rudy, T. Tobin, J. Chromatogr.B 829 (2005) 56–68.

16] Y.R. Nii, P.O. Edlund, J. Pharm. Biomed. Anal. 37 (2005) 341–350.17] P. Munoz, J. Blanca, M. Ramos, M. Bartolome, E. Garcıa, N. Mendez,

J. Gomez, M. Martın de Pozuelo, Anal. Chim. Acta 529 (2005)137–144.

18] G. Brambilla, F. Buiarelli, G.P. Cartoni, F. Coccioli, C. Colamonici, A.

Fagiolo, C. Giannini, B. Neri, J. Chromatogr. B 755 (2001) 265–278.

19] R. Draisci, C. Marchiafava, L. Palleschi, P. Cammarata, S. Cavalli, J. Chro-matogr. B 753 (2001) 217–223.

20] M. Cherlet, S. De Baere, P. De Backer, J. Chromatogr. B 805 (2004) 57–65.21] Commission Decision 2002/657/EC, Off. J. Eur. Commun. (2002) L221.