You are on page 1of 12

Food Control 115 (2020) 107289

Contents lists available at ScienceDirect

Food Control
journal homepage: www.elsevier.com/locate/foodcont

Analysis of highly polar pesticides and their main metabolites in animal T


origin matrices by hydrophilic interaction liquid chromatography and mass
spectrometry
Sonia Herrera López∗, Jonatan Dias, André de Kok
WFSR – Wageningen Food Safety Research, National Reference Laboratory (NRL) for Pesticide Residues in Food and Feed, Akkermaalsbos 2, 6708, WB Wageningen, the
Netherlands

ARTICLE INFO ABSTRACT

Keywords: In this work, a method for the analysis of 14 highly polar pesticides in animal origin matrices by hydrophilic
Polar pesticides interaction liquid chromatography and mass spectrometry has been described. Three methanol-based extraction
Glyphosate procedures with slight modifications, depending on the matrix, have been used. The method has been validated
Isotopically labelled internal standards according to the performance criteria from the SANTE guideline for pesticide residues analysis in food and feed.
Animal origin matrices
Most of the compounds showed a limit of quantitation at the lowest spike levels tested (0.01–0.02 mg kg−1) in
Hydrophilic interaction liquid chromatography
the matrices evaluated. A SPE clean-up step with mixed mode cation exchange sorbent has been used with some
matrices, like chicken meat, eggs, liver and kidney, in order to improve the limits of quantification. Glufosinate,
glyphosate and HEPA were the compounds which presented more difficulties during the study, due to the lack of
sensitivity of the qualifier ions and/or because of the interference from the matrices.

1. Introduction their physical-chemical properties. They are mainly analysed by single


residue methods, special methods to analyse them individually or as a
Glyphosate is the active ingredient in the popular herbicide small group of compounds with similar properties (Rajski, Díaz
Roundup. Since its introduction in the market approximately 50 years Galiano, Cutillas, & Férnandez-Alba, 2018; Vass et al., 2016; Goscinny,
ago, glyphosate has become one of the world's most widely used her- Unterluggauer, Aldrian, Hanot, & Masselter, 2012), using LC methods
bicides due to its relatively low cost and high efficiency (Zhang, Rana, and a derivatization step (Guo, Riter, Wujcik, & Armstrong, 2016;
Shaffer, Taioli, & Sheppard, 2019). In the United States, the major in- Hanke, Singer, & Hollender, 2008). In a previous publication of our
crease occurred after the introduction of genetically modified glypho- group (Herrera López, Scholten, Kiedrowska, & de Kok, 2019), a
sate-resistant, “Roundup-ready” crops in 1996 (Thelin & Stone, 2013). method for the direct analysis of these pesticides and their metabolites
Before the use of “Roundup-ready” crops, the pesticide was applied (Dutch Polar Pesticides method, NL-PP) in fruit, vegetables and cereals
after harvest to kill all the plants and prepare the field for the new crop. was developed, using hydrophilic interaction chromatography (HILIC)
The application of glyphosate on the “Roundup-ready crops” shortly and mass spectrometry (MS). The method showed good robustness and
before harvest, so-called “green burndown,” to speed up their drying, it was successfully implemented in routine laboratories.
causes crops to have higher levels of glyphosate residues (Benbrook, For the majority of these polar pesticides, maximum residue levels
2016). Due to this “green burndown” procedure, the application rate (MRLs) have already been established with metabolites included in the
and residue concentrations of glyphosate have increased and resistant residue definition, while for glyphosate MRLs are under review, con-
weeds have emerged in Argentina and Brazil, but also in other parts of sidering inclusion of the metabolites (EFSA 2018; EFSA 2019). For
world. (Cerdeira et al., 2012; Vila-Aiub, Yu, & Powles, 2019). products of animal origin, the lack of information about the presence of
Glyphosate is the most known pesticide from the so-called “highly glyphosate and its metabolites causes the MRLs still to be considered as
polar pesticides”, due to the controversy about its toxicity in the last tentative. In the last report from EFSA (EFSA 2019), it has been men-
years. However, besides glyphosate, there are some other highly polar tioned that confirmatory methods for glyphosate, AMPA and N-acetyl-
pesticides, as glufosinate, fosetyl, ethephon and their metabolites which glyphosate in fat, liver and kidney are still missing.
form an extremely challenging group of pesticides to be analysed due to Most of the pesticides included in our previous publication (Herrera


Corresponding author.
E-mail addresses: sonia.herreralopez@wur.nl (S.H. López), jonatan.dias@wur.nl (J. Dias), andre.dekok@wur.nl (A. de Kok).

https://doi.org/10.1016/j.foodcont.2020.107289
Received 12 February 2020; Received in revised form 28 March 2020; Accepted 30 March 2020
Available online 06 April 2020
0956-7135/ © 2020 Elsevier Ltd. All rights reserved.
S.H. López, et al. Food Control 115 (2020) 107289

López et al., 2019) showed a strong matrix effect. Fruit, vegetables and animal origin: cow milk, swine fat, liver (bovine, swine and poultry),
cereals were evaluated and it was already observed that the matrix kidney (bovine and swine), chicken meat and chicken eggs. The sample
effects were depending on the individual matrix and not on a com- types have been selected for the study because they are included in the
modity group. There are different quantitation approaches to com- EU pesticide residues monitoring programme for samples of animal
pensate for matrix effect and recovery losses according to the SANTE origin for 2020 to 2022 (European Commission, 2019). Cow milk, fat,
guideline (European Commission, 2017), matrix-matched calibration liver and kidney samples have been collected directly from a slaugh-
(Reichert et al., 2018; Song et al., 2019), procedural standard calibra- terhouse in the Netherlands where samples are taken for our residue
tion (Martins et al., 2016), standard addition (Machado, Gérez, Pistón, monitoring programs. Chicken meat and eggs were purchased in a local
Heinzen, & Cesio, 2017) and isotopically labelled internal standards store for biological products, in Wageningen, The Netherlands. The
(Herrera López et al., 2019; Zhao, Zhao, Lei, Guo, & Zhao, 2018). Iso- samples were homogenised and then stored at −20 °C until spiking and
topically labelled internal standards (ILIS) is the most effective ap- sample extraction. Samples were prepared and processed according to
proach to correct for matrix effect and recovery losses but the cost of the SANTE document for pesticide residues analysis in food and feed
the analysis is increasing when the ILIS are used, because they use to be (European Commission, 2017). For quantitation purposes, internal
expensive. For some compounds, an ILIS is not commercially available standards were used for all compounds. For N-acetyl-AMPA and bro-
and then the other alternatives have to be used to correct for matrix mide, isotopically-labelled standards were not commercially available,
effects. therefore, the same internal standards as for ethephon and chlorate
In the present study, our developed Dutch Polar Pesticides (NL-PP) were used, respectively.
method (Herrera López et al., 2019) has been used for the analysis of
products of animal origin: liver, kidney, fat, chicken meat, chicken eggs 2.2.1. Extraction procedure for liver, kidney, chicken meat and chicken eggs
and cow milk. Slight modifications in the procedure have been in- 2.0 g of sample were weighed in a 50 mL centrifuge tube and spiked
cluded, in order to improve the results for some compounds in some of with isotopically labelled internal standard (ILIS) solution at
the matrices. Freezing out at −80 °C, a clean-up step with Mixed mode 250 μg kg−1. Then, 8 mL of ultra-pure water were added and the
Cation Exchanged (MCX) sorbent or addition of EDTA solution before samples were vortexed during 30 s. Then, 10 mL of MeOH with 1% of
the extraction are some of the changes included in the procedure. The formic acid were added. The tubes were shaken in an automatic axial
method has been validated for 14 highly polar pesticides (parents and extractor (AGYTAX®, Cirta Lab. S.L., Spain) during 5 min and they were
metabolites) according to the European Union (EU) SANTE Guideline placed in the freezer at −80 °C for 15 min. Thereafter, the extracts were
(European Commission, 2017). Different quantitation approaches, iso- centrifuged at 4150 R.C.F. for 10 min, at 10 °C, and passed through an
topically labelled internal standards, procedure standard calibration Oasis® (Mixed mode cation exchange) MCX cartridge from Waters
and matrix matched calibration, have been tested in order to find the (Milford, New Zealand) for the clean-up procedure. Finally, they were
best way to quantify the analytes in these type of matrices. diluted 10 times extra with dilution solvent (mixture of ACN/H2O
(60:40) and 0.2% TFA), prior to injection into the LC–MS/MS system.
2. Experimental Thereto, a 50 μL aliquot of the extract was diluted with 450 μL of in-
jection solvent. The final matrix equivalent concentration in the diluted
2.1. Chemicals and reagents extract in the vial was 0.01 g mL−1, which corresponds with a dilution
factor of 100. In the case of chicken eggs, the method without MCX was
High purity pesticide (> 98%) standards of ethephon, glufosinate, validated.
fosetyl, phosphonic acid, glyphosate and HEPA were purchased from
LGC-Dr. Ehrenstorfer (Augsburg, Germany), N-acetyl-AMPA, N-acetyl- 2.2.2. Extraction procedure for cow milk
glufosinate, N-acetyl-glyphosate and MPPA from Toronto Research 2.0 g of sample were weighed in a 50 mL centrifuge tube and spiked
Chemicals, TRC (North York, Canada), AMPA was purchased from with isotopically labelled internal standard (ILIS) solution at
Sigma–Aldrich (Steinheim, Germany) and bromide, chlorate and per- 250 μg kg−1. Then, 2 mL of EDTA solution (5 mM) and 6 mL of ultra-
chlorate from Inorganic Ventures (Christiansburg, Virginia, USA). pure water were added and the samples were vortexed during 30 s.
Isotopically labelled internal standards, AMPA 13C15N, ethephon D4, Then, 10 mL of MeOH with 1% of formic acid were added. The tubes
fosetyl–aluminium D15, HEPA D4, glufosinate D3 hydrochloride, N- were shaken in an automatic axial extractor (AGYTAX®, Cirta Lab. S.L.,
acetyl-glufosinate D3, MPPA D3, glyphosate 1,2–13C2, 15N, N-acetyl- Spain) during 5 min and they were placed in the freezer at −80 °C for
glyphosate D3, were purchased from LGC-Dr. Ehrenstorfer and from 15 min. Thereafter, the extracts were centrifuged at 4150 R.C.F. for
Toronto Research Chemicals. Phosphonic acid- 18O3, 18O3-chlorate, 10 min, at 10 °C and filtered using a 0.2 μm PVDF syringe filter from
18
O4-perchlorate were supplied by the EURL-SRM in Stuttgart, Pall corporation (New York, United States). Finally, they were diluted 5
Germany. Ethylenediaminetetraacetic acid (EDTA) disodium salt de- times extra with dilution solvent (mixture of ACN/H2O (60:40) and
hydrate, Titriplex, was obtained from Merck (Darmstadt, Germany). 0.2% TFA), prior to injection into the LC–MS/MS system. Thereto, a100
HPLC-grade water from a Water Purification System of Millipore μL aliquot of the extract was diluted with 400 μL of injection solvent.
(Burlington, MA, United States) was used. Formic acid was purchased The final matrix equivalent concentration in the diluted extract in the
from VWR (Lutterworth, United Kingdom) and trifluoroacetic acid from vial was 0.02 g mL−1, which corresponds with a dilution factor of 50.
Merck (Darmstadt, Germany). For the optimisation of experimental
conditions (ion-source-dependent parameters, and for MS/MS opera- 2.2.3. Extraction procedure for fat
tion), working solutions of individual standards were prepared in me- Fat samples were melted in a water bath at 60 °C. Then, 2.0 g of
thanol or water at 100 ng mL−1. For the linearity study, working so- sample were weighed in a 50 mL centrifuge tube and spiked with iso-
lutions of a mixture of the standards were prepared at different topically labelled internal standard (ILIS) solution at 250 μg kg−1.
concentration levels in ACN:H2O (60:40, 0.2% TFA) and these were Then, 10 mL of ultra-pure hot water (60 °C) were added and the sam-
kept at 4 °C. UPLC-grade acetonitrile and LC-grade methanol were ples were vortexed during 30 s. Then, 10 mL of MeOH with 1% of
supplied by Biosolve (Dieuze, France) and Merck (Darmstadt, formic acid were added. The tubes were shaken in an automatic axial
Germany), respectively. extractor (AGYTAX®, Cirta Lab. S.L., Spain) during 5 min and they were
placed in the freezer at −80 °C for 15 min. Thereafter, the extracts were
2.2. Sample preparation and extraction procedure centrifuged at 4150 R.C.F. for 10 min, at 10 °C and filtered, using a
0.2 μm PVDF syringe filter from Pall corporation (New York, United
This study has been carried out with different kind of samples of States). Finally, they were diluted 5 times extra with dilution solvent

2
S.H. López, et al. Food Control 115 (2020) 107289

(mixture of ACN/H2O (60:40) and 0.2% TFA), prior to injection into the especially for some pesticides/commodity combinations.
LC–MS/MS system. Thereto, a100 μL aliquot of the extract was diluted The test has been carried out for two animal product matrices, but
with 400 μL of injection solvent. The final matrix equivalent con- from different individual animals: liver (bovine, swine and poultry) and
centration in the diluted extract in the vial was 0.02 g mL−1, which kidney (bovine and swine). The different approaches have been tested
corresponds with a dilution factor of 50. in order to evaluate the difference in the quantitation for this group of
polar pesticides in matrices from the same commodity group and the
same type of matrix, but from different animal origin.
2.3. LC-MS/MS system

A hybrid quadrupole/linear ion trap mass spectrometer system


3. Results and discussion
(6500+ QTRAP®, Sciex Instruments, Concord, Ontario, Canada) with
an electrospray ion (ESI) source, coupled to a Shimadzu LC-system
3.1. Clean-up
consisting of two Nexera X2 LC-30AD LC-pumps and a SIL-30AC au-
tosampler (Shimadzu, Kyoto, Japan) has been used for method devel-
Different SPE clean-up sorbents were tested in our previous study
opment and validation. The LC analysis was performed with a HILIC-
for fruits, vegetables and cereals, in order to improve the LOQs for some
column, Obelisc-N (5 μm, 100 Å), L = 150 mm, ID = 2,1 mm (SIELC,
compounds in some of the matrices evaluated (Herrera López et al.,
Wheeling, IL, USA). The LC system operated with mobile phase A
2019). No significant improvement was observed in that study for any
(water with 1% formic acid) and mobile phase B (ACN). The LC-ESI-TQ-
pesticide/matrix combination. Considering the difference between
MS system was operated in the MRM (multiple reaction monitoring)
products of animal origin and those of plant origin, the effect of clean-
and negative ionization mode with a unit mass resolution set for Q1 and
up sorbents was also studied for difficult matrices as chicken eggs, liver
Q3. Declustering potential (DP), entrance potential (EP), collision en-
and kidney. Four different clean-up methods were tested: dispersive
ergy (CE) and collision cell exit potential (CXP), were optimized using
SPE with Enhanced Matrix Removal for lipid (EMR-lipid) and Primary-
flow injection analysis (FIA). The acquisition method used for the
Secondary Amine (PSA) sorbent, Mixed-mode Cation-eXchange (MCX)
analysis has been described in our previous publication (Herrera López
SPE cartridges, and freezing out at −80 °C. During all the experiments,
et al., 2019).
visually cleaner extracts were obtained after the clean-up step. As with
the products of plant origin, for most of the compounds, also no sig-
2.4. Different quantitation approaches nificant differences were observed now between the various clean-up
steps tested. However, it was beneficial in order to reduce the instru-
Due to the variable and strong matrix effects observed in a previous ment maintenance time and to increase the robustness of the final
evaluation of this group of polar pesticides (Herrera López et al., 2019) method. There were three compounds, glufosinate, glyphosate and
and after observing inconclusive results for absolute recoveries of HEPA, which presented more difficulties. Significantly different results
AMPA in liver samples during the recovery study, an extra evaluation of were obtained when a clean-up was applied. In Fig. 1, chromatograms
different quantitation approaches has been carried out. Isotopically for these compounds in two of the six matrices evaluated (liver and
labelled internal standards calibration, matrix-matched calibration and chicken eggs) are shown. For liver, slightly better results were observed
procedural standard calibration have been compared for quantitation. It for glyphosate in terms of detection of the qualifier ion, using MCX in
is well known, that the most effective way to compensate for matrix the clean-up step (see Fig. 1b). However, due to some remaining in-
effects, is the use of standard addition or isotopically labelled internal terference from the matrix, the ion ratio did still not meet the criterion
standards. When isotopically labelled standards are not available or are from the SANTE guideline ( ± 30%).
too costly, the use of procedural standards is often considered as the HEPA showed also better results when MCX sorbent was used (see
best alternative to compensate for matrix effects and low recoveries, Fig. 1c). For liver, the positive effect on the response of both the

Fig. 1. Extracted ion chromatograms of problematic compounds, (a) glufosinate, m/z: 180.0 → 95.0 (blue line), 180.0 → 85.0 (pink line in the back), (b) glyphosate,
m/z: 168.0 → 63.0 (blue line), 168.0 → 81.0 (pink line in the back) and (c) HEPA, m/z: 125.0 → 95.0 (blue line), 125.0 → 107.0 (pink line in the back). (For
interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

3
S.H. López, et al. Food Control 115 (2020) 107289

Fig. 1. (continued)

quantifier and qualifier ions allowed to reach a lower LOQ and for 3.2. Use of ethylenediaminetetraacetic acid (EDTA) disodium salt in the
chicken egg, the use of MCX sorbent was essential to detect the com- extraction procedure
pound in this matrix (see Fig. 1c). However, as a disadvantage, MCX
clean-up caused the partial loss of glufosinate for liver, chicken meat The use of EDTA in the extraction procedure has also been evaluated
and other validated matrices, and even the complete loss of glufosinate in this work. It is well known that glyphosate can behave as a chelator
for chicken eggs (see Fig. 1S). These losses are likely to be related with for metal ions due to its chemical structure. It may bind with metal ions
the interactions between the sulphite functional group of the sorbent present in the samples. In our study, there are two specific matrices,
(MCX) and the amino group in the glufosinate structure. The interaction milk and liver, that may contain high concentrations of polyvalent ca-
seems to depend on the percentage of proteins in the matrix. Glufosi- tions, such as calcium and magnesium. The use of EDTA as chelating
nate was completely lost during the clean-up step in matrices as oat and agent in milk samples for the determination of veterinary drugs, like the
egg, with a percentage of proteins between 11 and 17. However, in antibiotic tetracycline, has already been described in previous pub-
matrices as chicken meat, kidney, liver and soya beans, with a per- lications, because this antibiotic may also bind to the metal ions present
centage of proteins in the range 20–38, it was only partially lost. This in the samples, as can happen with glyphosate (Bruno, Curini, Di
could be explained by the competition between the amino group in the Corcia, Nazzari, & Pallagrosi, 2002; Pena, Lino, & Silveira, 1999). Also,
proteins and in the glufosinate molecule for the sulphite functional a previous application for anionic polar pesticides using EDTA in the
group in the sorbent. Therefore, when the amount of proteins in the extraction solvent has been reported for milk samples (Chamkasem,
matrix increases, the sulphite group is completely blocked by the amino Morrisa, & Harmona, 2015). In our study, the addition of EDTA solution
groups of the proteins and glufosinate will not be irreversibly bound to before the extraction was tested together with the different clean-up
the sorbent. steps mentioned in the previous section. For liver samples, no

Fig. 1. (continued)

4
S.H. López, et al. Food Control 115 (2020) 107289

Fig. 2. Extracted ion chromatograms (quantifier and qualifier ions) for glyphosate in cow milk with and without addition of EDTA solution before the extraction.

improvements were observed in terms of recovery when EDTA was +60% have been observed. Chicken eggs and swine fat showed a high
used. However, for milk samples, the use of EDTA was critical to assure percentage of compounds, 86 and 79%, respectively, with an insignif-
the good recovery of glyphosate from the samples. Fig. 2 shows the icant matrix effect (< ± 20%). Cow milk and chicken meat showed the
extracted ion chromatograms of the quantifier and qualifier ions for highest number of compounds with strong negative matrix effect. HEPA
glyphosate in cow milk extracts, using EDTA or not in the extraction showed a strong negative matrix effect in both matrices (−86 and
procedure. It is clear from this figure, that glyphosate is not recovered −50%) and AMPA a stronger negative matrix effect in cow milk
from the sample when EDTA is not used. However, when EDTA is added (−58%) than in chicken meat (−35%). Glyphosate and N-acetyl-glu-
to the samples before extraction, EDTA competes for the binding to the fosinate were also compounds with significant suppression of the re-
polyvalent cations and displaces glyphosate molecules attached to the sponse in cow milk, −45 and −27% respectively. The other two
cations, resulting in the full recovery from the samples. compounds with suppression in the chicken meat were N-acetyl-AMPA
and ethephon (−35 and −24%, respectively).
3.3. Validation Liver and chicken meat showed the highest number of compounds
with a positive matrix effect. Fosetyl and MPPA had a matrix effect
3.3.1. Linearity around 45% in liver and 27 and 35%, in chicken meat, respectively.
In Table 1S, the results from the linearity study have been sum- Glyphosate had a matrix effect of 38% in chicken meat.
marized for the six matrices evaluated (cow milk, swine fat, liver, Glyphosate showed a negative matrix effect for swine fat and cow
kidney, chicken meat and chicken eggs). Calibration standards in sol- milk. On the other hand, it showed a positive or non-significant matrix
vent at 5 concentration levels were injected 4 times in order to get the effect in liver, kidney, chicken eggs and chicken meat. This positive
results for linearity of the calibration curve, instrument detection limits, effect is a consequence of the clean-up used for these four matrices.
estimated method detection and quantitation limits. The calibration When using MCX as sorbent in the clean-up, some interfering ions from
curve constructed from single injections of matrix-matched calibration the matrix were removed and the response of the target ion improved.
standards was also checked for linearity. For all pesticides, the criterion However, the lack of sensitivity of the qualifier ion made the identifi-
for a linear calibration curve (r2 > 0.98) is met, both for standards in cation of glyphosate impossible at the lower levels evaluated. The same
solvent and in matrix extract (see Table 1S). The deviation of the back- effect was already observed before in our previous validation study for
calculated concentrations of the calibration standards from the true oat and soy beans, when different clean-up approaches were tested to
concentrations is always in the range of ± 20%, as required according improve the LOQs of the method (Herrera López et al., 2019). This is
to the criterion in the European guideline for the analysis of pesticide the reason why glyphosate was not fully validated according to the
residues in food and feed (European Commission, 2017). SANTE guideline criteria for these two matrices (kidney and liver) at
the lowest concentration level. Glyphosate showed the most extreme
3.3.2. Matrix effect and opposite matrix effect in cow milk (−45%) and in kidney (47%), as
The matrix effects have been calculated from the slope of the cali- is plotted in the graph in the supplementary material (Fig. 2S). Due to
bration curves from standards in solvent and in matrix for the six the very variable results of the matrix effects, isotopically-labelled in-
evaluated matrices. The results are summarized in Table 1. ternal standards (ILIS) were required for the correction for matrix effect
Only fosetyl and chlorate showed consistently a significant positive and obtain a reliable quantitation in the final method.
matrix effect (> 20%) in all the matrices. For chlorate, the matrix effect
ranged from 8% in kidney to 36% in fat, for fosetyl, from 17% in kidney 3.3.3. Accuracy (trueness and precision) and method quantitation limits
to 55% in swine fat. The accuracy (trueness and precision) of the method has been de-
In general, variable matrix effects within the range of −86 to termined by analysing recovery samples at 4 different spike levels,

5
S.H. López, et al. Food Control 115 (2020) 107289

Table 1
Matrix effects calculated from the slopes of calibration curves from standards in solvent and in matrix.
Pesticides Matrix effect (%)

Swine fat Cow milk Liver Chicken eggs Kidney Chicken meat

AMPA −2 −58 −77 −6 −45 −35


Bromide 10 −1 4 −1 −13 7
Chlorate 36 33 19 10 8 8
Ethephon −6 −12 −1 −4 −3 −24
Fosetyl 55 48 44 27 17 27
Glufosinate −5 1 6 1 4 8
Glyphosate −13 −45 23 −10 47 38
HEPA −25 −86 −47 −13 −40 −50
MPPA −4 59 48 2 60 35
N-acetyl-AMPA −1 −13 3 −1 −7 −35
N-acetyl-glufosinate 2 −27 1 3 −7 1
N-acetyl-glyphosate 20 5 14 36 1 −5
Perchlorate 0 −11 2 −1 −6 −4
Phosphonic acid 5 0 9 2 −1 0

depending on the matrix evaluated, and with 6 replicates (n = 6) for the interferences from the matrix, also mainly for the qualifier ion.
each spike level. The trueness (as % recovery, n = 6) and precision (as However, they could easily be detected at that concentration with one
% RSDr, n = 6) of the method were determined. The results for “ab- ion and perfect matching with the target ion from the isotopically la-
solute” and ILIS-corrected recoveries are summarized in table 2S, 3S, belled internal standards. That concentration was then set as screening
4S, 5S, 6S, 7S and 8S. Absolute recoveries have been calculated without detection limit (SDL).
using internal standards and with matrix-matched calibration stan- Blank samples of all matrices to be used in the validation study have
dards. The corrected recoveries were calculated using the standards in been analysed to determine the possible background contamination for
solvent and isotopically labelled internal standards. the target pesticides. Bromide was present at different concentrations in
Some compounds could not be validated at the lower concentration milk, eggs, chicken meat and kidney within the range of 0.8–3 mg kg−1.
level 0.02 mg kg−1, fulfilling all the requirements from the SANTE The least contaminated samples were used for the validation and the
document, due to the lack of sensitivity of the qualifier ion or because of background concentrations present in the samples were subtracted to

Fig. 3. Extracted ion chromatograms (quantifier and qualifier ions) and ion ratios for glyphosate in bovine and swine kidney, at different spike levels, (a) glyphosate
(0.02 mg kg-1) - Kidney, (b) glyphosate (0.2 mg kg-1) - Kidney and (c) glyphosate (0.5 mg kg-1) – Kidney.

6
S.H. López, et al. Food Control 115 (2020) 107289

Fig. 3. (continued)

calculate the real recoveries. When the concentration in the sample was glyphosate, validated at 0.05 mg kg−1 (SDL 0.01 mg kg−1) and HEPA,
very high, it was not possible to validate this compound at the lowest which could not be validated at any concentration (see Table 4S), it was
levels, even though detection and correct identification at these levels decided to employ a clean-up step in order to improve the results for
was not a problem. these compounds. Fig. 3Sa and 3Sb show the chromatograms at 3
concentration levels (0.02, 0.05 and 0.1 mg kg−1) for both compounds
3.3.3.1. Milk. From the 14 evaluated compounds, 12 showed with and without MCX in the clean-up step.
recoveries between 70 and 120% and RSD ≤20% for all the The use of MCX clean-up improved the results for glyphosate,
concentrations (0.01, 0.02, 0.05 and 0.1 mg kg−1), as shown in especially for the response of the second ion, and also for HEPA that
Table 2S. For bromide, an ILIS was not commercially available and could be detected at 0.02 mg kg−1 despite poor sensitivity for the
the recoveries were calculated using the ILIS for chlorate (18O3- second ion. On the other hand, the use of MCX caused the loss of glu-
chlorate). Because the unlabelled chlorate and bromide showed fosinate (see Fig. 3Sc).
different matrix effects (33 and −1%, respectively), the calculated In the case of eggs validation, the use of MCX clean-up or no the
bromide recoveries obtained were thus lower than they should be, that clean-up affected the individual compounds differently. The final de-
is around 70%. HEPA was only validated at the highest level, cision about the use or deletion of MCX clean-up will depend on the
0.1 mg kg−1, due to the high suppression of the response and the preferred scope of the target analytes for each analysis.
lack of sensitivity of the qualifier ion. But its SDL could be set as
0.02 mg kg−1. 3.3.3.4. Chicken meat. In the case of chicken meat, as shown in
Table 6S, glufosinate and HEPA were validated from the level of
3.3.3.2. Fat. For swine fat, with the exception of N-acetyl-glyphosate, 0.1 mg kg−1. However, the SDL for these compounds could be set at
all the compounds showed recoveries and RSDs within the acceptable 0.02 and 0.05 mg kg−1, respectively. Glyphosate and MPPA complied
range for all concentration levels, 0.01, 0.02 and 0.05 mg kg−1 (see with the criteria for recovery and precision at 0.05 mg kg−1, but a SDL
Table 3S), and could be fully validated. could be set at 0.02 mg kg−1. N-acetyl-AMPA showed lower ILIS-
corrected recoveries than the absolute recoveries, but they were still
3.3.3.3. Chicken eggs. For chicken eggs, the validation was carried out within the range 70–120%. This could be explained by the
with and without clean-up with MCX sorbent (see results in Tables 4S unavailability of the corresponding ILIS for N-acetyl-AMPA and the
and 5S). The first validation was performed without clean-up and most use of Ethephon-D4 as a substitute ILIS. The different matrix effects
of the compounds met the criteria of the SANTE document at the lowest shown for N-acetyl-AMPA and Ethephon-D4 is probably the reason why
validated level (0.02 mg kg−1). However, considering the results for the ILIS-corrected recoveries are lower than the uncorrected ones.

7
S.H. López, et al. Food Control 115 (2020) 107289

Fig. 3. (continued)

3.3.3.5. Kidney. Several samples of bovine and swine kidney were three livers tested, as it was also observed during the evaluation of
tested to evaluate possible background levels of some of the target kidney. The worst results were for swine liver. In Fig. 4, the extracted
compounds in the blank samples. As has been mentioned above, all the ion chromatograms of the three livers, at three different spike levels,
samples tested were positive for bromide. The samples with the lower are shown. At the lowest spike level, 0.02 mg kg−1, the full
level (3 mg kg−1) for bovine and swine kidney were chosen to use for identification was not possible in any type of liver evaluated. At the
the validation study. Bromide could thus only be validated at two higher levels tested (0.2, 0.5 mg kg−1), the identification was
5 mg kg−1. HEPA was also detected in the blank samples of bovine possible for bovine and poultry liver, but not for swine liver. For swine
kidney at 0.07 mg kg−1. This concentration was subtracted from the liver, the full identification was only possible at the highest level for
blank samples for the recovery calculation. As can been seen in glyphosate. However, the SDL for glyphosate could be set for all three
Table 7S, glufosinate was validated at 0.05, 0.2 and 0.5 mg kg−1 and types of liver at 0.02 mg kg−1 (see Fig. 4a). In Table 8S the recovery
its SDL could be set at 0.02 mg kg−1. data at 0.2 mg kg−1 are for bovine and poultry liver and at 0.5 mg
In this matrix, glyphosate showed different results in the two dif- kg−1for all three of types.
ferent types of kidney evaluated. The interferences from the matrix The LOQs and screening detection limits of the final method using
affected the qualifier ion in different ways depending on the type of the ILIS for quantitation are shown in Table 2.
kidney (see Fig. 3). The recoveries for glyphosate in Table 7S at In summary, the following validation results have been obtained.
0.2 mg kg−1 are only for swine kidney because the second ion could not Most of the compounds have been validated at 0.01 mg kg−1 for cow
be resolved from the interferences from the bovine kidney matrix and milk and swine fat. Only N-acetyl-glufosinate and N-acetyl-glyphosate
hence these data were not considered. However the results at were validated at 0.02 mg kg−1 in swine fat and HEPA at 0.1 mg kg−1
0.5 mg kg−1 are for swine and bovine kidney, because the ion ratio in cow milk. In the case of kidney and liver 70 and 64% of the com-
criterion ( ± 30%) is met. See also Fig. 3b and c. Despite the validated pounds, respectively, were validated at 0.02 mg kg−1. Glufosinate was
LOQs have been set at the two higher concentrations, glyphosate could validated in kidney at 0.05 mg kg−1 and HEPA at 0.2 mg kg−1 in liver.
be detected (quantifier ion) at the lowest level evaluated for both types Glyphosate showed LOQs of 0.5 mg kg−1 for these two matrices. In
of kidney (Fig. 3a). Therefore, its SDL could be set at 0.02 mg kg−1. chicken meat, 8 compounds were validated at 0.02 mg kg−1, two
compounds (glyphosate and MPPA) at 0.05 mg kg−1 and another two
3.3.3.6. Liver. The validation for the matrix liver was performed using (glufosinate and HEPA) at 0.1 mg kg−1. Bromide and phosphonic acid
3 different types of liver (bovine, swine and poultry, n = 2 for each) were validated at 0.2 mg kg−1, the lowest spike levels tested for these
(see results in Table 8S). Ten out of fourteen compounds were validated compounds. For chicken eggs, all the compounds were validated at the
from the lowest level tested (0.02 mg kg−1). During validation using lowest spike level tested. In order to be able to detect glufosinate, no
the clean-up with MCX, glyphosate showed a different behaviour in the clean-up should be applied and in that case its LOQ is 0.01 mg kg−1.

8
S.H. López, et al. Food Control 115 (2020) 107289

Fig. 4. Extracted ion chromatograms (quantifier and qualifier ions) and ion ratios for glyphosate in bovine, poultry and swine liver, at different spike levels, (a)
glyphosate (0.02 mg kg-1) - Liver, (b) glyphosate (0.2 mg kg-1) - Liver and (c) Glyphosate (0.5 mg kg-1) – Liver.

Fig. 4. (continued)

9
S.H. López, et al. Food Control 115 (2020) 107289

Fig. 4. (continued)

Table 2
Limits of quantification (LOQs) and screening detection limits (SDLs) of the final validated method for the six evaluated matrices.
Pesticide LOQs (mg kg−1)

Cow Milk Swine Fat Kidney Liver Chicken meat Chicken eggs

AMPA 0.01 0.01 0.02 0.02 0.02 0.02


Bromide 0.5 0.1 5 5 0.2 2
Chlorate 0.01 0.01 0.02 0.02 0.02 0.02
Ethephon 0.01 0.01 0.02 0.02 0.02 0.02
Fosetyl 0.01 0.01 0.02 0.02 0.02 0.02
Glufosinate 0.01 0.01 0.05 (0.02c) 0.2 0.1 (0.02c) 0.01d
Glyphosate 0.01 0.01 0.2 (0.02c)
a
0.2 (0.02c)
b
0.05 (0.02c) 0.02
HEPA 0.1 (0.02c) 0.01 0.02 0.2 0.1 (0.02c) 0.02
MPPA 0.01 0.01 0.02 0.02 0.05 (0.02c) 0.02
N-acetyl-AMPA 0.01 0.01 0.02 0.02 0.02 0.02
N-acetyl-glufosinate 0.01 0.01 0.02 0.02 0.02 0.02
N-acetyl-glyphosate 0.01 0.02 0.02 0.02 0.02 0.02
Perchlorate 0.01 0.01 0.02 0.02 0.02 0.02
Phosphonic acid 0.1 0.1 0.2 0.2 0.2 0.2

a
LOQ for swine kidney.
b
LOQ for bovine and poultry liver.
c
SDL= Screening detection limit.
d
LOQ for method without clean-up.

3.4. Different quantitation approaches constructed with external standards solutions prepared in solvent or
matrix-matched with poultry liver extract and procedural standards in
During the recovery study of liver matrix, the absolute recoveries bovine, swine and poultry liver extract. The different quantitative re-
for all pesticides were also calculated using matrix-matched, external sults obtained using the different quantitation approaches were com-
calibration standards using bovine liver extracts. It was observed that, pared. In Fig. 5, the five different calibration curves have been plotted
for AMPA, at all four levels, absolute recoveries above 200% were for (a) AMPA and (b) fosetyl. Fig. 5a shows the different response for
calculated (see Table 9S). This effect was not observed for all the pes- AMPA in solvent and in the three types of kidney. But it is also clear in
ticides included in the method, as is shown in Table 9S for fosetyl, as an the graph, that the response for poultry liver is different than the re-
example. The results for AMPA may be explained by different matrix sponse for bovine and swine liver. Due to this difference, the absolute
effects in the three types of liver tested. In order to prove that ILIS are recoveries calculated using matrix-matched standards using bovine
required for quantitation in routine batches of samples, where different liver showed recoveries above 200% for poultry liver. These results
types of liver have to be analysed, different calibration curves were confirm that isotopically labelled internal standards are essential for the

10
S.H. López, et al. Food Control 115 (2020) 107289

Fig. 5. Five calibration curves constructed from different calibration approaches: standards in solvent, matrix-matched standards with poultry liver extract and
procedural standard calibration with poultry, bovine and swine liver. (a) AMPA and (b) fosetyl.

correct quantitation for some compound/matrix combinations, because eggs because no blank samples were available.
other approaches such as procedural standard calibration and matrix- In general, AMPA, glufosinate, glyphosate and HEPA were the
matched calibration would only work when the samples to be analysed compounds which showed relatively more difficulties during the vali-
will be of exact the same type as the blank sample used for calibration dation study. AMPA and HEPA showed the strongest matrix effects,
standards. HEPA in cow milk, −86%, and AMPA in liver, −77%. The lack of
sensitivity of the qualifier ions of Glyphosate and HEPA did not allow
4. Conclusion the full identification at the lower spike levels studied. The recoveries of
glufosinate were affected by the use of SPE cartridges with MCX as
The method for the determination of fourteen polar pesticides and clean-up. The compound was retained on the sorbent due to the in-
their metabolites in cow milk, swine fat, chicken meat, chicken eggs, teraction between functional groups of the polymer and amino groups
liver and kidney has been validated according to the EU SANTE docu- in the glufosinate molecule. The retention was not the same for all
ment, using isotopically labelled internal standards for quantitation. matrices evaluated. The compound was fully retained during the ex-
This is the first publication on the analysis of these fourteen anionic periments with chicken eggs, but for the other matrices glyphosate was
polar pesticides in products of animal origin, using one extraction at least detectable at the higher concentrations evaluated.
method and a single, fast chromatographic run. Isotopically labelled internal standards are essential in terms of
The main conclusion of this study is that the results are totally de- quantitation for some compound/matrix combinations, because it has
pendent on the individual matrix types and not on the commodity been observed that other approaches such as procedural standard ca-
group. This was already observed in the previous validation for fruit, libration and matrix matched calibration did not work when the sample
vegetables, oat and soya, but in this study not only differences between matrices to be analysed are different from the type of matrix used to
matrices belonging to the same commodity group as liver and kidney prepare the calibration standard solutions.
were observed, but also differences between individual types of liver or The validated method appears to be suitable for the intended pur-
kidney. pose, that is the determination of 14 polar pesticides/metabolites, in the
The method quantitation limits are, for most of the compounds and six matrices mentioned above. The method can behave different for the
in most of the matrices evaluated, 0.01 or 0.02 mg kg−1. Some com- various individual matrices from the same SANTE commodity group to
pounds had higher LOQs, because they were intentionally validated at which kidney and liver belong (group 7, offal). Therefore, the on-going
higher levels (phosphonic acid and bromide), or because of their lower validation with different types of matrices should be carried out, to be
sensitivity, especially for the qualifier ion (as for glufosinate, glyphosate able to evaluate the target compounds in other matrices during the
and HEPA). Bromide could not be validated in liver, kidney and chicken application of the method in the routine analysis.

11
S.H. López, et al. Food Control 115 (2020) 107289

CRediT authorship contribution statement determination of glyphosate and aminomethylphosphonic acid in environmental
water samples by high performance liquid chromatography coupled to electrospray
tandem mass spectrometry. Journal of Chromatography A, 1443, 93–100. https://doi.
Sonia Herrera López: Conceptualization, Data curation, Formal org/10.1016/j.chroma.2016.03.020.
analysis, Investigation, Methodology, Validation, Writing - original Hanke, I., Singer, H., & Hollender, J. (2008). Ultratrace-level determination of glypho-
draft, Writing - review & editing. Jonatan Dias: Data curation, Formal sate, aminomethylphosphonic acid and glufosinate in natural waters by solid-phase
extraction followed by liquid chromatography-tandem mass spectrometry:
analysis, Investigation, Methodology, Validation, Writing - review & Performance tuning of derivatization, enrichment and detection. Analytical and
editing. André de Kok: Investigation, Project administration, Bioanalytical Chemistry, 391, 2265–2276. https://doi.org/10.1007/s00216-008-
Supervision, Validation, Writing - review & editing. 2134-5.
Herrera López, S., Scholten, J., Kiedrowska, B., & de Kok, A. (2019). Method validation
and application of a selective multiresidue analysis of highly polar pesticides in food
Declaration of competing interest matrices using hydrophilic interaction liquid chromatography and mass spectro-
metry. Journal of Chromatography A, 1594, 93–104. https://doi.org/10.1016/j.
chroma.2019.02.024.
The authors declare that they have no known competing financial Machado, I., Gérez, N., Pistón, M., Heinzen, H., & Cesio, M. V. (2017). Determination of
interests or personal relationships that could have appeared to influ- pesticide residues in globe artichoke leaves and fruits by GC–MS and LC–MS/MS
ence the work reported in this paper. using the same QuEChERS procedure. Food Chemistry, 227, 227–236. https://doi.org/
10.1016/j.foodchem.2017.01.025.
Martins, M. L., Rizzetti, T. M., Kemmerich, M., Saibt, N., Prestes, O. D., Adaime, M. B.,
Acknowledgement et al. (2016). Dilution standard addition calibration: A practical calibration strategy
for multiresidue organic compounds determination. Journal of Chromatography A,
1460, 84–91. https://doi.org/10.1016/j.chroma.2016.07.013.
This work was funded by the Netherlands Food and Consumer Pena, A. L., Lino, C. M., & Silveira, I. N. (1999). Determination of oxytetracycline, tet-
Product Safety Authority (NVWA). racycline, and chlortetracycline in milk by liquid chromatography with post-column
derivatization and fluorescence detection. Journal of AOAC International, 82, 55–60.
https://doi.org/10.1093/jaoac/82.1.55.
Appendix A. Supplementary data Rajski, L., Díaz Galiano, F. J., Cutillas, V., & Férnandez-Alba, A. R. (2018). Coupling ion
chromatography to Q-orbitrap for the fast and robust analysis of anionic pesticides in
Supplementary data to this article can be found online at https:// fruits and vegetables. Journal of AOAC International, 101, 352–359. https://doi.org/
10.5740/jaoacint.17-0410.
doi.org/10.1016/j.foodcont.2020.107289.
Reichert, B., de Kok, A., Pizzutti, I. R., Scholten, J., Cardoso, C. D., & Spanjer, M. (2018).
Simultaneous determination of 117 pesticides and 30 mycotoxins in raw coffee,
References without clean-up, by LC-ESI-MS/MS analysis. Analytica Chimica Acta, 1004, 40–50.
https://doi.org/10.1016/j.aca.2017.11.077.
Song, N., Lee, J. Y., Mansur, A. R., Jang, H. W., Lim, M., Lee, Y., et al. (2019).
Benbrook, C. M. (2016). Trends in glyphosate herbicide use in the United States and Determination of 60 pesticides in hen eggs using the QuEChERS procedure followed
globally. Environmental Sciences Europe, 28. https://doi.org/10.1186/s12302-016- by LC-MS/MS and GC-MS/MS. Food Chemistry, 298, 125050. https://doi.org/10.
0070-0. 1016/j.foodchem.2019.125050.
Bruno, F., Curini, R., Di Corcia, A., Nazzari, M., & Pallagrosi, M. (2002). An original Thelin, G. P., & Stone, W. W. (2013). Estimation of annual agricultural pesticide use for
approach to determining traces of tetracycline antibiotics in milk and eggs by so- counties of the conterminous United States, 1992–2009. National water-quality assessment
lid‐phase extraction and liquid chromatography/mass spectrometry. Rapid programU. S. Geological Survey Scientific Investigations54p. https://doi.org/10.
Communications in Mass Spectrometry, 16, 1365–1376. https://doi.org/10.1002/rcm. 3133/sir20135009 Report 2013-5009.
724. Vass, A., Robles-Molina, J., Pérez-Ortega, P., Gilbert-López, B., Dernovics, M., Molina-
Cerdeira, A. L., Gazziero, D. L. P., Duke, S. O., & Matallo, M. B. (2011). Agricultural Díaz, A., et al. (2016). Study of different HILIC, mixed-mode, and other aqueous
impacts of glyphosate-resistant soybean cultivation in south America. Journal of normal-phase approaches for the liquid chromatography/mass spectrometry-based
Agricultural and Food Chemistry, 59, 5799–5807. https://doi.org/10.1021/jf102652y. determination of challenging polar pesticides. Analytical and Bioanalytical Chemistry,
Chamkasem, N., Morrisa, C., & Harmona, T. (2015). Direct Determination of Glyphosate, 408, 4857–4869.
Glufosinate, and AMPA in milk by Liquid chromatography/tandem mass spectro- Vila-Aiub, M. M., Yu, Q., & Powles, S. B. (2019). Do plants pay a fitness cost to be resistant
metry. Journal of Regulatory Science, 2, 20–26. https://doi.org/10.1007/s00216-016- to glyphosate? New Phytologist, 223, 532–547. https://doi.org/10.1111/nph.15733.
9597-6. Zhang, L., Rana, I., Shaffer, R. M., Taioli, E., & Sheppard, L. (2019). Exposure to gly-
European Commission (2017). Guidance document on analytical quality control and method phosate-based herbicides and risk for non-hodgkin lymphoma: A meta-analysis and
validation procedures for pesticides residues analysis in food and feed. SANTE. Document supporting evidence. Mutation Research: Reviews in Mutation Research, 781, 186–206.
No. SANTE/11813/2017. https://doi.org/10.1016/j.mrrev.2019.02.001.
European Food Safety Authority (EFSA) (2018). Review of the existing maximum residue Zhao, P., Zhao, J., Lei, S., Guo, X., & Zhao, L. (2018). Simultaneous enantiomeric analysis
levels for glyphosate according to Article 12 of Regulation (EC) No 396/2005. EFSA of eight pesticides in soils and river sediments by chiral liquid chromatography-
Journal, 16. https://doi.org/10.2903/j.efsa.2018.5263. tandem mass spectrometry. Chemosphere, 204, 210–219. https://doi.org/10.1016/j.
European Food Safety Authority (EFSA) (2019). Review of the existing maximum residue chemosphere.2018.03.204.
levels for glyphosate according to Article 12 of Regulation (EC) No 396/2005 – re- European Commision (2019). COMMISSION IMPLEMENTING REGULATION (EU) 2019/
vised version to take into account omitted data. EFSA Journal, 17. https://doi.org/10. 533 of 28 March 2019concerning a coordinated multiannual control programme of
2903/j.efsa.2019.5862. the Union for 2020, 2021 and 2022 to ensure compliance with maximum residue
Goscinny, S., Unterluggauer, H., Aldrian, J., Hanot, V., & Masselter, S. (2012). levels of pesticides and to assess the consumer exposure to pesticide residues in and
Determination of glyphosate and its metabolite AMPA (aminomethylphosphonic on food of plant and animal origin. Official Journal of European Commission. L88,
acid) in cereals after derivatization by isotope dilution and UPLC-MS/MS. Food 28-41. Available at: http://data.europa.eu/eli/reg_impl/2019/533/oj.
Analytical Methods, 5, 1177–1185. https://doi.org/10.1007/s12161-011-9361-7.
Guo, H., Riter, L. S., Wujcik, C. E., & Armstrong, D. W. (2016). Direct and sensitive

12

You might also like