HiO

Wa

b

c

a

ARRAA

KMQTBV

1

wdpidopfff(2

af

h0

Journal of Chromatography A, 1347 (2014) 122–128

Contents lists available at ScienceDirect

Journal of Chromatography A

jo ur nal ho me pag e: www.elsev ier .com/ locate /chroma

igh-throughput screening of pesticide and veterinary drug residuesn baby food by liquid chromatography coupled to quadrupolerbitrap mass spectrometry

ei Jiaa,b, Xiaogang Chua,b,∗, Yun Lingb, Junrong Huanga, James Changc

College of Chemistry & Chemical Engineering, Shaanxi University of Science & Technology, Xi’an 710021, ChinaInstitute of Food Safety, Chinese Academy of Inspection and Quarantine, Beijing 100123, ChinaThermo Fisher Scientific, 355 River Oaks Parkway, San Jose, CA 95134, USA

r t i c l e i n f o

rticle history:eceived 10 March 2014eceived in revised form 19 April 2014ccepted 22 April 2014vailable online 1 May 2014

eywords:ulti-residue methods-Orbitrap

a b s t r a c t

A new analytical method was developed and validated for simultaneous analysis of 333 pesticide andveterinary drug residues in baby food. Response surface methodology was employed to optimize ageneric extraction method. Ultrahigh-performance liquid chromatography and electrospray ionizationquadrupole Orbitrap high-resolution mass spectrometry (UHPLC-ESI Q-Orbitrap) was used for the sep-aration and detection of all the analytes. The method was validated by taking into consideration theguidelines specified in Commission Decision 2002/657/EC and SANCO/12571/2013. The extraction recov-eries were in a range of 79.8–110.7%, with coefficient of variation <8.3%. The 333 compounds behavedynamic in the range 0.1–1000 �g kg−1 concentration, with correlation coefficient >0.99. The limits of

−1

argeted-screeningaby foodalidation

detection for the analytes are in the range 0.01–5.35 �g kg . The limits of quantification for the analytesare in the range 0.01–9.27 �g kg−1. This method has been successfully applied on screening of pesti-cide and veterinary drugs in ninety-three commercial baby food samples, and tilmicosin, fenbendazole,tylosin tartrate and thiabendazole were detected in some samples tested in this study. The present studyis very useful for fast screening of different food contaminants.

. Introduction

Pesticide and veterinary drugs are usually small moleculareight chemicals essential for treating infections, increasing pro-uction, and improving animal husbandry. However, the potentialresence of contaminants and drug residues is an important issue

n the field of food and animal feed safety [1,2]. Infants and chil-ren represent a vulnerable risk group of the population in termsf multi-residue toxicity and stringent regulations have been set torotect them from dietary exposure to these chemicals [3–5]. Babyoods combine a wide range of different matrices: cereal-based

ood (CBF), meat-based food (MBF), powdered milk-based infantormulae (PMBIF), non-fatty based on fruit (FBF) and vegetableVBF) [6,7]. The European Commission (Directives 2006/141/EC and003/89/EC) specified the general Maximum Residue Levels (MRLs)

∗ Corresponding author at: Institute of Food Safety, Chinese Academy of Inspectionnd Quarantine, Beijing 100123, China. Tel.: +86 010 85778904;ax: +86 010 857707750.

E-mail addresses: foodjiawei@aliyun.com, foodjiawei@yahoo.com.cn (X. Chu).

ttp://dx.doi.org/10.1016/j.chroma.2014.04.081021-9673/© 2014 Elsevier B.V. All rights reserved.

© 2014 Elsevier B.V. All rights reserved.

of 10 �g kg−1 for any individual pesticide residue in baby food. Asa result of lack of regulation for veterinary drug residues, a “zero-tolerance” policy is applied for the veterinary drug residues in babyfood, which means that the presence of these compounds is ille-gal at any level [8]. Consequently, these regulated limits requirethe analysis of extremely high number multi-residues in baby food[9–11].

Therefore, efficient, sensitive and accurate methods have beendeveloped for the analysis of each group of residues separately.The determination of pesticide or veterinary drugs in baby foodhas been based on pressurized liquid extraction (PLE), liquid–liquidextraction (LLE), matrix solid-phase dispersion (MPSD) or modifiedQuEChERS methodology [12–17]. Some of these procedures havebeen followed by a clean-up step based on solid-phase extraction(SPE) or dispersive SPE, and subsequent chromatographic analy-sis by ultra-high-performance liquid chromatography coupled totandem mass spectrometry (UHPLC–MS/MS) [12,14].

Normally, most of these methods are focused on specific groups

of residues, not being suitable for wide-scope multi-residue anal-ysis. To be able to analyze pesticide and veterinary drugs with awide variety of physicochemical properties simultaneously, non-selective, generic sample-preparation procedures are applied. The

togr. A

m“bemtbiqimtUitrpmomhFiimibolo

tftotwsf

2

2

cSfIptm−BfrT1atmsStw

W. Jia et al. / J. Chroma

ost frequently reported generic sample-preparation methods aredilute and shoot” and QuEChERS methods [18–21]. A clear draw-ack of these strategies is the occurrence of abundant matrixffects, which compromise method selectivity, detection limits,aintenance frequency and quantitative aspects [22,23]. Never-

heless, the lack of selectivity in generic sample preparation cane compensated by selectivity in instrumental analysis. Although

n terms of sensitivity, the use of LC or GC coupled to tripleuadrupole tandem MS (QqQ) is usually preferred. However, QqQ

nstruments are not sufficiently selective and only provide unitass resolution. In each retention time window, the number of

ransitions to be acquired can as such be kept relatively low.sing this approach, the number of analytes that can be acquired

n one single run is limited [24–26]. An attractive alternative ishe use of full scan high resolution mass spectrometry (theo-etically, no limitations in number of monitored analytes). Mostublished high resolution mass spectrometry based multi-class,ulti-residue methods are based on Orbitrap, time-of-flight (TOF),

r quadrupole-time-of-flight (QqTOF) [27–29]. Typically, Orbitrapass spectrometry is expected to provide resolving powers of not

igher than 100,000–200,000 with detection times of 1–2 s [30].rom last year the role of UHPLC-Q-Orbitrap and related techniquess increasingly built up as enabling tool in food safety analysis fort can provide product-ion spectra with accurate mass measure-

ent, that permit unequivocal confirmation of detailed structuralnformation. In spite of the potential value of the application, to theest of our knowledge, so far no one has reported the applicationf Q-Orbitrap mass spectrometry combined with high performanceiquid chromatography for simultaneous determination for a groupf pesticide and veterinary drugs in foods [31,32].

Bearing in mind the lack of works related to the determina-ion of several classes of pesticide and veterinary drugs in babyood, in this paper, we describe the development of a cost-effective,ime-efficient and easy-to-use sample preparation method basedn QuEChERS for the simultaneous extraction of more than 330 pes-icide and veterinary drugs in different types of baby foods. Coupledith an optimized UHPLC-Q-Orbitrap method, this method was

uccessfully applied on screening of multi-residues in baby foodrom local market.

. Experimental

.1. Chemicals and reagents

Pesticide and veterinary drugs analytical standards were pur-hased from LGC Standards (Teddington, UK), Fluka (Buchs,witzerland), Sigma–Aldrich (Steinheim, Germany), Dr. Ehrenstor-er GmbH (Augsburg, Germany) and Witega (Berlin, Germany).ndividual stock standard solutions (500–1000 mg L−1) were pre-ared, depending on the specific solubility properties, by dissolvinghe appropriate quantity of the compound mainly in acetone (ACE),

ethanol (MeOH) or acetonitrile (MeCN), and were stored at20 ◦C. HPLC-grade ACE, MeCN and MeOH were sourced from J.T.aker (Deventer, Holland). A solution (100–300 mg L−1) for each

amily of pesticide and veterinary drug was prepared from cor-esponding individual stock standard solution in MeCN or MeOH.hen, a multi-compound working solution at a concentration of0 mg L−1 of each compound was prepared by combining suitableliquots of each individual standard stock solution and dilutinghem with MeOH. Acetic acid, formic acid (FAc), ammonium for-

ate, sodium acetate, sodium chloride and anhydrous sodium

ulfate (Na2SO4) were of analytical grade and purchased fromigma–Aldrich (Steinheim, Germany). Ultrafree-MC centrifugal fil-er devices (0.22 �m) of Millipore (Millipore, Brussels, Belgium)ere used. Trifluoro acetic acid was obtained from Fluka (Buch,

1347 (2014) 122–128 123

Switzerland). Ultrapure Water (resistivity, 18.2 M�) was purifiedon a Milli-Q Plus apparatus (Millipore, Brussels, Belgium).

2.2. Instrumentation

The UHPLC-Q-Orbitrap system consisted of an Accela 1250 LCpump and a CTC Analytics PAL open autosampler coupled with aQ Exactive mass spectrometer (Thermo Fisher Scientific, Bremen,Germany). The system was controlled by Exactive Tune 1.1 andXcalibur 2.2 software (Thermo Fisher Scientific, San Jose, USA).

2.3. Analytical procedure

2.3.1. Sample preparationThe sample preparation protocol was developed using modified

literature protocols and personal experience [31,32]. After homog-enization on a Polytron PT-2000 (Kinematica, Switzerland) for 30 s,5.0 g of each sample was weighed in polypropylene centrifuge tubes(50 mL), fortified with the three hundred and thirty-three differentanalytes and let to stand for 15 min. 10 mL volume of a MeCN/watersolution (84/16, v/v) with 1% acetic acid was added as an extractionsolvent and the tube was tightly capped and vigorously mixed for1 min using a vortex (Scientific Industries, New York, USA) mixer atmaximum speed. Na2SO4 (6 g), sodium acetate anhydrous (1.45 g)and ceramic homogenizers were added to the tube, to induce phaseseparation. After that, the tube was immediately shaken for 1 min,and then centrifuged for 5 min at 2264 × g at 4 ◦C (Beckman Couler,Brea, USA). An aliquot of the final upper layer (200 �L) was trans-ferred into a Mini-UniPrep vial, 300 �L MeOH and 500 �L 8 mMammonium formate buffer were added. After the vial was capped,vortexed for 30 s. 1 mL of the sample extract was taken and filteredthrough a Millex-GN nylon filter (0.22 �m, Pall Corporation, Har-bor, USA). The cleaned extract was collected in a vial for injectioninto the UHPLC-Q-Orbitrap system.

A total of ninety-three different baby foods (including VBF, MBF,CBF, FBF and PMBIF) were analyzed. These samples were obtainedfrom different markets and all of them were analyzed followingthe procedure described above. Those samples found to contain noresponse at the retention times of reference compounds or metabo-lite were selected for use as negative controls and stored at 4 ◦Cprior to analysis.

2.3.2. Experimental design for response surface methodology(RSM)

Response surface methodology (RSM) was employed to investi-gate the variations in recovery rates with respect to the preparationof conditions including extraction solvent volume, the amounts ofsodium acetate, and MeCN. The optimal composition of the 3 vari-ables was determined by using a central composite design (CCD)approach. In this work, the full CCD consisted of (1) a completetwo-factorial design; (2) n0, center point (n0 > 1), and (3) two axialpoints on the axis of each design variable at a distance of = 2.000from the design center. Hence, a total number of design points ofN = 2k + 2k + n0 was used. The actual variable was coded to facili-tate multiple regression analysis. The complete design consistedof 15 combinations including seven replicates of the center pointwith five degrees of freedom for calculation of errors in the exper-iments. The optimal values of response Y (individual recovery ofinterest compounds) were obtained by solving the regression equa-tion and by analyzing the response surface contour plots. Table 1

indicates the coded and CCD-processed variables for the opti-mization of the QuEChERS method for samples. The resulting 15experiments were carried out randomly. The goodness of fit of theregression model and the significance of parameter estimates were

124 W. Jia et al. / J. Chromatogr. A 1347 (2014) 122–128

Table 1Variables and levels evaluated in the central composite design to optimize the extraction conditions.

Independent variable Unit Symbol Coded levels

− −1 0 +1 +˛

6

dt

2

TcUfnbT0Td5

eaet3f(tlwowt2fta

3

tagicc

3

teiwttaot

Extraction solvent volume mL X1

Na acetate quantity g X2

MeCN content % X3

etermined through appropriate statistical methods. Design Expertrial version 8.0.6.0 was used (Stat-Ease Inc., Minneapolis, MN).

.3.3. Instruments and analytical conditionsThe analytical column used was a 100 mm × 2.1 mm, i.d., 2.6 �m,

hermo Accucore C-18 aQ connected to a 10 mm × 2.1 mm, Accu-ore C-18 aQ guard column (Thermo Fisher Scientific, San Jose,SA). The mobile phase consisting of 0.1% (v/v) FAc and ammonium

ormate 4 mM in water (eluent A) and 0.1% (v/v) FAc and ammo-ium formate 4 mM in MeOH (eluent B). The gradient method isased on the procedure previously described by Zhang et al. [33].he following gradient was used: 0 min 100% A, 1 min 100% A, 7 min% A, 12 min 0% A, 13 min 100% A, until the end of the run at15 min.he sample injection volume was set at 10 �L. All the three hun-red and thirty-three analytes eluting over 0–9.5 min while the last.5 min were used for column cleaning and re-equilibration.

The mass spectrometer Q-Orbitrap was equipped with a heatedlectrospray ionization (HESI) source. The optimized HESI temper-ture was set at 350 ◦C, the capillary temperature at 320 ◦C, thelectrospray voltage at 3.5 kV and 3.0 kV for positive and nega-ive modes, respectively. Sheath and auxiliary gas were 18 and

L min−1. All quantitative data in this study were acquired usingull MS scan mode. The mass resolution was set at 70,000 FWHMm/z 200). In full MS/dd-MS2 (TopN), which is used for confirma-ory purpose, the Q-Orbitrap performs data-dependent scans. Asong as the targeted compounds were detected, precursor ions that

ere selected by the quadrupole were sent to the HCD collision cellf the Q-Orbitrap mass spectrometer. Here, they were fragmentedith normalized collision energy (NCE) to obtain product ion spec-

ra. At this stage, the mass resolution was set at 17,500 FWHM (m/z00) and NCE 35%. Using this strategy, coeluting matrix compoundsrom the matrix or noisy peaks can be easily excluded, facilitatinghe identification and quantification of known or novel analytes in

single run analysis.

. Results and discussion

In this work, 258 pesticides have been studied, includingriazoles, neocotinoids, organophosphorus, carbamates, triazinesnd organochlorines. Furthermore, 75 veterinary drugs includingestagens, macrolides, androgens, quinolones, non-steroidal anti-nflammatory drugs, tetracyclines, ionophores, sulphonamides,orticoids, avermectines, tranquilisers, nitroimidazoles, ampheni-ols, coccidiostats, �-agonists and penicillins were also selected.

.1. Optimization of the HPLC-Q-Orbitrap conditions

Chromatographic conditions were studied in order to achievehe best separation and retention for the compounds. First, severalxperiments were performed on different mobile phases consist-ng of MeCN or MeOH as organic phase and water as polar phase

ith different concentrations of acetic and formic acid (from 0.05o 0.5%), ammonium formate, ammonium acetate (from 1 mM

o 10 mM). MeOH was chosen as the organic phase because itchieved a better resolution and sensitivity than MeCN. On thether hand, the addition of FAc–ammonium formate provided bet-er results than acetic acid–ammonium acetate and it was used to

2 6 10 14 180.5 1.0 1.5 2.0 2.50 70 80 90 100

improve the ionization efficiency and achieve better chromatogra-phy. Moreover, the addition of FAc also helped in the separationof the sulphonamides isomers (sulfadimethoxine and sulfadox-ine, sulfaclozine sodium monohydrate and sulfachloropyridazine),without which, these isomers overlapped and could not be quan-tified separately. Finally, the best results were obtained whenMeOH was used as organic modifier and aqueous solution of FAc(0.1%)–ammonium formate (4 mM) was employed. We evaluatedvarious HPLC columns (Thermo Scientific Accucore: aQ C18, RP-MS,C18, PFP, Phenyl-Hexyl) in order to optimize the chromatographicseparation of the above mentioned three hundred and thirty-threeanalytes and the aQ C18 column yielded the best results. Severalgradient profiles were studied, obtaining good response with thegradient described in Section 2. Under these conditions retentiontimes of the analytes were constant, ranging from 3.87 (amoxicillintrihydrate) to 9.19 (narasin) min (Table S1, Electronic Supplemen-tary Material).

Supplementary Table S1 related to this article can befound, in the online version, at http://dx.doi.org/10.1016/j.chroma.2014.04.081.

The optimum mass spectrometric parameters for the identifi-cation and quantification of the three hundred and thirty-threeanalytes were first obtained after analyzing the compounds byflow injection analysis respectively. Three consecutive injectionson-column at 10 �g kg−1 were carried out to detect any decreasein mass accuracy. Sensitivity of target analytes was checkedby recording chromatograms in full scan method in both pos-itive and negative ionization mode. Due to adduct formationwith FAc/ammonium formate buffer, some analytes exhibit strongammonium adduct species ([M+NH4]+) which appear to be themost predominant ions in the mass spectrum.

3.2. Optimization of the extraction procedure

Despite QuEChERS uses MeCN acidified with acetic acid, andbearing in mind that conventional extraction procedures of multi-residues from different samples use a mixture of MeCN/water orMeOH/water, the extraction solvent was evaluated first, using theextraction procedure described in Section 2.3, except that the typeof solvent was varied. Thus, different solvents such a mixture ofMeCN/water (84:16, v/v), a mixture of ACE/water (84:16, v/v), anda mixture of MeOH/water (84:16, v/v) were checked. The obtainedresults are shown in Table 2. It can be observed many analyteshad low recovery or could not be recovered with the mixturecontaining ACE. Furthermore, often a pronounced matrix-effectand RSD > 10% were observed. Probably, the mixture of ACE/watercauses a very rapid denaturation of proteins, with possible for-mation of aggregates that could adsorb/occlude the compoundsof interest, reducing extraction reproducibility. The MeOH/watermixture performed worst in term of matrix interferences, the finalextracts were turbid, probably due to insufficient protein precipi-tation and the amount of fat present in the final extract. In terms of

matrix-effect and extraction recoveries, MeCN/water (84:16, v/v)was found to be the most suitable extraction solvent across alltested baby foods. MeCN/water (84:16, v/v) was chosen as theextraction solvent because it good miscibility with water, and its

W. Jia et al. / J. Chromatogr. A

Table 2Comparison of recoveries for three proposed extraction methods for of three hun-dred and thirty-three pesticide and veterinary drugs in five baby foods.

Matrix Extraction Percent of 333 analytes in recovery range

<30 30–50 50–70 70–120 >120

CBF MeCN/water 0 0 0 99 1MeOH/water 0 0 11 89 0ACE/water 2 6 13 74 5

MBF MeCN/water 0 0 0 100 0MeOH/water 0 0 15 85 0ACE/water 4 7 21 62 6

PMBIF MeCN/water 0 0 0 93 7MeOH/water 0 0 16 84 0ACE/water 3 5 18 66 8

FBF MeCN/water 0 0 0 100 0MeOH/water 0 0 2 94 4ACE/water 0 3 1 91 5

VBF MeCN/water 0 0 0 100 0MeOH/water 0 1 2 94 3

al

cloi(tswai

rcsbrvtcspwecia

sFmwtdavFrecc

used for their confirmation are characteristic of each one. Thus,

ACE/water 0 2 3 88 7

bility to recover the analytes without extracting high quantities ofipophilic material.

It is well known that macrolides and tetracyclines can formhelation complexes with cations, especially those containing highevels of Ca2+ and Mg2+. For this reason, a solution of EDTA isften added in the extracting buffer as a competing agent, to facil-tate extraction of tetracyclines [34]. We tried adding Na2EDTA0.1 M) to the optimized extracting MeCN/water (84:16, v/v) mix-ure, but recovery of macrolides and tetracyclines showed only alight increase, while a general decrease for the other compoundsas observed. For the same chelation problems as for macrolides

nd tetracyclines, in the extraction procedure Na2SO4 was chosennstead of the more commonly used MgSO4.

Critical factors which could have an influence on the analyteecovery and method sensitivity were investigated by a centralomposite design (CCD). In general, higher amount of extractionolvent leads to a higher analytical signal in almost all analytes,ut there was no preliminary study performed testing differentatios of mass (g) of sample per volume (mL) of extraction sol-ent. Compared with those from a non-buffered QuEChERS method,he acetate buffered method improved stabilities and recoveries ofertain pH-dependent compound (e.g. �-lactams, quinolones andome macrolides). Sodium acetate also has a dissolving effect onrotein and fat globules, which could affect recovery rates. Theater amount was critical in order to achieve satisfactory recov-

ries for macrolides and tetracyclines without having losses for theoccidiostats which are unstable in water. Therefore, the factorsncluded (1) volume of extraction solvent, (2) amount of sodiumcetate, (3) MeCN content in the mixture solvent.

In order to verify if the models generated and explained aignificant part of the variations in the experimental data, the-test for significance was carried out. It can conclude that theodel was well-fitted to the observations and the regressionas statistically significant at the 99% confidence level. Among

he linear, quadratic, and cross-product forms of the indepen-ent variables, the coefficient constant, X1, X2, X3, X1

2, X22,

nd X32 were significant at the level of p < 0.05. The Model F-

alue of 2178.2 implies the model is significant. The “Lack ofit F-value” of 0.251 implies the Lack of Fit is not significantelative to the pure error. Therefore, after the response recov-

ry rates were experimentally determined under the 14 sets ofonditions, the regression coefficients for the recoveries were cal-ulated and a polynomial regression model equation was fitted as

1347 (2014) 122–128 125

follow: Y = 99.12 + 7.82X1 + 0.52X2 + 6.40X3 + 0.15X1 × X2 − 0.15X1 ×X3 − 0.85X2 × X3 − 9.42X1

2 − 17.42X22 − 13.80X3

2.The individual recoveries of all analytes were introduced sepa-

rately as the response in the statistical program. Fig. 1 shows thedifferent response plots for enoxacin (as an example) for differentcombinations of the parameters investigated. The optimum con-ditions were chosen by taking into consideration data obtainedfrom the response surface pots and the regression coefficient plots.Therefore, the response surfaces generated suggest that the bestextraction conditions for enoxacin were 10 mL MeCN/water (84:16,v/v) mixture with sodium acetate 1.45 g. Multi-residual meth-ods for recovery were tested under the conditions optimized forenoxacin, which gave the highest observed recovery in the singlerecovery tests and were therefore used to test the multi-residualcompound quantification. The results of the multi-residue meth-ods recovery test and validation information for each analyte areshown in Table S2. In summary, the resulting conditions allowedreliable simultaneous analysis of three hundred and thirty-three ofthe target analytes with recoveries in the range of 79.8–110.7%.

Supplementary Table S2 related to this article can befound, in the online version, at http://dx.doi.org/10.1016/j.chroma.2014.04.081.

3.3. Development of screening and confirmation methods

In order to increase sample throughput, only one injection wascarried out and fragment spectra were acquired in order to distin-guish between negative and positive samples (screening method)and to confirm the presence of the analytes (confirmation method).

Full MS scan mode allowed for screening and quantifying theanalytes or retrospectively looking into unknowns. The identifica-tion of the target compounds was based on the measurement ofthe accurate mass and the retention time (RT). The RT window tol-erance was set at a very conservative ±3 SD (RT average minusor plus 3 times the standard deviation of the RT). The effect ofthe mass extraction window on selectivity for analytes in babyfoods at a concentration 50 �g kg−1 (2–10 ppm mass deviations)was tested. The best results were obtained when mass extractionwindows of 2.5 ppm were employed, providing a high selectivityand reduced probability of false positives. If a signal within the RTwindow showing a mass error <2.5 ppm was found within the RTwindow, and the intensity of this signal was equal or above thelower value of the uncertain region (8.3 × 104), this sample wasconsidered as a non-negative sample and it was re-processed toconfirm the results by the confirmation method. On the contrary, ifno signal was found within the RT window or the measured masserror was higher than 2.5 ppm, the sample was considered negativeand it was not re-processed by the confirmation method.

In order to develop the confirmation method, it was neces-sary to establish fragments for each compound. When operatedin full MS/dd-MS2 mode, a product ion spectrum with accuratemass measurement is obtained automatically according to inclu-sion list (a list of targeted accurate masses), and this defined as adata dependent acquisition (dd-MS2). The resolving power (R) is acritical parameter which has an impact on the correct assignmentof the masses for the analytes. To eliminate matrix interferencesthe resolving power of the Q-Orbitrap was evaluated. With an Rof 70,000 FWHM (17,500 FWHM for dd-MS2) no overlapping inmasses was observed between analytes and matrix components.Table S1 summarizes the optimal parameters of the UHPLC-Q-Orbitrap. Furthermore, it is important to notice, that despite somestructural analogs and multiple isomers can co-elute, the fragments

betamethasone and dexamethasone are co-eluting multiple iso-mers but unequivocal fragments (m/z 237.12737, 135.08054) havebeen selected for their confirmation (Fig. 2). Specific fragments

126 W. Jia et al. / J. Chromatogr. A 1347 (2014) 122–128

rface

hw((cipM

F(

Fig. 1. Response su

ave been found for each analytes, minimizing the possibility of arong compound confirmation. On the other hand, sulphonamides

e.g. sulfadimethoxine and sulfadoxine) have a common fragmentsm/z 311.08069, 156.07678, 108.04475, 92.04999), but they do not

o-elute (Table S1). In order to minimize probability of false pos-tives, the non-negative samples were carried out using isotopicatterns. The relative isotope abundance (RIA) of the M+1 (being

the corresponding [M+H]+) peak, which is mainly due to the

ig. 2. Extracted ion chromatogram and spectra from a full MS/dd-MS2 experiment usingdisplayed as a stick per scan) the protonated molecule of betamethasone and dexametha

plot for enoxacin.

presence of 13C, was useful to confirm analytes with a high num-ber of carbons. When characteristic atoms, such as S, Br, and Clwere present in the molecule, M+2 were used for identificationpurposes. For the one hundred and ninety-six sulphur-, bromine-,

or chlorine-containing pesticide and veterinary drugs, M+2 werethe most sensitive isotope. In all other cases, M+1 were the mostabundant isotope. For instance, Fig. 3 shows the theoretical andexperimental spectra of triclabendazole in a spiked MBF sample

UHPLC-Q-Orbitrap, corresponding to the theoretical extracted ion chromatogramssone (m/z 393.20718) and fragments in a spiked MBF sample (50 �g kg−1).

W. Jia et al. / J. Chromatogr. A 1347 (2014) 122–128 127

riclabe

(M

3

ssu

mc(

C

Fig. 3. Experimental and theoretical spectra of t

50 �g kg−1). It can be observed that both spectra are similar and+1 and M+2 ions can be used for confirmation purposes.

.4. Validation of the proposed method

Method validation was performed in terms of matrix-effect,pecificity, linearity, trueness, precision, CC� and CC�. Commis-ion Decision 2002/657/EC and SANCO/12571/2013 protocol weresed as guidelines for the validation studies.

The matrix effect was tested by comparing the slopes of theatrix-free calibration curves to the matrix-matched calibration

urves. Matrix effect was investigated by calculating the percentage

C%) of signal enhancement or suppression, according to Eq. (1).

% =(

1 − ss

sm

)× 100 (1)

Fig. 4. UHPLC-Q-Orbitrap matrix effects of five baby foods. The

ndazolein in a spiked MBF sample (50 �g kg−1).

where ss is the slope of calibration plot with matrix-matchedcalibration solutions and sm is the slope of calibration plot withcalibration solutions in solvent.

Fig. 4 shows the obtained results for each baby food, and itcan be observed that the results depend on the type of matrixevaluated. In the three hundred and thirty-three analytes, threehundred and thirty-one did not display significant changes in theslopes between standard calibration curves and matrix-matchedcalibration curves (−50% < C% < +50%) [35]. The remaining two com-pounds that had significant matrix effect were namely levamisolhydrochloride and albendazole (C% < −50% or C% > +50%). In conse-quence, matrix matched calibration should be used to compensatematrix effect.

Extraction recoveries were assessed by spiking blank dairysamples before and after extraction at three concentration lev-els (CC�, 2 × CC�, 4 × CC�) with five replicates at each level. Thematrix matched standard calibration curves from the extracts were

evaluation was done at 10 �g kg−1 equivalent in samples.

128 W. Jia et al. / J. Chromatogr. A

Table 3Quantification results for target analytes in positive baby foods analyzed by UHPLC-Q-Orbitrap.

Sample Compound Concentration (�g kg−1)

FBF 2 Coumaphos 1.92CBF 8 Azoxystrobin 2.31VBF 6 Thiacloprid 2.12MBF 3 Tilmicosin 22.34MBF 9 Fenbendazole 4.90

Tylosin tartrate 15.81MBF 14 Levamisol hydrochloride 3.67PMBIF 1 Albendazole 1.45PMBIF 17 Trimethoprim 4.97PMBIF 18 Oxfendazole 21.0PMBIF 22 Tilmicosin 16.53PMBIF 23 Thiabendazole 3.04

t0rapatWfnactcsstaodwaoimdoFetts

3

iantadttbto

[

[

[[

[

[[

[[[[[[

[[

[[[[

[

[

[[[

[Chem. 138 (2013) 827.

[35] J. Wang, W. Chow, D. Leung, J. AOAC Int. 94 (2011) 1685.

herefore evaluated with 11 calibration levels S0–S11, namely 0,.1, 0.5, 5, 20, 50, 100, 200, 500, 750, 1000 �g kg−1, where “0” cor-esponded to the extract from the sample analyzed as such. Forll analytes, r2 was greater than 0.99 and the deviation of eachoint from the calibration line was lower than 15%. Specificity wasssessed by verifying the presence of interference at the reten-ion time of analytes greater than a signal-to-noise ratio of three.

ithin-laboratory reproducibility was assessed by spiking babyood samples at three different concentrations nine times. True-ess was calculated as the percentage of error between spikednd found concentrations. CC� was estimated from the calibrationurve prepared by spiking baby food samples at four concentra-ion levels in the low concentration range. CC� is calculated as theoncentration corresponding to the y-intercept plus 2.33 times itstandard deviation. In the case of CC�, the concentration corre-ponds to CC� + 1.64 s, s being the standard deviation obtained athe CC� level. CC� ranged between 0.01 �g kg−1 and 5.35 �g kg−1,nd CC� ranged between 0.01 �g kg−1 and 9.27 �g kg−1.The resultsf this validation are summarized in Table S2. Compared with tra-itional methods, the sensitivity was enhanced, and the accuracyas improved, leading to a powerful method for screening pesticide

nd veterinary drug residues in foods. Three consecutive injectionsn-column at 10 �g kg−1 were carried out to detect any decreasen mass accuracy. No significant decrease was observed and the

aximum mass deviation ranged from 0.1 to 2.0 ppm. This minoreviation in mass accuracy demonstrated the wide dynamic rangef Q-Orbitrap for qualitative analysis at a resolving power of 70,000WHM. The CC� values of the proposed Q-Orbitrap mass spectrom-ter method are comparable to the limits of the previously reportedriple quadrupole mass spectrometer methods. Comparing withhe detection limits reported in literatures [36–38], the detectionensitivity and accuracy was improved more than 5 times.

.5. Sample analysis

One of the proposed methods were optimized and validated,t was applied to investigate the occurrence of the three hundrednd thirty-three pesticide and veterinary drug residues in a total ofinety-three commercial baby food samples. As shown in Table 3,ilmicosin, fenbendazole, tylosin tartrate, levamisol hydrochloride,lbendazole, trimethoprim, oxfendazole, and thiabendazole wereetected in MBF and PMIF samples, only coumaphos, azoxystrobin,hiaclopridand were detected in CBF, FBF and VBF samples. Withhis UHPLC-Q-Orbitrap method, not only accuracy was enhanced,ut also the low concentration preservative, this suggested that

he UHPLC-Q-Orbitrap method was appropriate for the screeningf pesticide and veterinary drug residues in foods.

[[[

1347 (2014) 122–128

4. Conclusions

A new analytical method has been developed and applied inroutine for screening and quantitation of pesticide and veterinarydrug residues in baby food. In summary, by combining QuEChERS-based extraction procedure and UHPLC-Q-Orbitrap, an accurateand highly sensitive method was developed to screen three hun-dred and thirty-three pesticide and veterinary drug residues inbaby food.

Acknowledgments

We wish to thank Thermo Fisher Scientific (San Jose, USA) forproviding us with UHPLC-Q-Orbitrap. We would also like to thankJames Chang for the technical support. The present research wasfinancially supported by the grants from the project of GeneralAdministration of Quality Supervision, Inspection and Quarantineof the People’s Republic of China (Project number: 2012104002;2013IK198), and the Graduate Innovation Fund of Shaanxi Univer-sity of Science and Technology.

References

[1] A. Rico, P.J. Van den Brink, Sci. Total Environ. 468/469 (2014) 630.[2] Y.D. Zhang, N. Zheng, R.W. Han, B.Q. Zheng, Z.N. Yu, S.L. Li, S.S. Zheng, J.Q. Wang,

Food Control 36 (2014) 238.[3] F. Della Betta, L. Vitali, R. Fett, A.C.O. Costa, Talanta 122 (2014) 23.[4] Y. Li, J.C. Beghin, Food Policy 45 (2014) 57.[5] E.M. Rintala, P. Ekholm, P. Koivisto, K. Peltonen, E.R. Venäläinen, Food Chem.

150 (2014) 199.[6] A. López, G.A. Javier, J. Fenoll, P. Hellín, P. Flores, J. Food Compos. Anal. 33 (2014)

39.[7] G.B. Pouokam, G.C. Ajaezi, A. Mantovani, O.E. Orisakwe, C. Frazzoli, Sci. Total

Environ. 481 (2014) 296.[8] M.M. Aguilera-Luiz, J.L. Martinez-Vidal, R. Romero-Gonzalez, A. Garrido

Frenich, Food Chem. 132 (2012) 2171.[9] R.R. Bouhouch, S. Bouhouch, M. Cherkaoui, A. Aboussad, S. Stinca, M.

Haldimann, M. Andersson, M.B. Zimmermann, Lancet Diabetes Endocrinol. 2(2014) 197.

10] L. Matumba, M. Monjerezi, T. Biswick, J. Mwatseteza, W. Makumba, D. Kaman-gira, A. Mtukuso, Food Control 39 (2014) 87.

11] C.M. Mesch, M. Stimming, K. Foterek, A. Hilbig, U. Alexy, M. Kersting, L. Libuda,Appetite 76 (2014) 113.

12] X. Hou, S. Lei, S. Qiu, L. Guo, S. Yi, W. Liu, Food Chem. 153 (2014) 121.13] F.Y. Hu, L.M. He, J.W. Yang, K. Bian, Z.N. Wang, H.C. Yang, Y.H. Liu, J. Chromatogr.

B 949/950 (2014) 79.14] M.W. Kujawski, Z. Barganska, K. Marciniak, E. Miedzianowska, J.K. Kujawski, M.

Slebioda, J. Namiesnik, LWT-Food Sci. Technol. 56 (2014) 517.15] X. Ma, J. Wang, Q. Wu, C. Wang, Z. Wang, Food Chem. 157 (2014) 119.16] S.K. Wadhwa, M. Tuzen, T.G. Kazi, M. Soylak, B. Hazer, Food Chem. 152 (2014)

75.17] E. Watanabe, Y. Kobara, K. Baba, H. Eun, Food Chem. 154 (2014) 7.18] D.Y. Bang, S.K. Byeon, M.H. Moon, J. Chromatogr. A 1331 (2014) 19.19] C. Caballo, M.D. Sicilia, S. Rubio, Talanta 119 (2014) 46.20] J. Li, D. Liu, T. Wu, W. Zhao, Z. Zhou, P. Wang, Food Chem. 151 (2014) 47.21] S. Walorczyk, Talanta 120 (2014) 106.22] G. Aslan-Sungur, E.O. Gaga, S. Yenisoy-Karakas , J. Chromatogr. A 1325 (2014)

40.23] M.T. Jafari, M. Saraji, H. Sherafatmand, Anal. Chim. Acta 814 (2014) 69.24] E. Engel, J. Ratel, P. Blinet, S.T. Chin, G. Rose, P.J. Marriott, J. Chromatogr. A 1311

(2013) 140.25] M. Farré, Y. Picó, D. Barceló, J. Chromatogr. A 1328 (2014) 66.26] P. Steinbach, W. Schwack, J. Chromatogr. A 1323 (2014) 28.27] R. Diaz, M. Ibánez, J.V. Sancho, F. Hernández, J. Chromatogr. A 1276 (2013) 47.28] M.M. Gómez-Ramos, C. Ferrer, O. Malato, A. Agüera, A.R. Fernández-Alba, J.

Chromatogr. A 1287 (2013) 24.29] M. Rodriguez-Aller, R. Gurny, J.L. Veuthey, D. Guillarme, J. Chromatogr. A 1292

(2013) 2.30] M.L. Gómez-Pérez, P. Plaza-Bolanos, R. Romero-González, J.L. Martínez-Vidal,

A. Garrido-Frenich, J. Chromatogr. A 1248 (2012) 130.31] M. Castro-Puyana, M. Herrero, Trend Anal. Chem. 52 (2013) 74.32] W. Jia, X.G. Chu, Y. Ling, J.R. Huang, Y.H. Lin, J. Chang, J. Sep. Sci. 37 (2014) 782.33] A. Zhang, J.S. Chang, C. Gu, M. Sanders, Thermo Scientific Application Note

51878, 2010.34] J. Zhan, Y.Y. Zhong, X.J. Yu, J.F. Peng, S.B. Chen, J.Y. Yin, J.J. Zhang, Y. Zhu, Food

36] X. Wang, S. Wang, Z. Cai, Trend Anal. Chem. 52 (2013) 170.37] G. Kaklamanos, U. Vincent, C. von Holst, J. Chromatogr. A 1293 (2013) 60.38] G. Kaklamanos, U. Vincent, C. von Holst, J. Chromatogr. A 1322 (2013) 38.