G Model

CHROMA-359034; No. of Pages 10 ARTICLE IN PRESS

Journal of Chromatography A, xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

Full Length Article

Optimization for QuEChERS extraction of mycotoxins and veterinary

drugs by response surface methodology for application to egg
and milk
Jian Zhou a,b , Jiao-Jiao Xu a,∗ , Jin-Mi Cong b , Zeng-Xuan Cai a , Jing-Shun Zhang a ,
Jun-Lin Wang a , Yi-Ping Ren c,∗
a
Lab of Physicochemical Research, Department of Physicochemical & Toxicology, Zhejiang Provincial Center for Disease Control and Prevention, Zhejiang
310051, China
b
Zhejiang Provincial Key Lab of Health Risk Appraisal for Trace Toxic Chemicals, Ningbo Municipal Center for Disease Control and Prevention, Ningbo
315010, China
c
National Center for Food Safety Risk Assessment Application Technology Cooperation Center, Yangtze Delta Region Institute of Tsinghua University,
Zhejiang 314006, China

a r t i c l e i n f o a b s t r a c t

Article history: A multiclass method was proposed for the simultaneous determination of various classes of veterinary
Received 12 June 2017 drugs (n = 65), mycotoxins and metabolites (n = 39) in egg and milk by ultra-high performance liquid
Received in revised form chromatography-tandem mass spectrometry. The contaminants were extracted by QuEChERS-based
19 November 2017
strategy including salt-out partitioning and dispersive solid-phase extraction for cleanup further. With
Accepted 21 November 2017
the aim of maximizing throughput and extraction efficiency, Plackett-Burman design was employed ini-
Available online xxx
tially for screening significant variables. And response surface methodology based on central composite
design was conducted to achieve optimal conditions in details: 3.35% (v/v) of formic acid in acetonitrile,
Keywords:
QuEChERS
1.2 g of NaCl, 0.5 g of anhydrous NaAc, 300 mg of C18 and 140 mg of primary secondary amine. Satisfactory
Egg analytical characteristics in validation, in aspects of accuracy (70%–105% for mycotoxins and quinolones,
Milk 55%–80% for sulphonamides and 40%–105% for other veterinary drugs), precision (inter-day RSDs < 14%)
Mycotoxins and sensitivity (LOQs ranged from 0.01 ␮g/kg to 31 ␮g/kg), were achieved under the optimized con-
Veterinary drugs ditions. The matrix effects were evaluated and compensated by the use of matrix-matched calibration
Response surface methodology curves (R2 > 0.987). In practice, 45 eggs and 30 milk samples were investigated by the established method,
of which positive finding aflatoxin in milk and sterigmatocystin in eggs.
© 2017 Elsevier B.V. All rights reserved.

1. Introduction health, which could cause problems of allergy, antibiotic resistance

Veterinary drugs have been widely prescribed for both
therapeutic and prophylactic reasons. But for improper usage, non-
respect of withdrawal periods and cross-contamination, the drugs
could be transferred and accumulated in animal-derived products
[1]. Because of high consumption of animal-derived products, like
egg and milk, the presence of drug residues poses a risk to public
Abbreviations: ACN, acetonitrile; CCD, central composite design; ESI, elec-
trospray ionization; FA, formic acid; LOQ, limit of quantification; ME, Matrix
effect; MeOH, methanol; MRL, maximum residue limit; MRM, multiple-reaction-
monitoring; PSA, primary secondary amine; QNL, quinolone; RSM, response surface
methodology; S/N, signal-to-noise; SA, sulphonamide; SPE, solid-phase extraction.
∗ Corresponding authors.
E-mail addresses: jjxucdc@163.com (J.-J. Xu), renyiping@263.net (Y.-P. Ren).
and even carcinogenic character.
Mycotoxins contamination is another serious worldwide prob-
lem on food and feed, which contaminated 25% of grain and oil crops
approximately every year reported by FAO. The mainly concerned
toxins includes aflatoxins, ochratoxin A and trichothescenes, which
were classified in group 1, group 2 and group 3 by the Interna-
tional Agency of Research on Cancer, respectively. These substrates
could be transferred to animal-derived food by contaminated feed
and bio-transformed into various metabolites, e.g. aflatoxin M1 and
hydrolyzed fumonisin B1 /B2 , which also represents related parental
compound toxicity and risk to human health [2–4].
The maximum residue limits (MRLs) for veterinary drugs and
maximum limit (MLs) for mycotoxins in food and feed have been
established in EU [5,6] and China [7,8]. Hereinto, MRLs for QNLs
and SAs in animal-derived food are in the range of 10 ␮g/kg to
1900 ␮g/kg, and 100 ␮g/kg for total amount, respectively. However,

https://doi.org/10.1016/j.chroma.2017.11.050
0021-9673/© 2017 Elsevier B.V. All rights reserved.

Please cite this article in press as: J. Zhou, et al., Optimization for QuEChERS extraction of mycotoxins and veterinary drugs by response
surface methodology for application to egg and milk, J. Chromatogr. A (2017), https://doi.org/10.1016/j.chroma.2017.11.050
G Model
CHROMA-359034; No. of Pages 10 ARTICLE IN PRESS
2 J. Zhou et al. / J. Chromatogr. A xxx (2017) xxx–xxx

no specific criteria have been legally established for mycotoxins in to design the fitted models, estimate the interaction effects and
animal-derived matrices, due to the lack of representative data and determine the optimum conditions, respectively [23].
studies. In this case, it is of great importance to develop generic, The aim of this present study was to develop a reliable and uni-
accurate and sensitive methods for these contaminants analysis in versal QuEChERS method coupling with ultra-high performance
food of animal origin. liquid chromatography tandem mass spectrometry (UPLC–MS/MS),
With the increasing number and class of organic residues and for the simultaneous determination of 104 veterinary drugs
contaminants to be found in raw materials and/or processed food, (mainly QNLs and SAs), mycotoxins and its metabolites in egg and
consumers have strong willing to get more information about milk. To obtain the optimal extraction efficiencies, the statisti-
toxicological implications and other potential risk relating to the cal methodologies were implemented in details. The established
presence of these residues and contaminants. In this sense, it is method was expected to contribute to the risk monitoring and
essential to approach reliable and suitable analytical methods that investigation of contamination level in eggs and milk.
allow to analyze a huge number of compounds simultaneously,
including several classes or more than one family.
2. Experimental
To date, few studies have been proposed for simultaneous deter-
mination of multi-classes in food and feed matrices, especially for
2.1. Standards and chemicals
veterinary drugs and mycotoxins together. Mol et al [9] proposed
one of the first methods for the simultaneous analysis of pesticides,
HPLC-grade acetonitrile (ACN) and methanol (MeOH) were
mycotoxins and plant toxins in feed and several types of food matri-
purchased from Merck (Darmstadt, Germany). Formic acid (FA,
ces, as well as Dominicis et al [10] established targeted screening
purity > 99%) was purchased from Anaqua Chemicals (Houston,
method applied for bakery ingredients and food commodities. A
USA). Analytical grade of MgSO4 , NaCl, Na2 SO4 , anhydrous NaAc
similar approach was used by Capriotti et al [11] for the simulta-
and QuEChERS sorbents (C18 and primary secondary amine (PSA))
neous determination of antimicrobials and mycotoxins in egg. In
were provided by local suppliers. Ultrapure water was obtained
order to achieve a suitable elution of the target compounds and
from Millipore (Bedford, MA, USA). Immunoaffinity columns for
to reach good sensitivity, more than one injection was performed.
sterigmatocystin were provided by R-Biopharm (Glasgow, Scot-
Zhan et al determined 255 veterinary drugs, pesticides and myco-
land). Analytical standards of mycotoxins were purchased from
toxins in milk [12], 220 chemical residues and mycotoxins in infant
Sigma-Aldrich (St. Louis, USA), Biopure (Tulln, Austria), Toronto
formula [13] and 226 veterinary drugs and other contaminants
Research Chemicals Inc. (Toronto, Canada), ALEXIS (Lausen,
in muscle [14]. However, two or three injections were necessary,
Switzerland) and Pribolab (Qingdao, China), respectively. The stan-
analysis time was longer. And in the previous reports, there is less
dards of hydrolyzed fumonisin B1 /B2 were prepared by the basic
residues’ or contaminants’ metabolites considered.
hydrolysis of fumonisin B1 /B2 as described in a previous report
Bearing in mind that complex components in matrices and
by G. Pagliuca et al [24]. The achieved product solution was fur-
a great variability in physicochemical properties of residues
ther confirmed by UPLC–MS/MS and as a result, no residues of
and contaminants, generic extraction and chromatographic con-
the parental fumonisin toxins were detectable. The standards of
ditions would be used and optimized. Besides of solid phase
veterinary drugs were all purchased from Dr. Ehrenstorfer GmbH
extraction and dilute-and-shoot for purification and enrichment,
(Augsburg, Germany).
QuEChERS (acronym of Quick, Easy, Cheap, Effective, Rugged, and
The standard stock solution of each analyte was prepared indi-
Safe) methodology was a significant method for high-throughput
vidually in the suitable solvents: ACN/0.02% FA aqueous solution
determinations and widely utilized in multi-classes targets pre-
(50/50, v/v) for QNLs; ACN for all other analytes. The solu-
treatment [11,15]. It was initially proposed for pesticide residues
tions were stored at −20 ◦ C in the dark and were stable for
[16], and had subsequently been expanded to antibiotics, food addi-
at least 6 months. Then, two different working solutions were
tives and mycotoxins due to the merits of rapidity, universality and
prepared by combining suitable aliquots of each stock solu-
effectiveness. For individual class of veterinary drugs or mycotox-
tion and diluting them with appropriate volume of ACN for
ins analysis, Stubbings et al. established a multiclass UPLC–MS/MS
mycotoxins and 50% ACN for drugs. The specific concentrations
procedure in animal tissue using QuEChERS approach [17], as well
were set as follows: aflatoxin B1 /B2 /G1 /G2 and ochratoxin A/B
as 10 mycotoxins in eggs at trace level reported by Frenich et al. [18].
(500 ng/mL); sterigmatocystin (2 ␮g/mL); T-2 toxin, diacetoxyscir-
And Moretti et al. developed a multiclass method for screening and
penol, neosolaniol, deoxynivalenol-3-glucuronide and tentoxin
confirmatory analysis of 62 antibiotics in milk [19]; Capriotti et al.
(5 ␮g/mL); HT-2 toxin, de-epoxydeoxynivalenol, alternariol methyl
proposed a multiclass screening method based on QuEChERS tech-
ether, chaetoglobosin and ochratoxin ␣ (10 ␮g/mL); deoxyni-
nique for the determination of antimicrobials and mycotoxins in
valenol, 3/15-acetyl-deoxynivalenol, fusarenone X and fumonisin
eggs [11]. Although there is a growing tendency to develop multi-
B1 (100 ␮g/mL); fumonisin B2 (50 ␮g/mL) and fumonisin B3
class methods in the field of food safety, the greater diversity in the
(25 ␮g/mL); zearalenone and its derivatives (6 ␮g/mL); afla-
chemical properties between veterinary drugs and mycotoxins, has
toxin M1 /M2 (50 ng/mL); verruculogen and gliotoxin (40 ␮g/mL);
impeded progress to combine them into analytical experiments.
cyclopiazonic acid (0.1 ␮g/mL); penicillic acid (12.7 ␮g/mL) and
Otherwise, in the most of previous studies, every single factor
hydrolyzed fumonisin B1 /B2 (11.2 ␮g/mL); veterinary drugs
was usually optimized using one-factor-at-a-time approach whilst
(1 ␮g/mL, except 200 ng/mL of amantadine and chloramphenicol,
all the other factors were fixed at a constant level [20,21]. The
100 ng/mL of metronidazole). In this paper, the concentration of
single-dimensional optimization is incapable of distinguishing the
aflatoxin B1 was expressed as a representative for convenience.
importance of each factor and reaching the true optimum, due to
the ignorance of the interaction effects among factors [22]. In this
context, statistical experimental approaches, e.g. fractional facto- 2.2. Samples
rial (Plackett-Burman) design and response surface methodology
(RSM) could optimize the parameters collectively and eliminate the Totally, 75 fresh samples including eggs (n = 45) and milk (n = 30)
limitations of conventional optimization process. Plackett-Burman were purchased from local markets in Zhejiang province. Egg sam-
design provides an effective way to identify the main factors among ples (both albumen and yolk) were homogenized thoroughly for
a large number of variables and reduce the number of experiments. 3 min using a blender (IKA, German). In spite of milk, as homoge-
RSM, which includes factorial design and regression analysis, helps neous liquid, no extra operation step was done. All the samples

Please cite this article in press as: J. Zhou, et al., Optimization for QuEChERS extraction of mycotoxins and veterinary drugs by response
surface methodology for application to egg and milk, J. Chromatogr. A (2017), https://doi.org/10.1016/j.chroma.2017.11.050
G Model
CHROMA-359034; No. of Pages 10 ARTICLE IN PRESS
J. Zhou et al. / J. Chromatogr. A xxx (2017) xxx–xxx
were analyzed before the expiration date and stored in the dark at
−20 ◦ C.
2.3. Instrument
The UPLC–MS/MS system was composed of Waters AcquityTM
UPLC and mass spectrometer, namely Xevo TQ-S (Manchester, UK),
which equipped with Z-spray electrospray ionization (ESI) inter-
face. Acquisition was performed in multiple-reaction-monitoring
(MRM) mode, and the specific MS/MS parameters such as the selec-
tion of precursor and product ion, retention time, cone voltage
and collision energy, were optimized by direct infusion of standard
solutions into mass spectrometer and summarized in Supplemen-
tary Table 1. Chromatographic separation was performed with an
Acquity UPLC CORTECS column (3.0 mm × 150 mm, 1.6 ␮m) at 40 ◦ C
column temperature. A flow rate of 0.4 mL/min was used, and the
injection volume was set at 10 ␮L. The detailed gradient elution
program is reported in Supplementary Table 2 as well as the ioniza-
tion source parameters in Supplementary Table 3. Data acquisition
and processing was performed using MassLynx 4.1 and TargetLynx
4.1 program.
2.4. Samples preparation
The extraction of analytes from matrices was achieved by
a QuEChERS-based approach. Briefly, aliquots 2.50 g ± 0.01 g of
homogenized eggs (5.00 g milk) were weighted into a 50-mL cen-
trifuge tube. Subsequently, 5.0 mL of water (2.5 mL of water for
milk) and 10.0 mL of ACN containing 3.35% of FA (v/v) were added
in tube. The mixture was vortexed vigorously for 2 min and then
subjected to ultrasonic extraction for 25 min (vortexed in every
five-minute interval). After that, the tube involved the addition of
4.0 g of anhydrous Na2 SO4 , 1.2 g of NaCl and 0.5 g of anhydrous
NaAc, which was shaken by hand for 3 min persistently and then
centrifuged at 8500 rpm for 3 min (7606 rcf, 4 ◦ C). Subsequently,
3.0 mL of the upper organic layer was transferred to a 15-mL dis-
persive tube containing 300 mg of C18 , 140 mg of PSA and 1.5 g
of anhydrous Na2 SO4 . The mixture was vortexed for 3 min and
then centrifuged. An aliquot of 2.0 mL of the resulting supernatant
(equivalent to 0.50 g egg and 1.00 g milk respectively) was trans-
ferred into a new glass tube and evaporated under a gentle nitrogen
flow at 45 ◦ C. Finally, the residue was reconstituted with 1.0 mL
of ACN/water (20/80, v/v) and the obtained solution was passed
through a 0.2-␮m polytetrafluoroethylene filter.
2.5. Method validation
Analytical characteristics of the proposed method were estab-
lished by a validation procedure with spiked egg and milk samples,
in terms of linearity, matrix effect (ME), accuracy, repeatability,
inter-day precision, limits of quantification (LOQs) and selectivity.
The methodological linearity was assessed by spiking blank
extract (both two matrices) at six different concentration levels
prior to analysis. The concentration ranges for analytes were as fol-
lows: 0.5, 1.0, 2.0, 5.0, 20.0 and 50.0 ng/mL for veterinary drugs (0.1,
0.2, 0.4, 1.2, 4.0 and 10.0 ng/mL for amantadine and chlorampheni-
col; 0.05, 0.01, 0.2, 0.5, 2.0 and 5.0 ng/mL for metronidazole) and
0.062, 0.12, 0.25, 0.5, 1.2 and 5.0 ng/mL for aflatoxin B1 . Signal sup-
pression or enhancement effect that caused by matrix co-elution
interferences, i.e. ME, was assessed for each analyte by comparing
the slope of pure standard curve with the slope of matrix-matched
standard (spiked extract) curve.
Accuracy was studied by spiking blank samples at three concen-
tration levels (10, 20 and 50 ␮g/kg for veterinary drugs; 2, 4 and
10 ␮g/kg for amantadine and chloramphenicol; 1, 2 and 5 ␮g/kg
for metronidazole; 2, 4 and 6 ␮g/kg for aflatoxin B1 ), processing
Please cite this article in press as: J. Zhou, et al., Optimization for QuEChERS extraction of mycotoxins and veterinary drugs by response
surface methodology for application to egg and milk, J. Chromatogr. A (2017), https://doi.org/10.1016/j.chroma.2017.11.050
3
six replications each level on the same day (repeatability) whereas
three consecutive days for inter-day precision (n = 6 × 3). LOQs of
analytes in two matrices were calculated by spiking various low
concentration levels and determined as the lowest concentrations
that produce chromatographic peaks at signal-to-noise ratio (S/N)
of 10. The selectivity of method was analyzed by comparing the
chromatograms of analyte-free samples and the spiked ones, to
check if interference peaks existed at the same retention time as
targets of interest.
3. Results and discussion
Given that the wide range of chemical properties for analytes,
unknown concentration level of contaminants (maybe at low level),
and the complex components in egg and milk (mainly protein
and lipid), the extraction and clean-up procedure was of great
significance in developing multiclass methods. To obtain the opti-
mal experimental conditions, a two-step design combination, i.e.
Plackett-Burman design and central composite design (CCD), was
used for screening and further optimization, respectively.
3.1. Optimization of extraction procedure
To reduce the handling steps and increase sample throughput,
a universal QuEChERS approach was performed. Several variables
that affected the extraction efficiency of analytes from matri-
ces, were optimized by analyte-spiking at medium level. Based
on the previous report, a mixture of MeOH/water (80:20, v/v)
with 1% acetic acid was initially tested [18]. Nevertheless, inad-
equate phase separation and protein precipitation were actually
observed after the addition of drying salts. Subsequently, the mix-
tures of ACN/water (80:20, v/v) and ACN/water (80:20, v/v) with
1% FA were checked as possible candidates. Interestingly, with the
employment of ACN-based extractant, efficient separation between
water and organic phase were achieved. By comparison, it could
be found that there were no significant differences on the recov-
eries of most analytes using the two extraction systems, except
carboxylic compounds, e.g. fumonisins and ochratoxin A. There-
fore, the acidified acetonitrile aqueous solution was selected for
further experiments. Additionally, EDTA-Mcllvaine solution was
usually applied as chelating agent for the extraction of veterinary
drugs from protein-rich matrices, e.g. eggs and milk. However, the
employment of EDTA-Mcllvaine solution exhibited poor extrac-
tion recoveries for most mycotoxins, which might be related to the
buffer effect.
Another important characteristic of QuEChERS procedure
focused on the selection of drying salts, which could induce phase
separation and affect analytes partition. Until now, numerous com-
binations of QuEChERS salts had been proposed. MgSO4 , NaCl,
Na2 SO4 and NaAc were the most suitable candidates [11,18,25].
Thus six groups of combined salts were taken into account:
(a) 4.0 g Na2 SO4 + 1.0 g NaCl; (b) 4.0 g Na2 SO4 + 1.0 g NaAc; (c)
4.0 g NaAc + 1.0 g NaCl; (d) 4.0 g MgSO4 + 1.0 g NaCl; (e) 4.0 g
MgSO4 + 1.0 g NaAc; and (f) 4.0 g MgSO4 + 1.0 g NaCl + 1.0 g NaC-
itrate + 0.5 g NaSequihydrate. The above salt combinations were
added to the spiked extractant and then the separated organic layer
was collected and analyzed. As presented in Supplementary Fig.
S1, the combinations of either NaAc or NaCl with Na2 SO4 provided
more favorable extraction efficiency (recovery > 60%) than that of
MgSO4 , which might be related to the complexing reaction between
veterinary drugs and metallic ions (Mg2+ ). Although both NaAc
and NaCl provoked distinct layering effects, superfluous addition
of NaAc could reduce the recoveries of fumonisins sharply due to
the buffering effect. In conclusion, the combination of Na2 SO4 , NaCl
and NaAc was adopted as QuEChERS salts whilst the amount of NaCl
and NaAc should be further optimized in details.
G Model
CHROMA-359034; No. of Pages 10 ARTICLE IN PRESS
4 J. Zhou et al. / J. Chromatogr. A xxx (2017) xxx–xxx
3.2. Optimization strategy for preparation conditions
Various parameters may affect the extraction efficiency of ana-
lytes by QuEChERS approach, e.g. the volume of aqueous solution,
the acidity of extractant, extraction time, salting-out effect and
the amount of QuEChERS sorbents. In this context, a Plackett-
Burman design was conducted at the beginning of experimental
optimization and it is intended to narrow the investigation range of
candidate variables. After the significant factors are selected, opti-
mization design, i.e. RSM based on CCD follow, which is used to
search the optimal conditions. Experimental design and statistical
analysis were performed by Design-expert 8.0.6.0 (Stat-Ease Inc.,
Minneapolis, USA), Minitab 17.1.0 (Minitab Inc., USA) and IBM SPSS
Statistics 19 (SPSS Inc., Chicago, USA).
3.2.1. Screening the significant parameters
A two-level Plackett-Burman design, was implemented to
screen the significant parameters among eight parameters: the
volume of water (X1 ), FA percentage in ACN (X2 ), extraction time
(X3 ), amount of QuEChERS salts and sorbents, i.e. Na2 SO4 (X4 ),
NaCl (X5 ), NaAc (X6 ), C18 (X7 ) and PSA (X8 ), which may affect the
QuEChERS method efficiency. The responses, i.e. the recoveries
of each analyte, were obtained in accordance with the designed
matrix (n = 20,) in Table 1 and t-test with 95% confidence level
was conducted to determine the significant factors. The effects of
the investigated variables are shown in the form of Pareto charts
(the evaluation results for aflatoxin B1 were represented in Sup-
plementary Fig. S2). The standardized effect of investigated factor
exceeding the reference line is considered significant to the results
with 95% probability. All analytes were subjected to t-test in turn
and from the obtained results, the parameters of X8 , X5 , X6 and X2
showed statistically significant influence to 72, 60, 52 and 60 ana-
lytes, respectively. These four factors were selected and subjected
to further optimization by CCD. By contrast, the other four fac-
tors exhibited no significant effects on the responses (be significant
variables to less than 15 analytes).
3.2.2. Central composite design
After selecting the four important factors, those were optimized
using a CCD approach. Totally, thirty experimental runs with six
replications at the center point of cubic domain were conducted
to reduce the uncontrollable influences. Each factor was studied at
five levels and the designed matrix was shown in Supplementary
Table 4. The results of each analyte obtained from CCD were fitted to
a polynomial equation with quadratic multiple regressions, which
express the relationship between response and variable as follow
[26]:

f

f

f

f additive forms. In this case, the 104 analytes were infused directly
Y = ı0 + ıi Xi + ıii Xi2 + ıij Xi Xj + ␧
i=1 i=1 i=1 j=1
Where, Y is the predicted response, Xi and Xj represent independent
variables, ı0 is the compensation term while ␧ is the experimental
error. The coefficient ıi , ıii and ıij represents the linear, quadratic
and interaction coefficient, respectively. For instance, the second-
order polynomial equation of aflatoxin B1 was generated as below:
Recovery (%) = 81.94 − 8.21 × A + 5.48 × B + 2.57 × C + 1.12
×D + 1.84 × AB + 0.23 × AC + 0.88 × AD − 2.54 × BC
−1.15 × BD − 1.28 × CD − 1.26 × A2 − 2.79 × B2
−0.40 × C 2 + 0.29 × D2
Please cite this article in press as: J. Zhou, et al., Optimization for QuEChERS extraction of mycotoxins and veterinary drugs by response
surface methodology for application to egg and milk, J. Chromatogr. A (2017), https://doi.org/10.1016/j.chroma.2017.11.050
The model summary statistics (P-value < 0.0001) and lack-of-fit
test (P-value of 0.1402) indicated that the conducted model was
significant, and the variations occur in responses should be associ-
ated to the polynomial model, rather than with the pure error. The
equation suitability and the importance of model items determined
using the analysis of variance (ANOVA) approach are summarized
in Supplementary Table 5. Hereinto, the independent variables of
A, B and C, interaction terms of AB and BC, and quadratic effects of
B2 had significant effects on the response of aflatoxin B1 at 95% con-
fidence level (P < 0.05). The coefficients of determination (R2 ) were
calculated by least-square regression analyses and applied to eval-
uate the model fitness and the quality of the equation. According to
Ranjbari [27], the R2 should be at least 0.800 for validation of good
agreement between the experimental data and the fitted models.
As for aflatoxin B1 , R2 obtained from ANOVA was 0.944, which indi-
cated that the established polynomial equation could explain 94.4%
of the variability in response. In conclusion, all analytes were sub-
jected to ANOVA and satisfactory R2 values were achieved, which
varied from 0.829–0.980.
The three-dimensional plots were highly recommended as
graphical interpretations for the interaction effects, which were
mapped against two experimental parameters while the others
held constant at the central levels. The response surface plots for
aflatoxin B1 (in Fig. 1) were built by the experimental data and the
generated equation. Fig. 1A–C depict the interactions of FA concen-
tration (%) versus other three factors. The recovery reaches to the
maximum when the term of FA content is about at −␣ level, i.e.,
no addition of acid is preferable for the extraction of aflatoxin B1 .
Response surface plots for NaCl addition versus other three factors
are shown in Fig. 1A, D, and E. According to the plots, the recov-
ery reaches to the maximum when the NaCl addition is about at +1
level. In the same way, higher levels (approximately + level) of NaAc
and PSA are proven to help improve the extraction efficiency, based
on the corresponding plots (Fig. 1B–F). These phenomena may be
related to the fact that the increased ionic strength and clean-up
effect, which reduce the co-extractive interferences and enhance
the transference of aflatoxin B1 to the extractant . Subsequently,
each analyte was matched to corresponding model and subject to
ANOVA with the same weight and importance. As a compromise,
the optimal extraction conditions for the analytes were determined
by calculation of numerical solutions: 3.35% FA in ACN extractant,
and 1.2 g of NaCl, 0.5 g of NaAc and 140 mg of PSA for QuEChERS
purification procedure.
3.3. Optimization of UPLC-MS/MS conditions
To obtain favorable resolution and high sensitivity, primary
optimization step focused on the selection of mobile phase and
into mass spectrometer under the full scan mode. The results
showed that all of the analytes could generate protonated adducts
[M+H]+ or the sodium adducts [M+Na]+ under ESI+ mode, whereas
the de-protonated molecules [M−H]− were obtained under ESI-
mode for some analytes. For most of veterinary drugs and mycotox-
ins, the responses of [M+H]+ ions were much more intense than the
[M−H]− ions. And protonated adducts exhibited better ionic stabil-
ity and fragment reproducibility than the [M+Na]+ ions, although
the responses of the latter ones were significantly higher.
For chromatographic separation, several frequently-used ana-
lytical columns including HSS T3 (2.1 × 150 mm, 1.7 ␮m), BEH C18
(2.1 × 150 mm, 1.7 ␮m) and CORTECS C18 were evaluated. Prefer-
able peak profiles and lower instrumental pressure were obtained
using the CORTECS column, which mainly due to the higher column
efficiency (particle size 1.6 ␮m) and the larger inner diameter. As
for the choice of the elution mobile phase, MeOH and ACN were
considered as candidates and initially tested in terms of separa-
G Model
CHROMA-359034; No. of Pages 10 ARTICLE IN PRESS
J. Zhou et al. / J. Chromatogr. A xxx (2017) xxx–xxx 5

Table 1
Experimental factors, coded levels and screening runs of Plackett-Burman design.

No. Factors Coded levels

Level -1 Level +1

1 Volume of water (mL) 5.0 7.5

2 Percentage of formic acid (%, v/v) 1.0 3.0
3 Ultrasonic extraction time (min) 20 30
4 Amount of anhydrous Na2 SO4 (g) 4.0 6.0
5 Amount of NaCl (g) 1.0 2.0
6 Amount of anhydrous NaAc (g) 0.0 0.5
7 Amount of C18 sorbent (mg) 200 300
8 Amount of PSA sorbent (mg) 0 100

Run X1 X2 X3 X4 X5 X6 X7 X8

1 7.5 1.0 30 6.0 1.0 0.0 200 0

2 7.5 3.0 20 6.0 2.0 0.0 200 0
3 5.0 3.0 30 4.0 2.0 0.5 200 0
4 5.0 1.0 30 6.0 1.0 0.5 300 0
5 7.5 1.0 20 6.0 2.0 0.0 300 100
6 7.5 3.0 20 4.0 2.0 0.5 200 100
7 7.5 3.0 30 4.0 1.0 0.5 300 0
8 7.5 3.0 30 6.0 1.0 0.0 300 100
9 5.0 3.0 30 6.0 2.0 0.0 200 100
10 7.5 1.0 30 6.0 2.0 0.5 200 0
11 5.0 3.0 20 6.0 2.0 0.5 300 0
12 7.5 1.0 30 4.0 2.0 0.5 300 100
13 5.0 3.0 20 6.0 1.0 0.5 300 100
14 5.0 1.0 30 4.0 2.0 0.0 300 100
15 5.0 1.0 20 6.0 1.0 0.5 200 100
16 5.0 1.0 20 4.0 2.0 0.0 300 0
17 7.5 1.0 20 4.0 1.0 0.5 200 100
18 7.5 3.0 20 4.0 1.0 0.0 300 0
19 5.0 3.0 30 4.0 1.0 0.0 200 100
20 5.0 1.0 20 4.0 1.0 0.0 200 0

tion effect, system pressure and analysis time. The results showed
that compared with MeOH, ACN provided sharper peak profiles and
facilitated the elution of analytes. Subsequently, different additives
in the aqueous phase, e.g. FA, ammonia and ammonium acetate
were assessed. As known, the addition of FA in aqueous solution
could promote the formation of protonated ions, thereby improv-
ing the detection sensitivity. This was especially for fumonisins and
some drugs, which could hardly ionize under the neutral or alkaline
conditions. Moreover, the addition of ammonium acetate, was eval-
uated. However, no significant improvement of sensitivities was
found. On the contrary, the ionization efficiencies for most of the
analytes were suppressed. Finally, the chromatographic separation
was conducted on CORTECS C18 column with ACN and FA mobile
phase system.
3.4. Validation of the proposed method
3.4.1. Linearity and MEs
Because of ionization interrupting in ESI source by co-elution
matrix components, ME was chose to evaluate the effect to target
analyte. The MEs were calculated by comparing the slope of matrix-
matched curve versus the slope of pure standard curve. The values
for absolute MEs were generally considered tolerable within ±20%
[28]. Moreover, different individual samples, feeding patterns and
sampling periods might cause the variations of endogenous com-
ponents in matrix extract, which could also affect the accuracy of
results. In this case, the relative ME of each analyte was estimated
in five different blank samples (eggs and milk). Sequentially, the
average value of the five measurements was used as final results.
According to the previous report by Viswanathan et al., the diver-
sities of matrix composition were considered ignorable with RSDs
of the absolute MEs lower than 15% [29].
As demonstrated in Fig. 2 (specific data in Table 2), the relative
MEs of analytes were acceptable with the highest RSD value of 12%
(sulfanilamide in eggs) in all case. Totally, 8 mycotoxins, 18 QNLs, 11
SAs and 10 other drugs were subject to strong suppression effects
in egg matrices whilst 16 mycotoxins, 17 QNLs, 15 SAs and 7 other
drugs in milk matrices. Significant MEs for most QNLs and SAs in
both two matrices were observed and the effect ratios were almost
the same. It could be explained by the coexistence of strongly polar
interferences during the same elution period (especially for the
eluted QNLs in the first five minutes). But interestingly, there were
obvious differences between the MEs for mycotoxins in eggs and
milk. Significant MEs were observed in milk matrices, which were
signal suppression for fusarenone X, neosolaniol, sterigmatocystin,
aflatoxins, Alternariol toxins, deoxynivalenol and its derivatives,
and signal enhancement for fumonisin B1 and ochratoxin ␣. On
the other hand, penicilic acid, fusarenone X, neosolaniol, deoxyni-
valenol and its derivatives were suppressed significantly in eggs.
Furthermore, the matrix interferences in eggs (−53%) and milk
(+51%) affected the ionization efficiency of verruculogen contrarily.
These experimental phenomena might be related to the difference
in sample composition. The matrix-matched curves were utilized
to compensate the MEs by spiking blank sample extracts with
appropriate volumes of the composite working solutions (prior to
filtration). Linearity was then evaluated and as a result, satisfac-
tory linear relationships within the concentration range were found
with R2 higher than 0.987 in all case (Table 2).
3.4.2. Accuracy and precision
The methodological accuracy was assessed through the three-
level spiking samples while the precision was studied in terms
of repeatability and inter-day (three consecutive days) precision.
As results shown in Fig. 3, 95% of the recovery values were
within 54%–109%, which ranged from 62% to 105% for mycotox-
ins (except for deoxynivalenol-3-glucuronide from 20% to 26%),
from 63% to 109% for QNLs, 46% to 104% for SAs and from 30% to
106% for other drugs. With respect to precision, the repeatability

Please cite this article in press as: J. Zhou, et al., Optimization for QuEChERS extraction of mycotoxins and veterinary drugs by response
surface methodology for application to egg and milk, J. Chromatogr. A (2017), https://doi.org/10.1016/j.chroma.2017.11.050
G Model
CHROMA-359034; No. of Pages 10 ARTICLE IN PRESS
6 J. Zhou et al. / J. Chromatogr. A xxx (2017) xxx–xxx

Table 2
Overview of the methodological characteristics including LOQ, ME for each analyte, and recoveries for three spiking levels.

Family Analyte No. Analytes Egg matrix Milk matrix

a,b
LOQ (␮g/kg) ME (%) Recoveries (%) for LOQ (␮g/kg) ME (%)a,b Recoveries (%) for
low, medium and low, medium and
high levelb high levelb

Mycotoxins 1 15-acetyl-deoxynivalenol 31.0 −14 (3) 101 (7) / 98 (8) / 88 (7) 10.0 −7 (4) 102 (4) / 98 (7) / 95 (3)
2 3-acetyl-deoxynivalenol 5.0 −10 (1) 91 (4) / 101 (3) / 102 (5) 2.0 −17 (4) 103 (5) / 99 (5) / 94 (3)
3 aflatoxin B1 0.02 −13 (5) 82 (10) / 86 (11) / 80 (10) 0.01 −35 (3) 80 (5) / 81 (7) / 80 (5)
4 aflatoxin B2 0.02 16 (5) 100 (7) / 100 (9) / 87 (5) 0.01 −28 (4) 102 (8) / 95 (5) / 88 (3)
5 aflatoxin G1 0.02 −4 (5) 83 (8) / 81 (9) / 75 (10) 0.01 −21 (5) 89 (6) / 84 (6) / 88 (4)
6 aflatoxin G2 0.02 −5 (4) 102 (5) / 100 (9) / 84 (11) 0.02 −47 (6) 105 (5) / 101 (8) / 93 (5)
7 aflatoxin M1 0.02 2 (4) 95 (6) / 93 (5) / 89 (5) 0.01 −18 (3) 94 (7) / 93 (7) / 90 (4)
8 aflatoxin M2 0.05 −16 (2) 99 (5) / 98 (6) / 89 (4) 0.02 −26 (4) 96 (6) / 97 (7) / 90 (5)
9 alternariol 6.0 8 (7) 97 (11) / 93 (11) / 91 (10) 4.0 −21 (6) 102 (10) / 98 (8) / 90 (3)
10 alternariol methyl ether 0.2 1 (5) 89 (4) / 87 (5) / 81 (3) 0.2 −45 (4) 92 (9) / 94 (9) / 97 (5)
11 chaetoglobosin 1.0 18 (4) 91 (11) / 90 (9) / 91 (11) 0.6 −3 (6) 77 (4) / 81 (7) / 79 (4)
12 cyclopiazonic acid 0.02 4 (3) 89 (6) / 82 (6) / 81 (9) 0.01 −5 (5) 98 (9) / 95 (5) / 90 (3)
13 de-epoxydeoxynivalenol 1.0 −28 (4) 84 (4) / 84 (4) / 80 (8) 0.4 −49 (4) 86 (5) / 87 (6) / 87 (5)
14 deoxynivalenol 5.0 −30 (4) 77 (6) / 82 (5) / 86 (5) 5.0 −49 (5) 93 (6) / 90 (8) / 88 (6)
15 deoxynivalenol-3-glucoside 10.0 −53 (4) 23 (8) / 23 (7) / 23 (12) 10.0 −41 (4) 23 (12) / 24 (14) / 23 (6)
16 diacetoxyscirpenol 0.2 −14 (4) 98 (5) / 100 (7) / 94 (6) 0.1 −31 (3) 96 (7) / 93 (5) / 91 (4)
17 fumonisin B1 1.0 −5 (3) 91 (4) / 86 (4) / 91 (5) 0.5 34 (2) 101 (3) / 98 (4) / 97 (4)
18 fumonisin B2 0.5 −7 (1) 86 (4) / 84 (6) / 91 (4) 0.5 9 (3) 91 (8) / 92 (8) / 97 (5)
19 fumonisin B3 0.4 −8 (6) 84 (4) / 85 (4) / 79 (4) 0.2 −13 (3) 93 (5) / 93 (6) / 90 (3)
20 fusarenone X 4.0 −24 (5) 94 (6) / 92 (4) / 87 (6) 2.0 −39 (4) 96 (6) / 93 (6) / 87 (4)
21 gliotoxin 0.2 −8 (2) 100 (7) / 95 (11) / 80 (9) 0.2 −27 (7) 93 (7) / 89 (9) / 89 (4)
22 HT-2 0.5 −2 (5) 103 (5) / 103 (6) / 96 (6) 0.2 −10 (5) 104 (4) / 102 (5) / 97 (4)
23 hydrolyzed fumonisin B1 0.2 −11 (3) 82 (3) / 84 (4) / 80 (6) 0.1 2 (2) 92 (4) / 89 (6) / 87 (3)
24 hydrolyzed fumonisin B2 0.6 −20 (6) 85 (10) / 83 (8) / 78 (11) 0.2 −8 (7) 92 (10) / 94 (9) / 91 (3)
25 neosolaniol 1.0 −65 (6) 78 (5) / 93 (7) / 97 (7) 0.3 −54 (7) 104 (6) / 100 (8) / 88 (3)
26 ochratoxin A 0.01 2 (7) 95 (9) / 96 (10) / 93 (5) 0.01 −8 (3) 85 (8) / 84 (19) / 84 (5)
27 ochratoxin B 0.01 0 (5) 100 (8) / 101 (9) / 100 (7) 0.01 −10 (3) 91 (4) / 91 (6) / 91 (4)
28 ochratoxin-alpha 10.0 34 (7) 91 (9) / 89 (7) / 87 (8) 2.5 28 (6) 88 (6) / 90 (10) / 86 (4)
29 penicilic acid 1.5 −36 (6) 84 (7) / 95 (6) / 88 (3) 0.6 −34 (2) 90 (8) / 95 (8) / 89 (4)
30 sterigmatocystin 0.5 −16 (4) 86 (2) / 85 (1) / 91 (1) 0.05 −25 (5) 91 (2) / 94 (1) / 91 (2)
31 T-2 0.05 −3 (5) 99 (10) / 104 (13) / 93 (13) 0.05 0 (4) 99 (7) / 102 (6) / 94 (5)
32 tentoxin 0.06 −4 (4) 104 (6) / 102 (7) / 94 (5) 0.02 −6 (2) 97 (4) / 98 (6) / 98 (4)
33 verruculogen 2.0 −53 (4) 66 (10) / 62 (8) / 62 (8) 0.5 50 (3) 84 (8) / 89 (7) / 90 (8)
34 zearalanone 0.2 3 (7) 94 (8) / 95 (7) / 88 (5) 0.1 −15 (2) 90 (5) / 92 (6) / 87 (4)
35 zearalenone 0.3 0 (5) 102 (9) / 99 (7) / 90 (5) 0.2 −13 (4) 87 (7) / 89 (6) / 88 (4)
36 ␣-zearalanol 0.8 −10 (9) 95 (9) / 94 (8) / 89 (6) 0.5 −14 (4) 93 (5) / 91 (4) / 89 (3)
37 ␣-zearalenol 0.1 2 (5) 98 (8) / 100 (6) / 90 (5) 0.05 −10 (1) 103 (7) / 99 (4) / 96 (3)
38 ␤-zearalanol 0.7 −10 (8) 92 (7) / 93 (6) / 88 (6) 0.4 −9 (3) 95 (4) / 94 (2) / 88 (3)
39 ␤-zearalenol 0.1 −27 (7) 100 (10) / 91 (10) / 88 (5) 0.06 2 (4) 101 (7) / 100 (7) / 94 (3)

QNLs 40 cinoxacin 0.2 35 (5) 81 (5) / 77 (6) / 76 (6) 0.1 28 (3) 101 (6) / 100 (6) / 94 (5)
41 ciprofloxacin 0.2 −53 (7) 93 (3) / 82 (6) / 82 (5) 0.2 −50 (1) 89 (4) / 85 (4) / 80 (4)
42 danofloxacin 0.1 −32 (4) 103 (5) / 94 (9) / 98 (5) 0.1 −30 (3) 88 (9) / 88 (6) / 81 (5)
43 difloxacin 0.05 −30 (2) 80 (4) / 84 (6) / 82 (6) 0.05 −29 (2) 100 (7) / 100 (4) / 92 (5)
44 enoxacin 0.5 −56 (5) 74 (5) / 87 (6) / 89 (8) 0.4 −52 (3) 84 (4) / 83 (5) / 85 (4)
45 enrofloxacin 0.1 −57 (6) 97 (10) / 89 (9) / 92 (7) 0.05 −33 (2) 102 (4) / 97 (3) / 88 (4)
46 fleroxacin 0.1 −34 (5) 109 (3) / 92 (4) / 87 (5) 0.1 −38 (2) 98 (4) / 95 (4) / 90 (5)
47 flumequine 0.2 −2 (5) 83 (9) / 95 (7) / 95 (7) 0.1 5 (1) 102 (7) / 99 (3) / 92 (5)
48 gatifloxacin 1.0 −35 (3) 102 (6) / 93 (5) / 91 (4) 0.5 −33 (4) 100 (7) / 99 (6) / 93 (5)
49 lomefloxacin 0.1 −38 (4) 101 (3) / 88 (7) / 88 (4) 0.05 −52 (2) 95 (6) / 94 (4) / 82 (5)
50 marbofloxacin 0.2 −42 (3) 101 (6) / 91 (6) / 91 (4) 0.2 −59 (1) 88 (4) / 91 (9) / 92 (9)
51 moxifloxacin 0.2 −33 (7) 94 (5) / 85 (5) / 91 (5) 0.1 −13 (2) 98 (5) / 98 (4) / 92 (5)
52 nadifloxacin 0.1 −24 (5) 102 (7) / 94 (7) / 96 (4) 0.1 53 (10) 97 (6) / 95 (5) / 87 (5)
53 nalidixic acid 0.1 −5 (4) 97 (8) / 92 (9) / 93 (7) 0.1 −13 (4) 96 (5) / 92 (2) / 92 (5)
54 norfloxacin 0.5 −54 (5) 86 (7) / 77 (5) / 96 (6) 0.2 −52 (2) 84 (6) / 89 (6) / 88 (5)
55 ofloxacin (levofloxacin) 0.05 −41 (3) 98 (5) / 87 (8) / 93 (3) 0.05 −32 (6) 97 (4) / 92 (4) / 95 (6)
56 orbifloxacin 0.1 −31 (3) 105 (3) / 94 (5) / 87 (4) 0.1 −41 (4) 100 (3) / 96 (5) / 89 (4)
57 oxolinic acid 0.1 10 (2) 78 (2) / 98 (8) / 96 (3) 0.1 9 (2) 100 (9) / 99 (4) / 91 (4)
58 pazufloxacin 0.2 −47 (3) 90 (6) / 88 (8) / 97 (5) 0.1 −45 (3) 85 (8) / 85 (6) / 85 (5)
59 pefloxacin 0.2 −38 (5) 88 (7) / 101 (7) / 91 (4) 0.1 −39 (5) 98 (3) / 92 (4) / 87 (5)
60 pipemidic acid 0.2 −56 (5) 77 (5) / 63 (5) / 64 (3) 0.1 −57 (1) 71 (4) / 71 (6) / 73 (5)
61 sarafloxacin 0.1 −42 (4) 104 (5) / 91 (8) / 90 (3) 0.1 −38 (2) 103 (5) / 98 (3) / 87 (4)
62 sparfloxacin 0.1 −20 (3) 102 (3) / 93 (7) / 88 (5) 0.1 −25 (2) 100 (3) / 97 (4) / 89 (4)

SAs 63 sulfabenzamide 0.05 −24 (6) 57 (8) / 59 (9) / 67 (6) 0.05 −62 (1) 76 (4) / 76 (9) / 73 (4)
64 sulfacetamide 1.0 −56 (9) 47 (9) / 46 (9) / 48 (10) 1.0 −61 (3) 72 (5) / 76 (7) / 73 (2)
65 sulfachlorpydazine 0.1 −15 (7) 63 (5) / 69 (4) / 65 (6) 0.05 −17 (1) 73 (5) / 76 (6) / 69 (4)
66 sulfadimethoxine 0.05 −1 (5) 68 (9) / 62 (7) / 64 (7) 0.05 −22 (2) 81 (4) / 81 (5) / 75 (7)
67 sulfadoxine 0.05 −6 (2) 78 (7) / 88 (6) / 95 (5) 0.05 −10 (3) 89 (3) / 81 (5) / 72 (5)
68 sulfaguanidine 1.0 −58 (9) 53 (9) / 47 (10) / 47 (8) 3.0 −59 (9) 54 (6) / 52 (8) / 58 (6)
69 sulfamerazine 0.1 −42 (2) 69 (5) / 72 (9) / 76 (8) 0.1 −53 (2) 76 (3) / 75 (6) / 68 (4)
70 sulfamethazine 0.05 −19 (5) 73 (4) / 71 (12) / 73 (5) 0.05 −22 (1) 80 (3) / 75 (6) / 71 (6)

Please cite this article in press as: J. Zhou, et al., Optimization for QuEChERS extraction of mycotoxins and veterinary drugs by response
surface methodology for application to egg and milk, J. Chromatogr. A (2017), https://doi.org/10.1016/j.chroma.2017.11.050
G Model
CHROMA-359034; No. of Pages 10 ARTICLE IN PRESS
J. Zhou et al. / J. Chromatogr. A xxx (2017) xxx–xxx 7

Table 2 (Continued)

Family Analyte No. Analytes Egg matrix Milk matrix

a,b
LOQ (␮g/kg) ME (%) Recoveries (%) for LOQ (␮g/kg) ME (%)a,b Recoveries (%) for
low, medium and low, medium and
high levelb high levelb

71 sulfamethizole 0.1 −35 (3) 65 (6) / 61 (6) / 62 (6) 0.1 −38 (3) 72 (3) / 72 (7) / 71 (4)
72 sulfamethoxazole 0.2 6 (6) 65 (6) / 63 (7) / 61 (7) 0.1 −2 (2) 79 (7) / 79 (9) / 70 (5)
73 sulfamethoxypyridazine 0.1 −23 (4) 82 (5) / 81 (5) / 82 (6) 0.05 −28 (1) 78 (3) / 76 (6) / 71 (2)
74 sulfamonomethoxine 0.05 −20 (6) 69 (6) / 73 (6) / 75 (5) 0.05 −21 (1) 72 (4) / 73 (7) / 68 (5)
75 sulfamoxole 0.1 −37 (2) 66 (3) / 74 (5) / 73 (5) 0.1 −47 (2) 72 (3) / 71 (4) / 70 (4)
76 sulfanilamide 1.0 −66 (12) 57 (11) / 54 (12) / 51 (11) 1.0 −66 (5) 68 (8) / 67 (11) / 55 (7)
77 sulfanitran 1.0 −3 (8) 104 (6) / 100 (6) / 102 (5) 0.5 −12 (3) 98 (7) / 98 (6) / 92 (4)
78 sulfaphenazole 0.1 1 (3) 65 (7) / 62 (7) / 58 (7) 0.05 −28 (2) 75 (4) / 74 (5) / 75 (5)
79 sulfapyridine 0.05 −46 (7) 82 (6) / 96 (8) / 92 (6) 0.05 −43 (2) 73 (3) / 71 (6) / 65 (5)
80 sulfathiazole 0.3 −59 (4) 61 (7) / 69 (7) / 65 (6) 0.4 −59 (2) 66 (4) / 67 (9) / 62 (4)
81 sulfisomidine 0.1 −62 (4) 69 (8) / 67 (8) / 66 (7) 0.05 −43 (2) 65 (3) / 66 (4) / 57 (7)
82 sulfisoxazole 0.2 −17 (3) 70 (7) / 77 (9) / 71 (10) 0.2 −3 (5) 74 (12) / 78 (9) / 74 (5)

Others 83 amantadine 0.1 −17 (6) 86 (11) / 81 (12) / 85 (10) 0.05 −2 (6) 88 (5) / 90 (4) / 106 (6)
84 ampicillin 1.0 −47 (4) 61 (7) / 63 (6) / 64 (9) 0.5 −55 (2) 44 (12) / 37 (10) / 33 (9)
85 azithromycin 0.1 −23 (3) 97 (4) / 97 (5) / 97 (5) 0.05 32 (3) 98 (4) / 98 (4) / 101 (3)
86 azlocillin 0.1 −13 (10) 34 (13) / 38 (9) / 38 (13) 0.1 −9 (8) 36 (12) / 37 (12) / 37 (12)
87 ceftiofur 2.0 −26 (8) 93 (8) / 94 (9) / 94 (7) 1.5 −10 (6) 101 (6) / 104 (4) / 96 (4)
88 cephapirin 1.0 −48 (6) 68 (10) / 62 (5) / 62 (3) 1.0 −49 (2) 65 (5) / 62 (3) / 65 (4)
89 chloramphenicol 0.09 −21 (3) 101 (7) / 96 (7) / 91 (11) 0.3 −38 (5) 94 (8) / 95 (8) / 95 (6)
90 clarithromycin 1.0 2 (6) 84 (5) / 84 (4) / 86 (3) 0.2 1 (3) 84 (5) / 84 (5) / 85 (4)
91 clindamycin 0.06 −23 (4) 79 (4) / 72 (6) / 72 (5) 0.05 −21 (3) 83 (5) / 83 (8) / 78 (4)
92 cloxacillin 0.3 −3 (4) 56 (8) / 57 (6) / 56 (7) 0.3 10 (7) 56 (9) / 56 (7) / 56 (9)
93 josamycin 0.4 −11 (12) 86 (5) / 85 (6) / 84 (5) 0.2 27 (2) 86 (5) / 84 (5) / 88 (3)
94 kitasamycin 0.2 3 (7) 69 (12) / 74 (10) / 72 (8) 0.2 10 (4) 71 (9) / 70 (12) / 70 (9)
95 lincomycin 0.2 −50 (6) 42 (9) / 42 (11) / 41 (7) 0.05 −49 (4) 40 (10) / 39 (11) / 35 (7)
96 methicillin 0.2 −3 (5) 79 (5) / 78 (4) / 77 (5) 0.2 12 (2) 79 (6) / 79 (8) / 78 (4)
97 metronidazole 0.1 −25 (7) 86 (9) / 75 (8) / 75 (7) 0.06 −40 (3) 96 (7) / 94 (5) / 84 (4)
98 oleandomycin 0.1 1 (5) 99 (6) / 100 (6) / 94 (6) 0.2 0 (2) 97 (4) / 93 (5) / 91 (3)
99 oxacillin 0.2 −27 (1) 60 (9) / 58 (12) / 59 (8) 0.3 10 (4) 60 (8) / 60 (9) / 63 (9)
100 penicillin G 2.5 −21 (8) 71 (9) / 65 (11) / 71 (9) 2.0 −26 (3) 62 (10) / 67 (8) / 62 (7)
101 penicillin V 1.0 3 (2) 47 (15) / 47 (13) / 42 (11) 1.5 4 (6) 36 (12) / 32 (9) / 30 (9)
102 roxithromycin 0.2 23 (6) 98 (5) / 97 (4) / 98 (6) 0.2 −4 (3) 98 (5) / 99 (4) / 97 (5)
103 tilmicosin 0.1 −15 (2) 103 (5) / 98 (6) / 98 (7) 0.1 18 (7) 102 (12) / 102 (7) / 92 (5)
104 tylosin 0.2 6 (6) 85 (8) / 83 (7) / 83 (7) 0.2 8 (1) 87 (6) / 86 (7) / 86 (7)
a
Average value of ME results of five blank samples.
b
RSDs are given in the brackets.

and inter-day RSDs were equal or lower than 14% for all tested
assays, indicating the favorable stability of the proposed method.
The obtained LOQs of the investigated 104 contaminants in eggs
ranged from 0.01 ␮g/kg to 31 ␮g/kg (0.05 ␮g/kg to 2.5 ␮g/kg for
drugs) and from 0.01 ␮g/kg to 10 ␮g/kg (0.05 ␮g/kg to 3.0 ␮g/kg
for drugs) in milk (Table 2). Hereinto, for the mycotoxins that mon-
itored by most countries, e.g. aflatoxins, ochratoxin A, zearalenone,
deoxynivalenol and fumonisins, all the obtained LOQs met the strict
tolerance levels setting for mycotoxins in foodstuffs by EU [30].
Meanwhile, no interference peaks were observed in tested blank
samples, implying the favorable selectivity of the method.
3.4.3. Application of the validated method
In practice, 45 eggs and 30 milk samples were collected and
subjected to the established method. All identification criteria con-
sisting of retention time, peak shape and precursor-to-product ion
ratio on UPLC–MS/MS were used to check the suspect samples. In
this case, the determination values below LOD were equaled to 0
and the values between LOD and LOQ were equaled to LOD. The
obtained results in Table 3 showed that 9 eggs and 6 milk sam-
ples were free of all contaminants. 36 eggs (80.0% of incidence) and
20 milk (66.6%) samples were contaminated with 8 QNLs and 2
SAs, ranging in levels up to 32.8 ␮g/kg (ofloxacin) and 134.0 ␮g/kg
(sulfamethoxazole), respectively. Among the QNLs, danofloxacin,
ofloxacin and enrofloxacin were the most frequently detectable
items in eggs: the mean level (incidence, range of measure-
ment) were 1.4 ␮g/kg (31%, 0.4 ␮g/kg–5.2 ␮g/kg), 1.5 ␮g/kg (29%,
0.1 ␮g/kg–32.8 ␮g/kg) and 0.7 ␮g/kg (18%, 0.2 ␮g/kg–1.7 ␮g/kg),
respectively. On the other hand, danofloxacin, ofloxacin and
enoxacin were the most commonly reported in milk: 0.5 ␮g/kg
(43%, 0.1 ␮g/kg–2.7 ␮g/kg), 0.4 ␮g/kg (30%, 0.05 ␮g/kg–1.1 ␮g/kg)
and 0.6 ␮g/kg (10%, 0.5 ␮g/kg–0.8 ␮g/kg). The extensive contam-
ination of quinolones, usually in low levels, might be related to
the use of drugs as feed additives. Moreover, two eggs were pos-
itive for amantadine and metronidazole at the concentration of
1.0 ␮g/kg and 7.5 ␮g/kg respectively, which must be absent in the
food products of animal origin.
With respect to mycotoxins, extensive occurrences in these
matrices, typically zearalenone, deoxynivalenol and aflatoxins,
were reported according to previous publications [11,18,31–35]. In
fact, no trace level (<LOQs) of mycotoxins mentioned above were
detected in this case, except for aflatoxin M1 in milk (< 0.05 ␮g/kg).
These results were consistent with the findings report by Sypecka
et al. [36]. Interestingly, an unexpected mycotoxin, i.e. sterigmato-
cystin, was found in 10 egg samples with maximum concentration
exceeding 3000 ␮g/kg. To the best of our knowledge, this work is
the first report on the detection of sterigmatocystin in eggs with
remarkably high concentration. The experimental phenomena and
abnormal data may be attributed to different living conditions and
breeding patterns of the hens.
To ensure the accuracy of results, the calibration curve
of sterigmatocystin was newly constructed with wider ranges
(0.25–2000 ng/mL). Meanwhile, the national food safety standards
of China, namely GB 5009.25-2016 Determination of sterigma-

Please cite this article in press as: J. Zhou, et al., Optimization for QuEChERS extraction of mycotoxins and veterinary drugs by response
surface methodology for application to egg and milk, J. Chromatogr. A (2017), https://doi.org/10.1016/j.chroma.2017.11.050
G Model
CHROMA-359034; No. of Pages 10 ARTICLE IN PRESS
8 J. Zhou et al. / J. Chromatogr. A xxx (2017) xxx–xxx

Fig. 1. The estimated response surface plots of aflatoxin B1 recovery versus: (A) percentage of FA-amount of NaCl; (B) percentage of FA-amount of NaAc; (C) percentage of
FA-amount of PSA; (D) amount of NaCl-amount of NaAc; (E) amount of NaCl-amount of PSA; (F) amount of NaAc-amount of PSA.

Fig. 2. MEs of analytes in two investigated matrices: (A) egg and (B) milk.

tocystin in foodstuffs, which based on immunoaffinity column 4. Conclusions

methodology, was employed for confirmation analysis. As a result,
the quantifiable results of two different approaches that exceed the
LOQ of sterigmatocystin (0.5 ␮g/kg) were summarized and com-
pared (Supplementary Table 6). Since the limited data (n = 10) did
not accord with normal distribution, two-independent sample non-
parametric test (Kolmogorov-Smirnov test) was performed and no
significant diversity was found in the overall distribution between
two sets of results (P = 0.436 > 0.05).
In this study, a simple and universal multiclass method for the
simultaneous analysis of 104 contaminants and residues in eggs
and milk, i.e. veterinary drugs, mycotoxins and its metabolites,
was proposed. Compared with the previous methods for multi-
class analysis, the established method was rapid and efficient, and
capable of covering more detection items (especially mycotoxins
metabolites) that were scarcely monitored in animal foodstuffs.

Please cite this article in press as: J. Zhou, et al., Optimization for QuEChERS extraction of mycotoxins and veterinary drugs by response
surface methodology for application to egg and milk, J. Chromatogr. A (2017), https://doi.org/10.1016/j.chroma.2017.11.050
G Model
CHROMA-359034; No. of Pages 10 ARTICLE IN PRESS
J. Zhou et al. / J. Chromatogr. A xxx (2017) xxx–xxx 9

Fig. 3. Scatter diagrams of recoveries: (A) egg-low level; (B) egg-medium level: (C) egg-high level; (D) milk-low level; (E) milk-medium level; (F) milk-high level.

Table 3
Mycotoxins and veterinary drugs occurrence in the two investigated matrices.

Matrix Family Analyte detected Number of positive/analyzed samples Median (range) (␮g/kg)

eggs mycotoxins sterigmatocystin 10/45 2.9 (0.5–3608)

QNLs norfloxacin 1/45 2.4
ciprofloxacin 6/45 6.8 (2.4–13.3)
pefloxacin 3/45 3.6 (0.3–12.8)
danofloxacin 14/45 0.6 (0.4–5.2)
enrofloxacin 8/45 0.4 (0.2–1.7)
ofloxacin (levofloxacin) 14/45 0.1 (0.1–32.8)
orbifloxacin 13/45 0.08 (0.08–3.9)
SAs sulfamonomethoxine 3/45 1.0 (0.7–1.1)
sulfamethoxazole 1/45 134.0
others amantadine 1/45 1.0
metronidazole 1/45 7.5

milk mycotoxins aflatoxin M1 11/45 0.02 (0.002-0.03)

QNLs enoxacin 3/45 0.6 (0.5–0.8)
danofloxacin 13/45 0.2 (0.1–2.7)
ofloxacin (levofloxacin) 9/45 0.1 (0.01–1.1)

The chemometric approaches, i.e. Plackett-Burman design and CCD,
were employed and particular attention had been paid to the prepa-
ration procedure. The mathematical model and three-dimensional
plots showed detailed effects of factors. For validation, the estab-
lished method had yielded satisfactory analytical characteristics
in terms of linearity, accuracy, precision and sensitivity. In prac-
tice, the established method was successfully applied for 45 eggs
and 30 milk samples analysis, of which positive finding aflatoxin
in milk and sterigmatocystin in eggs. In the future, feasible sur-
vey of sterigmatocystin should be implemented to evaluate in this
matrix. And there are also existed some deficiencies and regrets,
for instance, inferior applicability to most ␤-lactam and macrolide,
and unsatisfactory extraction efficiency for strongly polar analytes.
Conflicts of interest
None.
Acknowledgments
This study was supported by the National Instrumentation Pro-
gram (No. 2011YQ060084) and the Medical and Health Science and
Technology Plan of Zhejiang Province (No. 2018266692-KY732).
Appendix A. Supplementary data
Supplementary material related to this article can be found, in
the online version, at doi:https://doi.org/10.1016/j.chroma.2017.
11.050
References
[1] A.L. Capriotti, C. Cavaliere, S. Piovesana, R. Samperi, A. Laganà, Multiclass
screening method based on solvent extraction and liquid
chromatography–tandem mass spectrometry for the determination of
antimicrobials and mycotoxins in egg, J. Chromatogr. A 1268 (2012) 84–90.
[2] P.C. Turner, B. Flannery, C. Isitt, M. Ali, J. Pestka, The role of biomarkers in
evaluating human health concerns from fungal contaminants in food, Nutr.
Res. Rev. 25 (2012) 162–179.
[3] E.N. Ediage, J.D. Di Mavungu, S. Song, A. Wu, C. Van Peteghem, S.De Saeger, A
direct assessment of mycotoxin biomarkers in human urine samples by liquid
chromatography tandem mass spectrometry, Anal. Chim. Acta 741 (2012)
58–69.
[4] G. Pagliuca, E. Zironi, A. Ceccolini, R. Matera, G.P. Serrazanetti, A. Piva, Simple
method for the simultaneous isolation and determination of fumonisin B1
and its metabolite aminopentol-1 in swine liver by liquid
chromatography–fluorescence detection, J. Chromatogr. B 819 (2005) 97–103.
[5] Commission regulation (EU) no 37/2010 of 22 December 2009 on
pharmacologically active substances and their classification regarding
maximum residue limits in foodstuffs of animal origin, Off. J. Eur. Union L 15
(2010) 1–72.

Please cite this article in press as: J. Zhou, et al., Optimization for QuEChERS extraction of mycotoxins and veterinary drugs by response
surface methodology for application to egg and milk, J. Chromatogr. A (2017), https://doi.org/10.1016/j.chroma.2017.11.050
G Model
CHROMA-359034; No. of Pages 10 ARTICLE IN PRESS
10 J. Zhou et al. / J. Chromatogr. A xxx (2017) xxx–xxx
[6] EC, Commission regulation (EU) no 165/2010 of 26 February 2010 amending
regulation (EC) no 1881/2006 setting maximum levels for certain
contaminants in foodstuffs as regards aflatoxins, Off. J. Eur Union. L 50 (2010)
8.
[7] GB 2761—2017 Maximum Levels of Mycotoxins in Foods, China’s Ministry of
Health, 2017.
[8] 235 Bulletin of the Ministry of Agriculture of the People’s Republic of China,
China’s Ministry of Agriculture, 2002.
[9] H.G.J. Mol, P. Plaza-Bolaños, P. Zomer, T.C. de Rijk, A.A.M. Stolker, P.P.J.
Mulder, Toward a generic extraction method for simultaneous determination
of pesticides, mycotoxins, plant toxins, and veterinary drugs in feed and food
matrixes, Anal. Chem. 80 (2008) 9450–9459.
[10] E. De Dominicis, I. Commissati, M. Suman, Targeted screening of pesticides,
veterinary drugs and mycotoxins in bakery ingredients and food commodities
by liquid chromatography-high-resolution single-stage orbitrap mass
spectrometry, J. Mass Spectrom. 47 (2012) 1232–1241.
[11] A.L. Capriotti, C. Cavaliere, S. Piovesana, R. Samperi, A. Laganà, Multiclass
screening method based on solvent extraction and liquid
chromatography–tandem mass spectrometry for the determination of
antimicrobials and mycotoxins in egg, J. Chromatogr. A 1268 (2012) 84–90.
[12] J. Zhan, X.J. Yu, Y.Y. Zhong, Z.T. Zhang, X.M. Cui, J.F. Peng, Generic and rapid
determination of veterinary drug residues and other contaminants in raw
milk by ultra performance liquid chromatography–tandem mass
spectrometry, J. Chromatogr. B 906 (2012) 48–57.
[13] J. Zhan, Y.-Y. Zhong, X.-J. Yu, J.-F. Peng, S. Chen, J.-Y. Yin, et al., Multi-class
method for determination of veterinary drug residues and other contaminants
in infant formula by ultra performance liquid chromatography–tandem mass
spectrometry, Food Chem. 138 (2013) 827–834.
[14] J. Zhan, D.M. Xu, S.J. Wang, J. Sun, Y.J. Xu, M.L. Ni, Comprehensive screening
for multi-class veterinary drug residues and other contaminants in muscle
using column-switching UPLC–MS/MS, Food Addit.Contam. Part. A 30 (2013)
1888–1899.
[15] A.G. Frenich, R. Romero-González, Comprehensive analysis of toxics
(pesticides, veterinary drugs and mycotoxins) in food by UHPLC–MS, TrAC
Trends Anal. Chem. 6 (2014) 158–169.
[16] M. Anastassiades, S.J. Lehotay, D. Stajnbaher, F.J. Schenck, Fast and easy
multi-residue method employing acetonitrile extraction/partitioning and
“dispersive solid-phase extraction” for the determination of pesticide
residues in produce, J. AOAC Int. 86 (2003) 412–431.
[17] G. Stubbings, T. Bigwood, The development and validation of a multiclass
liquid chromatography tandem mass spectrometry (LC–MS/MS) procedure
for the determination of veterinary drug residues in animal tissue using a
QuEChERS (quick, easy, cheap, effective, rugged and safe) approach, Anal.
Chim. Acta. 637 (2009) 68–78.
[18] A.G. Frenich, R. Romero-González, M.L. Gómez-Pérez, J.L.M. Vidal,
Multi-mycotoxin analysis in eggs using a QuEChERS-based extraction
procedure and ultra-high-pressure liquid chromatography coupled to triple
quadrupole mass spectrometry, J. Chromatogr. A. 1218 (2011) 4349–4356.
[19] S. Moretti, G. Cruciani, S. Romanelli, R. Rossi, G. Saluti, R. Galarini, Multiclass
method for the determination of 62 antibiotics in milk, J. Mass. Spectr. 51
(2016) 792–804.
[20] A. Jouyban, M.H. Sorouraddin, M.A. Farajzadeh, M.H. Somi, R.
Fazeli-Bakhtiyari, Determination of five antiarrhythmic drugs in human
plasma by dispersive liquid-liquid microextraction and high-performance
liquid chromatography, Talanta. 134 (2015) 681–689.
[21] X.W. Lai, D.L. Sun, C.Q. Ruan, H. Zhang, C.L. Liu, Rapid analysis of aflatoxins B1,
B2, and ochratoxin A in rice samples using dispersive liquid-liquid
microextraction combined with HPLC, J. Sep. Sci. 37 (2014) 92–98.
[22] M. Asadollahzadeh, H. Tavakoli, M. Torab-Mostaedi, G. Hosseini, A. Hemmati,
Response surface methodology based on central composite design as a
chemometric tool for optimization of dispersive-solidification liquid–liquid
Please cite this article in press as: J. Zhou, et al., Optimization for QuEChERS extraction of mycotoxins and veterinary drugs by response
surface methodology for application to egg and milk, J. Chromatogr. A (2017), https://doi.org/10.1016/j.chroma.2017.11.050
microextraction for speciation of inorganic arsenic in environmental water
samples, Talanta 123 (2014) 25–31.
[23] J. Zhou, J. Xu, B. Huang, Z. Cai, Y. Ren, High-performance liquid
chromatographic determination of multi-mycotoxin in cereals and bean
foodstuffs using interference-removal solid-phase extraction combined with
optimized dispersive liquid–liquid microextraction, J. Sep. Sci. 40 (2017)
2141–2150.
[24] G. Pagliuca, E. Zironi, A. Ceccolini, R. Matera, G.P. Serrazanetti, A. Piva, Simple
method for the simultaneous isolation and determination of fumonisin B, J.
Chromatogr. B 819 (2005) 1–7.
[25] A.G. Frenich, M. del Mar Aguilera-Luiz, J.L.M. Vidal, R. Romero-González,
Comparison of several extraction techniques for multiclass analysis of
veterinary drugs in eggs using ultra-high pressure liquid
chromatography-tandem mass spectrometry, Anal. Chim. Acta 661 (2010)
150–160.
[26] M. Bashiry, A. Mohammadi, H. Hosseini, M. Kamankesh, S. Aeenehvand, Z.
Mohammadi, Application and optimization of microwave-assisted extraction
and dispersive liquid-liquid microextraction followed by high-performance
liquid chromatography for sensitive determination of polyamines in turkey
breast meat samples, Food Chem. 190 (2016) 1168–1173.
[27] E. Ranjbari, M.R. Hadjmohammadi, Optimization of magnetic stirring assisted
dispersive liquid–liquid microextraction of rhodamine B and rhodamine 6G
by response surface methodology application in water samples, soft drink,
and cosmetic products, Talanta 139 (2015) 216–225.
[28] A.G. Frenich, R. Romero-González, M.L. Gómez-Pérez, J.L.M. Vidal,
Multi-mycotoxin analysis in eggs using a QuEChERS-based extraction
procedure and ultra-high-pressure liquid chromatography coupled to triple
quadrupole mass spectrometry, J. Chromatogr. A 1218 (2011) 4349–4356.
[29] C.T. Viswanathan, S. Bansal, B. Booth, A.J. DeStefano, M.J. Rose, J. Sailstad,
et al., Workshop/conference report—quantitative bioanalytical methods
validation and implementation: best practices for chromatographic and
ligand binding assays, AAPS J. 9 (2007) E30–E42.
[30] EC, Commission regulation (EU) no 165/2010 of 26 February 2010 amending
regulation (EC) no 1881/2006 setting maximum levels for certain
contaminants in foodstuffs as regards aflatoxins, Off. J. Eur. Union L 50 (2010)
8.
[31] R. Zhu, Z. Zhao, J. Wang, B. Bai, A. Wu, L. Yan, et al., A simple sample
pretreatment method for multi-mycotoxin determination in eggs by liquid
chromatography tandem mass spectrometry, J. Chromatogr. A 1417 (2015)
1–7.
[32] A.L. Capriotti, C. Cavaliere, P. Foglia, R. Samperi, S. Stampachiacchiere, S.
Ventura, et al., Multiclass analysis of mycotoxins in biscuits by high
performance liquid chromatography–tandem mass spectrometry.
Comparison of different extraction procedures, J. Chromatogr. A 1343 (2014)
69–78.
[33] D. Chen, X. Cao, Y. Tao, Q. Wu, Y. Pan, L. Huang, et al., Development of a
sensitive and robust liquid chromatography coupled with tandem mass
spectrometry and a pressurized liquid extraction for the determination of
aflatoxins and ochratoxin A in animal derived foods, J. Chromatogr. A 1253
(2012) 110–119.
[34] E.K. Tangni, N. Waegeneers, I. Van Overmeire, L. Goeyens, L. Pussemier,
Mycotoxin analyses in some home produced eggs in Belgium reveal small
contribution to the total daily intake, Sci. Total Environ. 407 (2009)
4411–4418.
[35] M.E. Zain, Impact of mycotoxins on humans and animals, J. Saudi Chem. Soc.
15 (2011) 129–144.
[36] Z. Sypecka, M. Kelly, P. Brereton, Deoxynivalenol and zearalenone residues in
eggs of laying hens fed with a naturally contaminated diet: effects on egg
production and estimation of transmission rates from feed to eggs, J. Agric.
Food Chem. 52 (2004) 5463–5471.