374 LOZANO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 101, NO.

2, 2018

SPECIAL GUEST EDITOR SECTION

Pesticide Residue Analysis in Fruit- and Vegetable-Based

Baby Foods Using GC-Orbitrap MS
ANA LOZANO, SAMANTA UCLÉS,1 ANA UCLÉS, CARMEN FERRER, and AMADEO R. FERNÁNDEZ-ALBA
University of Almerı́a, Department of Chemistry and Physics, Agrifood Campus of International Excellence (CeiA3), European
Union Reference Laboratory for Pesticide Residues in Fruits and Vegetables, 04120 Almerı́a, Spain

This paper presents an efficiency evaluation of GC Over the last few decades, special care has been given to the
coupled with quadrupole Orbitrap MS for safety of infants and children because they represent a vulnerable
identification and quantitation in the multiresidue population group. In the European Union (EU), the European
pesticide analysis of baby foods in full-scan mode. Commission has established specific and strict legislation, setting
The identification criteria were studied following maximum residue levels (MRLs) for pesticides in food for babies,
SANTE guidelines (retention time, mass accuracy, infants, and young children, under Commission Directive No.
and ion ratio), comfortably complying with the 2006/125/EC (4). This directive sets a very low common limit
values established, even at 0.003 mg/kg. Method for all pesticides, fixed at 0.01 mg/kg (this limit is normally, in
validation was carried out on 15 selected GC- practice, the minimum detectable level), with an acceptable daily
amenable pesticides covered by Commission intake of <0.0005 mg/kg body weight. Furthermore, it sets other,
Directive No. 2006/125/EC in three different baby lower levels for a list of very toxic pesticides and their degradation
food matrixes. Recovery studies were performed at products (from 0.004 to 0.008 mg/kg) and specifies a list of certain
0.003 and 0.006 mg/kg, with 96% of the cases pesticides that must not exceed 0.003 mg/kg in the product.
falling within the 70–120% range and with Significant efforts are taken to prevent pesticide residues in food
RSDs <15% for all the pesticides assayed. Linearity products; among these is the monitoring of real samples across the
over 3 orders of magnitude was verified, with EU. The monitoring of baby food samples is carried out by each
residuals <16% and correlation coefficient EU member state, and all the sample results are statistically treated
values >0.995. In general, matrix effect values by the European Food Safety Authority (EFSA). In the last
were >100%. The LOQ was 0.003 mg/kg for 97% of published EFSA annual report for pesticide residues (5), of the
the cases. The validated method was applied to 20 1812 baby food samples analyzed in 2014, 3.8% contained
real baby food samples from Spain and to the residues exceeding the MRLs. This percentage for baby food is
European Union Proficiency Test FV-BF01 sample, slightly higher compared to that for unprocessed food products,
in which the z-scores obtained were <1, thus in which 3.1% of samples contained residues exceeding the
demonstrating that this instrumentation has good MRLs (n = 72 808). However, 1.8% of the baby food samples
quantitation capabilities. were considered noncompliant, taking into account measurement
uncertainty. Bearing in mind the low concentration levels and the
need to monitor pesticide residues in infant foods accurately,

P
esticide residues can be present in fruits and vegetables sensitive and reliable identification and quantification methods
because of the protection-plan products used in the field to are required.
avoid pests and diseases or to improve productivity. Over the last decade, few works concerning baby food
Therefore, pesticides may also be present in processed pesticide residue analysis have been published. The scope of
products, such as ready-to-eat baby food. Infants (children analytes evaluated was usually in the range of 4–24 pesticides
under the age of 12 months) and young children (1–3 years along with their degradation products, but the number evaluated
old) are especially vulnerable to chemical exposure. They are was a few hundred compounds in four publications (6–9). In
more exposed to pesticides than adults because their diet most of the studies, acetonitrile-based extraction methods were
starting at 6 months of age comprises a higher percentage used with different approaches: the QuEChERS method was
of fruits and vegetables, and their consumption per kilogram used in its original version in two papers (10, 11); in another six
body weight is greater. Furthermore, children may also absorb studies, it was applied with acetate buffer (2, 6, 7, 9, 12, 13); and
pesticides more easily because their gastrointestinal tract is other authors performed the extraction with citrate buffer (14).
still developing (1) and their liver enzymes are still immature Petrarca et al. (13) compared dispersive SPE (d-SPE) cleanup
(2, 3). with dispersive liquid–liquid microextraction, but better LOQs
were obtained by d-SPE. González-Curbelo et al. (15) and
Gómez-Pérez et al. (8) chose acetonitrile as the extraction
Guest edited as a special report on “Application of Orbitrap Mass solvent, along with the use of ultrasound or a rotary agitator,
Spectrometry in Food Analysis” by Amadeo R. Fernández-Alba. respectively, to improve solvent–matrix interaction, followed by
1
Corresponding author’s e-mail: samantaucles@gmail.com SPE with multiwalled carbon nanotubes (15) or Florisil® (8).
We acknowledge funding support from the European Commission DG Only one work selected SPE for this kind of application (3) using
SANTE, Specific Agreement No. 5 of the Framework Partnership
Agreement No. SANCO/2005/FOOD SAFETY/0025 Pesticides in Fruits
an Oasis hydrophobic lipophilic balanced cartridge and methanol
and Vegetables. with 1% acetic acid. Regarding the chromatographic technique
DOI: https://doi.org/10.5740/jaoacint.17-0413 used, an even distribution between LC and GC was observed in
 LOZANO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 101, NO. 2, 2018 375

these publications, a necessary approach in order to cover the the citrate QuEChERS method without cleanup (to minimize
analytes’ entire range of physicochemical properties. Various sample handling), followed by quantitative analyses by GC-Q-
detectors were used, including classical detectors (such as a Orbitrap MS. Method validation in three fruit- and vegetable-
nitrogen phosphorus detector), but MS detectors were used most based baby foods was carried out in terms of sensitivity, linearity,
often. Of the latter type, triple quadrupole was the preferred matrix effect (ME), and LOQ. Lastly, the validated method was
choice because of its selectivity and sensitivity, accomplishing applied to a total of 20 baby food samples bought in Spain and
reliable identification and quantification, even at trace levels to a proficiency test (PT) sample.
(16). The benefits of high-resolution accurate mass
spectrometers were also explored on four occasions (6, 8, 9, Experimental
12). The use of accurate full-scan spectra throughout the
acquisition range permits unequivocal nontarget identification, Reagents and Materials
compound confirmation, and retrospective analysis in a single
workflow (17). The technology used in this equipment has Acetonitrile HPLC grade was purchased from Merck (Darmstadt,
substantially improved and currently allows better mass Germany), and ethyl acetate of pesticide residue analysis grade was
accuracy (≤1 mDa), higher resolving power [around 60 000 purchased from Fluka (Steinheim, Germany). Trisodium citrate
full width at half-maximum (FWHM) at m/z 200], and wider dehydrate, anhydrous magnesium sulfate, and sodium chloride
linear ranges, thus improving the quantification capabilities and were provided by Sigma-Aldrich (Steinheim, Germany). Sodium
lowering the LOQ to <0.01 mg/kg (9, 18, 19). hydrogencitrate sesquihydrate was obtained from Sigma-Aldrich.
In this study, the potential of GC coupled to quadrupole (Q)- Analytical reference standards of the 15 selected pesticides
Orbitrap MS in nontarget full-scan independent acquisition mode were purchased from Dr. Ehrenstorfer (Augsburg, Germany) or
was evaluated for identification and quantitation purposes. from Sigma-Aldrich and were stored at −30°C in dark
Fifteen GC-amenable pesticides with strict levels (<0.01 mg/kg) conditions. The isotope-labeled internal standards dichlorvos-
in baby food were selected from the Commission Directive (4). D6, malathion-D10, and triphenyl phosphate [triphenyl phosphate
The presented method consists of a sample treatment based on (TPP) - procedural internal standards, (PISs) and lindane-d6

Figure 1. Structures of the studied pesticides. HCB, hexachlorobenzene.

376 LOZANO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 101, NO. 2, 2018

(injection internal standard (IIS)] were purchased from Dr. injection (1 µL, 250°C, and 1 min splitless time). The injector
Ehrenstorfer–LGC Standards (Augsburg, Germany) and from was equipped with a single taper liner (78.5 mm × 4 mm id;
CDN Isotopes (Quebec, Canada). Stock solutions were prepared Thermo Fisher Scientific). ALPHAGAZ™ helium (99.999%)
in a suitable solvent (mostly ethyl acetate or acetonitrile) at a was used as the carrier gas at a constant flow of 1.2 mL/min. GC
concentration of 1000–2000 mg/L and were stored in amber separation was performed on a 30 m × 0.25 mm id, 0.25 µm
screw-capped glass vials in the dark at −20°C to minimize TraceGOLD™ TG-5SilMS column with a 5 m safeguard
degradation. From the stock solution, a standard mix was (Thermo Fisher Scientific) using the following temperature
prepared in acetonitrile at 10 mg/L. Further dilutions were program: 40°C (1.5 min), up to 90°C at 25°C/min (held for
prepared in ethyl acetate and used for the calibration, as needed. 1.5 min), up to 180°C at 25°C/min, up to 280°C at 5°C/min, and
lastly, up to 300°C at 10°C/min (held for 5 min). The auxyliary
Extraction Procedure temperatures were kept at 250°C.
The Q Exactive™ Orbitrap mass spectrometer (Thermo Fisher
Samples were extracted using the well-known citrate QuEChERS Scientific) was operated with electron ionization at 70 eV. The
method (20) without cleanup. After homogenization, 10 g sample transfer line and the source temperature were maintained at 250°C.
was weighed into a 50 mL PTFE centrifuge tube. Next, 10 mL Full-scan MS acquisition was carried out in profile mode using an
acetonitrile and a 10 µL aliquot of the PISs (dichlorvos-D6, m/z range of 50–550 Da. Nitrogen (99.999%) was used for the
malathion-D10, and TPP at 10 mg/L in acetonitrile) were added C-trap and the higher-energy collisional dissociation cell. Resolving
to the sample and shaken in an automatic axial extractor power was set to 60 000 FWHM (at m/z 200), and the automatic
(AGYTAX®; Cirta Lab, Madrid, Spain) for 4 min. Afterward, 4 g gain control was set to 1e6 ions, as optimized in a previous work
magnesium sulfate, 1 g sodium chloride, 1 g trisodium citrate (18). The maximum ion injection time was set to “AUTO,”
dihydrate, and 0.5 g sodium hydrogencitrate sesquihydrate were corresponding to 119 ms for the selected resolving power. The
added, and the sample was shaken again in the automatic extractor for scan speed under these conditions was approximately 7.4 scans/s.
4 min. The extract was then centrifuged at 3700 rpm for 5 min. A External mass calibration was performed before each sequence
150 µL aliquot of the extract was evaporated to dryness with a using perfluorotributylamine (m/z: 68.9945, 99.9928, 130.9911,
nitrogen stream and recomposed with the same volume of ethyl 196.9827, 218.9846, 263.9860, 413.9760, and 501.9694), with a
acetate. Lastly, lindane-d6 (the IIS) was added to obtain a final mass error tolerance of ±1 ppm (±0.2 mDa). During the
concentration of 0.05 mg/L. Therefore, at the end of the procedure, measurement, internal mass calibration was carried out by the
1 mL extract represented a 1 g portion of sample. instrument using three background ions from the column bleed as
the lock mass (C5H15O3Si3+, 207.03236; C7H21O4Si4+, 281.05115;
GC-Orbitrap MS Analysis and C9H27O5Si5+, 355.06994), with a search window of ±2 ppm (±1
mDa). If, in a particular scan, none of the three specified background
A GC system consisting of a TriPlus 100 LS Liquid ions were found within ±2 ppm of their exact mass, then no internal
Autosampler and a TRACE 1310 Gas Chromatograph with a locking took place for that scan.
hot split/splitless injector was used for the analysis (Thermo The instrument was controlled using Tune 2.7, Xcalibur 4.0,
Fisher Scientific, Bremen, Germany), performing hot splitless and TraceFinder 4.0-EFS (Thermo Fisher Scientific) software.

Table 1. CAS number, retention time, and selected ions for the 15 pesticides analyzed

Compound Chemical Abstracts Service No. tR, mina Quantification ion, m/z Identification ion, m/z

Aldrin 309-00-2 14.88 262.85641 186.02308

Cadusafos 95465-99-9 11.38 158.96978 130.93848
Dichlorvos-D6 203645-53-8 8.11 133.05313 191.01416
Dieldrin 60-57-1 17.83 262.85641 276.87210
Disulfoton 298-04-4 12.66 88.03412 108.98715
Disulfoton-sulfone 2497-06-5 16.80 114.96133 213.01673
Disulfoton-sulfoxide 06-07-2497 8.66 114.96133 96.95076
Endrin 72-20-8 18.47 242.95296 173.01525
Ethoprophos 13194-48-4 10.92 157.96196 114.96133
Fipronil 120068-37-3 15.82 366.94300 212.94801
Fipronil-desulfinyl 205650-65-3 13.62 332.99609 387.97117
HCB 118-74-1 11.74 283.80960 248.84080
Heptachlor 76-44-8 13.96 271.80962 100.00743
trans-Heptachlor epoxide 28044-83-9 16.07 182.91657 236.84076
Lindane-d6 86194-41-4 12.21 184.96242 223.94242
Malathion D10 347841-48-9 14.41 132.07035 142.99263
Nitrofen 1836-75-5 18.40 202.01799 282.97975
Terbufos 13071-79-9 12.29 230.97315 174.91055
TPP 115-86-6 20.78 233.03620 326.07030
a
tR = Retention time.
 LOZANO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 101, NO. 2, 2018 377

Studied Pesticides deviation was set as a 30% window of the target ion ratio.
Calibration curves were fitted by weighted linear regression
A group of 15 pesticides was selected considering Commission (1/x), ignoring the origin and having individual residuals not
Directive No. 125/2006/EC of the European Parliament on MRLs deviate by more than ±20%.
for pesticides in baby foods for infants and young children (4).
These compounds are GC-amenable pesticides of high priority in Validation of the Analytical Procedure
laboratories analyzing baby food for EU or national monitoring
programs. Their structures can be seen in Figure 1. The following The proposed method has been tested to assess trueness (mean
are the established MRLs: cadusafos (0.006 mg/kg), ethoprophos recovery) and precision, LOQs, linearity range, and ME. The
(0.008 mg/kg), and fipronil (0.004 mg/kg; the residue definition is method validation procedure was carried out using spiked blank
the sum of fipronil and fipronil-desulfinyl, expressed as fipronil). matrixes from three different baby food matrixes (apple/pear,
Under the Commission Directive, there is also a list of compounds peach/banana, and six-fruit) bought in local markets in Almeria,
that cannot be present above 0.003 mg/kg (this concentration is Spain. Five test sample portions from each matrix were spiked
considered the limit): aldrin and dieldrin (expressed as dieldrin); with each pesticide at 0.003 and 0.006 mg/kg to determine the
disulfoton (sum of disulfoton, disulfoton-sulfoxide, and disulfoton- recovery and precision (repeatability) of the extraction method.
sulfone, expressed as disulfoton); endrin; heptachlor and trans- The method’s LOQ was defined as the lowest validation spike
heptachlor epoxide (expressed as heptachlor); HCB; nitrofen; level that could be quantified with acceptable trueness
and terbufos. (70–120%) and precision (RSDs <20%), as described in
SANTE/11945/2015 (21). The linearity of the calibration
Identification Criteria and Quantification Method curves was evaluated using matrix-matched calibration
solutions prepared by spiking blank extracts with the working
solution at seven concentration levels: 0.001, 0.003, 0.006, 0.01,
In the present study, the identification criteria followed were
those set in the SANTE/11945/2015 document (21). When working 0.05, 0.1, and 0.2 mg/L. MEs were assessed by comparing the
in full-scan accurate mass measurement, the identification slopes of the seven-point calibration curves from the matrix-
requirements are the following: two ions (mostly fragment ions) matched standards in the three matrixes tested with the slopes of
with mass accuracy ≤5 ppm (1 mDa at 200 m/z), an ion ratio the calibration curve of standards in solvent at the same
within ±30% (relative) of the calibration standard average from the concentration levels. MEs were calculated with the following
same sequence, and an S/N of the analyte’s chromatographic equation:
peaks ≥3. slope of calibration curve standard in matrix
For quantitation purposes, an in-house database containing the MEð%Þ = × 100
slope of calibration curve standard in solvent
selected 15 pesticides was used (see Table 1). Chromatographic
peaks were extracted from full-scan MS data, with a mass
tolerance of 1 mDa within a retention time window of 12 s
(0.1 min on each side of the peak) with respect to the retention Real Samples and Proficiency Test Sample
time in the database. For quantification, the base peak of the
spectrum was selected, and the following fragment ion, in terms To evaluate the developed method, 20 samples purchased in
of sensitivity, was chosen as the identification ion. The ion ratio different local supermarkets, comprising 10 different matrixes

Table 2. Identification parameters for the studied pesticides in peach/banana matrix at 0.003 mg/kg

Mass error, mDa

a
Compound tR difference, min Quantification ion Identification ion Ion ratio deviation, %

Aldrin −0.01 0.1879 −0.0240 −7.2

Cadusafos 0.00 0.0686 0.1339 −5.4
Dieldrin 0.00 0.1879 0.2755 −8.4
Disulfoton −0.01 0.0215 0.1318 −11.3
Disulfoton-sulfone 0.00 0.0653 −0.0521 −3.9
Disulfoton-sulfoxide 0.00 0.0729 0.0762 −16.9
Endrin −0.01 −0.0276 −0.0107 −16.2
Ethoprophos 0.00 0.0151 0.0653 14.7
Fipronil 0.00 −0.1594 −0.0120 2.0
Fipronil-desulfinyl −0.01 −0.1488 −0.1312 7.1
HCB 0.00 0.1534 0.0051 −6.1
Heptachlor 0.00 0.1639 0.0315 4.6
trans-Heptachlor epoxide −0.01 0.0756 −0.0966 −6.4
Nitrofen 0.00 −0.0304 −0.0442 3.1
Terbufos −0.01 0.0403 0.0945 −5.5
a
tR = Retention time.
378 LOZANO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 101, NO. 2, 2018

(apple/pear, peach/banana, six-fruit, strawberry/banana, three- solvent. For instance, during the validation of the method,
fruit, four-fruit, peach, seven-vegetable, vegetables/turkey, and all the mass error values were <0.8820 mDa and the highest
vegetables/chicken) were analyzed. Samples were homogenized difference in retention time was +0.03 min. Regarding the ion
and stored at −18°C until analysis. Sample preparation was ratio, when the reference was set as the average of the values for
performed as described in Extraction Procedure. The method standards in solvent, the ion ratio deviation varied from –28.1%
was also applied to the EUPT-FV-BF01 sample (peach baby (aldrin at 0.001 mg/kg) to 27.1% (disulfoton-sulfoxide at 0.003
food; 22), and the z-score values for the obtained results were mg/kg), remaining under the established 30% difference. In
then calculated. addition, the peak shapes of the two ions from eight of the
compounds are shown in Figure 2.
Results and Discussion
Recovery Experiments and Precision of the Extraction
Identification Method

Shifts in the retention time, the mass error of the two Recovery tests for the presented method, using high-resolution
fragments, and the ion ratio deviation were obtained MS in full-scan mode acquisition, were performed on 15 pesticides
automatically from the software (TraceFinder 4.0-EFS). These in three baby food matrixes (apple/pear, peach/banana, and six-
parameters were checked for all the samples (and for the PT and fruit) at 0.003 mg/kg (MRL) and at 0.006 mg/kg (2 × MRL), in
real samples during validation). In a batch, the differences in accordance with the European Commission guidelines for analytical
retention time used to be as maximum ±0.05 min and mass error QC and method validation procedures (21).
below ±0.9 mDa. As an example, Table 2 shows the retention Recoveries were within the 70–120% range for 96% of the
time difference, the mass errors for the two ions, and their ion cases (Table 3). These results are comparable to and (almost
ratio deviation for the 15 pesticides at 0.003 mg/kg in the peach/ always) slightly better than those obtained by Leandro et al. (10)
banana matrix. Even at such a low concentration level, the and González-Curbelo et al. (15), in which different extraction
retention time difference was below 0.01 min, the mass error methods (both based on acetonitrile) and analyzers were used. This
was within ±0.28 mDa, and the ion ratio deviation was within fact can be explained by the removal of the cleanup step.
±17% (fully complying with the requirements); this Compounds with recoveries outside this range were disulfoton
demonstrates the significant capabilities of the system in (at 0.003 in peach/banana) and disulfoton-sulfoxide (at 0.003
terms of sensitivity and mass accuracy. and 0.006 mg/kg in apple/pear; and at 0.003 mg/kg in the six-fruit
Furthermore, identification parameters were still met when matrix). This might be from the rapid decrease in the disulfoton
mixing all three matrixes evaluated plus the standards in content in ethyl acetate (the injection solvent) due to its thioether

Figure 2. Extracted-ion chromatograms for the two selected ions from eight of the pesticides in the peach/banana matrix at 0.003 mg/kg.
 LOZANO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 101, NO. 2, 2018 379

Table 3. Results of the pesticide recovery study (n = 5), obtained using two spiking levels in three matrixes, along with their
LOQs

Apple/pear Peach/banana Six-fruit

0.003 mg/kg spike 0.006 mg/kg spike 0.003 mg/kg spike 0.006 mg/kg spike 0.003 mg/kg spike 0.006 mg/kg spike
LOQ, LOQ,
Recovery, RSDr, Recovery, RSDr, LOQ, Recovery, RSDr, Recovery, RSDr, mg/ Recovery, RSDr, Recovery, RSDr, mg/
Compound % % % % mg/kg % % % % kg % % % % kg

Aldrin 83 3 88 5 0.003 82 3 90 2 0.003 79 4 82 7 0.003

Cadusafos 88 3 90 3 0.003 100 5 94 1 0.003 87 6 85 10 0.003
Dieldrin 82 4 89 7 0.003 88 5 89 4 0.003 86 4 84 12 0.003
Disulfoton 113 10 117 3 0.003 124 4 104 5 0.006 112 7 117 7 0.003
Disulfoton- 84 3 84 4 0.003 88 5 88 3 0.003 81 3 85 10 0.003
sulfone
Disulfoton- 56 14 67 6 >0.006 76 3 73 8 0.003 64 7 70 12 0.006
sulfoxide
Endrin 85 6 86 5 0.003 91 10 90 3 0.003 85 3 81 14 0.003
Ethoprophos 91 2 91 5 0.003 101 4 94 2 0.003 84 5 84 12 0.003
Fipronil 96 4 97 6 0.003 106 8 97 2 0.003 88 3 89 14 0.003
Fipronil- 89 3 89 3 0.003 102 7 91 3 0.003 84 3 83 14 0.003
desulfinyl
HCB 73 8 74 12 0.003 82 12 79 2 0.003 75 14 73 10 0.003
Heptachlor 84 2 86 4 0.003 83 2 89 2 0.003 77 6 81 6 0.003
trans- 82 3 86 5 0.003 89 3 87 3 0.003 87 2 84 11 0.003
Heptachlor
epoxide
Nitrofen 88 2 94 2 0.003 101 3 96 2 0.003 89 2 86 12 0.003
Terbufos 88 4 93 5 0.003 104 5 100 2 0.003 82 9 88 6 0.003

group being prone to oxidation and thermodegradation (23). In the three matrixes, the RSDr values (n = 5) were ≤14% for
Furthermore, some compounds containing sulfoxide groups are all the analytes at all levels, proving the good precision of the
difficult to analyze by GC because they are susceptible to extraction method. Repeatability data (as RSDr) are presented in
decomposition at elevated temperatures (24). Table 3.

Table 4. Linearity and MEs in three baby food matrixes

Solvent Apple/pear Peach/banana Six-fruit

Linear Linear Linear Linear

range, range, ME, range, ME, range, ME,
Compound mg/kg Slope R2 mg/kg Slope R2 %a mg/kg Slope R2 %a mg/kg Slope R2 %a

Aldrin 0.001–0.2 46492 1.0000 0.001–0.2 66495 0.9992 43 0.001–0.2 63796 0.9996 37 0.001–0.2 62418 0.9997 34
Cadusafos 0.001–0.2 62462 0.9998 0.001–0.2 153014 0.9989 145 0.001–0.2 134424 1.0000 115 0.001–0.2 131710 0.9995 111
Dieldrin 0.003–0.2 18813 0.9999 0.003–0.2 32910 0.9996 75 0.003–0.2 28142 0.9999 50 0.003–0.2 29548 0.9983 57
Disulfoton 0.003–0.2 5029 0.9988 0.001–0.2 60392 0.9991 1101 0.001–0.2 36892 0.9993 634 0.001–0.2 34633 0.9967 589
Disulfoton- 0.001–0.2 52188 1.0000 0.001–0.2 109437 0.9993 110 0.001–0.2 98044 0.9994 88 0.001–0.2 99986 0.9984 92
sulfone
Disulfoton- 0.001–0.2 52661 0.9995 0.001–0.2 115815 0.9997 120 0.001–0.2 145638 0.9985 177 0.001–0.2 155272 0.9998 195
sulfoxide
Endrin 0.003–0.2 12307 1.0000 0.003–0.2 21852 0.9995 78 0.003–0.2 20965 0.9994 70 0.003–0.2 20515 0.9999 67
Ethoprophos 0.001–0.2 51309 0.9996 0.001–0.2 132110 0.9979 157 0.001–0.2 114858 0.9999 124 0.001–0.2 124409 0.9985 142
Fipronil 0.001–0.2 18185 0.9998 0.001–0.2 43870 0.9992 141 0.001–0.2 38166 0.9993 110 0.001–0.2 35995 0.9991 98
Fipronil- 0.001–0.2 31940 0.9997 0.001–0.2 77395 0.9985 142 0.001–0.2 72127 0.9999 126 0.001–0.2 64779 0.9993 103
desulfinyl
HCB 0.001–0.2 190045 1.0000 0.001–0.2 248186 0.9990 31 0.001–0.2 232630 0.9998 22 0.001–0.2 242870 0.9989 28
Heptachlor 0.001–0.2 48059 1.0000 0.001–0.2 73280 0.9994 52 0.001–0.2 72292 0.9999 50 0.001–0.2 69012 0.9995 44
trans- 0.001–0.2 20744 0.9998 0.001–0.2 30049 0.9992 45 0.001–0.2 28397 0.9989 37 0.001–0.2 28540 0.9996 38
Heptachlor
epoxide
Nitrofen 0.001–0.2 17681 1.0000 0.001–0.2 38060 0.9991 115 0.001–0.2 35051 0.9998 98 0.001–0.2 33325 0.9981 88
Terbufos 0.001–0.2 58868 0.9998 0.001–0.2 159362 0.9992 171 0.001–0.2 136597 0.9999 132 0.001–0.2 132976 0.9993 126
a
Positive values of MEs mean enhancement of the signal. In all cases, residuals of the calibration curve were below 16%.
380 LOZANO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 101, NO. 2, 2018

Figure 3. Heptachlor extracted-ion chromatograms for (a) the standard at 0.01 mg/kg in the baby food matrix, and (b) the EUPT-FV-BF01
sample, along with their identification parameters. Concentration level found in the sample: 0.008 mg/kg. Calculated z-score: 0.

LOQ reduction in analyte ions reaching the detector. In contrast, if

matrix components are present in the analysis, these compounds
According to the DG SANTE procedures, the method LOQ is block the active sites and more analytes can pass through.
defined as the lowest spike level tested meeting the acceptable Continuous sample extract injections would cause new active
recovery and precision criteria. sites to form because of the deposition of low-volatile matrix
The LOQs were 0.003 mg/kg for 97% of the cases (Table 3). components in the GC system. These phenomena cause more
There were four pesticides with LOQs of 0.006 mg/kg (disulfoton signal enhancement in extracts containing matrix components
in the peach/banana matrix and disulfoton-sulfoxide in the six-fruit than in solvent solutions and lead to peak tailing and/or analyte
matrix), because their recoveries at 0.003 mg/kg were outside the losses.
acceptable range. It was not possible to obtain the LOQ for one An ME assessment can be carried out by comparing the slopes
pesticide, due to unacceptable recovery values at both concentration of the analyte calibration curves of standards in solvent with
levels (disulfoton-sulfoxide in apple/pear). For the above-mentioned those of standards in the matrixes under investigation. The ME
compounds, it was not possible to reach the stringent MRL set in the values are shown in Table 4. MEs within the 0–20% range are
Commission Directive. Nonetheless, the RSDr was always <15%. generally considered an insignificant enhancement. For those
analytes with medium (20–50%) or strong (50–100 and >100%;
Linearity and MEs 27) enhancement, it is necessary to use either matrix-matched
calibration standards or standard addition.
None of the evaluated pesticides had an insignificant ME
All the pesticides had standard calibration curves in all the
in any of the matrixes studied. Therefore, matrix-matched
matrixes tested, with residuals <16% and correlation coefficient
calibration standards were necessary. Signal enhancement was
values ≥0.9967.
obtained in every case, as expected with GC-MS. Most of the
The linear range started at 0.001 mg/L for most of the
compounds had ME values of >100% (i.e., nine, seven, and six
compounds, except for dieldrin and endrin in the three
compounds in the apple/pear, peach/banana, and six-fruit
matrixes, for which the linear range started at 0.003 mg/L
matrixes, respectively). Medium MEs were obtained for three,
because of lower sensitivity. For disulfoton in solvent, the
five, and four analytes in the apple/pear, peach/banana, and six-
linear range also started at 0.003 mg/L, but not in the matrix,
fruit matrixes, respectively. The rest of the pesticides (i.e., three,
due to the signal enhancement for the compound. This is proof of
three, and five compounds in the apple/pear, peach/banana, and
the high sensitivity of the analytical method, allowing the low
six-fruit matrixes, respectively) had an ME value in the range of
levels established for baby food in the Commission Directive to
50–100%.
be achieved. For all of these pesticides, the response was linear
up to 0.2 mg/L (Table 4).
MEs in GC-MS occur because of undesired interactions with QCs
active sites in the inlet and the column (25, 26). When standards
in solvent are analyzed, the analytes bind to the active sites Four internal standards were added at different stages of the
(mainly silanol groups) on the liner or column. Hence, there is a process in order to check the correct performance of the analytical
 LOZANO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 101, NO. 2, 2018 381

procedure. The whole procedure, from the extraction step onward, The proposed analytical methodology allowed the
was controlled using dichlorvos-D6, malathion-D10, and TPP as quantification of 15 pesticides in fruit- and vegetable-based
the PISs. The procedure was considered to be performing properly baby foods. The results showed that the sensitivity of GC-Q-
when the recovery for the PISs was within the 70–120% range. If Orbitrap MS is appropriate for the required levels, including a
the values were outside the acceptable range, the extractions were suitable linear range, which enables unambiguous identification
repeated. The injection was also controlled by checking the peak (due to accurate mass analysis). The low LOQ values obtained
areas of lindane-d6 (the IIS). The sample extract was considered permitted the reliable quantification of baby foods at the stringent
correctly injected if the peak areas of the IIS did not deviate more MRL levels established by the European Commission, with the
than ±30% from the average area of the calibration standards. If exception of two compounds (disulfoton and disulfoton-
just one individual sample was above this value, it was sulfoxide). Method precision was always below the acceptable
considered a miss injection, and a correction was made. If criterion (experimental values were <15%). Quantification
there were more deviations, the whole sample set was reanalyzed. capabilities were proved by analyzing a PT sample, obtaining
The average recoveries for dichlorvos-D6, malathion-D10, z-scores <1. No pesticides from the scope were found in the real
and TPP were 76, 103, and 104%, with RSDs of 17, 9, and 9%, samples.
respectively, for the PT and real sample analyses. The IIS
deviation was within ±10% for each individual sequence. References

Real Samples and PT Sample (1) National Research Council (US) Committee on Pesticides in the
Diets of Infants and Children (1993) in Pesticides in the Diets of
To prove the effectiveness of the validated method and its Infants and Children, B.J. Rice, & J. Overton (Eds), National
suitability for routine analysis, it was applied to real samples as Academy Press, Washington, DC, pp 49–126
part of a survey. For the identification and quantification of the (2) Yang, A., Abd El-Aty, A.M., Park, J.-H., Goudah, A.,
positive findings, the parameters described in Identification Rahman, M.M., Do, J.-A., Choi, O.-J., & Shim, J.-H. (2014)
Criteria and Quantification Method were followed. A total of Biomed. Chromatogr. 28, 735–741. doi:10.1002/bmc.3092
20 real samples were analyzed with the present method, resulting (3) Radford, S.A., Panuwet, P., Hunter, R.E. Jr, Barr, D.B., &
Ryan, P.B. (2014) J. Agric. Food Chem. 62, 7085–7091.
in 100% of the samples able to be considered blanks (the
doi:10.1021/jf500779a
presence of no compound from the target list). Other authors (4) Commission Directive No. 2006/125/EC (2006) Off. J. Eur.
have also analyzed real samples (13–46 samples each); in half of Comm. L339, 16–35
these surveys, no tested pesticides were found (2, 8, 14), as was (5) European Food Safety Authority (2016) EFSA Journal 14, 4611
the case here. However, Gilbert-López et al. (12) found (6) Cajka, T., Hajslova, J., Lacina, O., Mastovska, K., &
carbendazim, thiabendazole, and imazalil at concentration Lehotay, S.J. (2008) J. Chromatogr. A 1186, 281–294.
levels lower than their MRLs. Radford et al. (3) detected at doi:10.1016/j.chroma.2007.12.009
least three degradation products in the 21 samples analyzed, and (7) Przybylski, C., & Segard, C. (2009) J. Sep. Sci. 32, 1858–1867.
Petrarca et al. (13) found procymidone in one sample at 0.069 mg/kg. doi:10.1002/jssc.200900016
Such results highlight the importance of continuing the (8) Gómez-Pérez, M.L., Romero-González, R., José Luis
Martı́nez, V., & Garrido Frenich, A. (2015) Talanta 131, 1–7.
monitoring of baby food samples.
doi:10.1016/j.talanta.2014.07.066
Furthermore, a PT sample was analyzed with the presented (9) Pérez-Ortega, P., Lara-Ortega, F.J., Gilbert-López, B., Moreno-
procedure. Of the 17 pesticides present in the test material, three González, D., Garcı́a-Reyes, J.F., & Molina-Dı́az, A. (2016)
were in the validated method (fipronil, HCB, and heptachlor). Food Anal. Methods 10, 1216–1244. doi:10.1007/s12161-016-
These three compounds were detected in the PT sample, with 0678-0
fipronil found at 0.012 mg/kg, HCB at 0.010 mg/kg, and (10) Leandro, C.C., Fussell, R.J., & Keely, B.J. (2005)
heptachlor at 0.008 mg/kg. The concentration levels obtained J. Chromatogr. A 1085, 207–212. doi:10.1016/j.
were very similar to the assigned values reported in the PT, as can be chroma.2005.06.054
observed with the z-score values: 0.8 for fipronil, 0.5 for HCB, and (11) Georgakopoulos, P., Zachari, R., Mataragas, M.,
0 for heptachlor (see Figure 3). All these values are considered Athanasopoulos, P., Drosinos, E.H., & Skandamis, P.N. (2011)
Food Chem. 128, 536–542. doi:10.1016/j.
acceptable (z-score values ≤2). In addition, a target screening
foodchem.2011.03.042
search was performed for the sample using an in-house database (12) Gilbert-López, B., Garcı́a-Reyes, J., Ortega-Barrales, P.,
developed in a previous work (18), which indicated seven more Molina-Dı́az, A., & Fernández-Alba, A. (2007) Rapid Commun.
detections: cypermethrin, cyprodinil, deltamethrin, etofenprox, Mass Spectrom. 21, 2059–2071. doi:10.1002/rcm.3062
iprodione, malathion, and tebuconazole. It is important to mention (13) Petrarca, M.H., Fernandes, J.O., Godoy, H.T., & Cunha, S.C.
that all of these were present in the PT material. (2016) Food Chem. 212, 528–536. doi:10.1016/j.
foodchem.2016.06.010
Conclusions (14) Cieślik, E., Sadowska-Rociek, A., Molina Ruiz, J.M., & Surma-
Zadora, M. (2011) Food Chem. 125, 773–778. doi:10.1016/j.
A study to evaluate the usefulness of GC-Q-Orbitrap MS foodchem.2010.09.019
was carried out, working in full-scan mode for the low-level (15) González-Curbelo, M.Á., Asensio-Ramos, M., Herrera-
Herrera, A.V., & Hernández-Borges, J. (2012) Anal. Bioanal.
(<0.01 mg/kg) quantitative analysis of pesticides in baby foods.
Chem. 404, 183–196. doi:10.1007/s00216-012-6103-7
It was proved that this method can be used as an accurate and (16) Picó, Y., Font, G., Ruiz, M.J., & Fernández, M. (2006) Mass
quantitative technique for routine analysis. The acquisition mode Spectrom. Rev. 25, 917–960. doi:10.1002/mas.20096
allows nontarget analysis using databases and/or libraries, as (17) Ferrer, I., Garcı́a-Reyes, J.F., & Fernandez-Alba, A. (2005)
well as retrospective analysis. Moreover, the scope of the TrAC, Trends Anal. Chem. 24, 671–682. doi:10.1016/j.
quantitation method can be rapidly and easily increased. trac.2005.04.004
382 LOZANO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 101, NO. 2, 2018

(18) Uclés, S., Uclés, A., Lozano, A., Martı́nez Bueno, M.J., & (22) EURL for Fruits and Vegetables (2016) EUPT-FV-BF01, http://
Fernández-Alba, A.R., (2017) J. Chromatogr. A 1501, 107–116. www.eurl-pesticides.eu/docs/public/tmplt_article.asp?LabID=
doi:10.1016/j.chroma.2017.04.025 500&CntID=1038&Theme_ID=1&Pdf=False&Lang=EN
(19) Rajski, Ł., Gómez-Ramos, M.M., & Fernández-Alba, A.R. (23) Maštovská, K., & Lehotay, S.J. (2004) J. Chromatogr. A 1040,
(2014) J. Chromatogr. A 1360, 119–127. doi:10.1016/j. 259–272. doi:10.1016/j.chroma.2004.04.017
chroma.2014.07.061 (24) Hill, A.R.C., Wilkins, J.P.G., Findlay, N.R.I., & Lontay, K.E.M.
(20) European Committee for Standardization (2008) CEN EN (1984) Analyst 109, 483–487. doi:10.1039/an9840900483
15662: Foods of plant origin-determination of pesticide residues (25) Banerjee, K., & Utture, S. (2015) in Mass Spectrometry for the
using GC-MS and/or LC-MS/MS following acetonitrile Analysis of Pesticide Residues and Their Metabolites, D. Tsipi,
extraction/partition and clean-up by dispersive SPE-QuEChERS- H. Botitsi, & A. Economou (Eds), John Wiley & Sons, Hoboken,
method, https://www.en-standard.eu/csn-en-15662-foods- NJ, pp 91–112
of-plant-origin-determination-of-pesticide-residues-using- (26) Agüera, A., & de Kok, A. (2004) in Chromatographic-Mass
gc-ms-and-or-lc-ms-ms-following-acetonitrile-extraction- Spectrometric Food Analysis for Trace Determination of
partitioning-and-clean-up-by-dispersive-spe-quechers- Pesticide Residues, A.R. Fernandez-Alba (Ed.), Elsevier Science,
method/ Amsterdam, The Netherlands, pp 339–368
(21) SANTE/11945/2015, Guidance document on analytical quality (27) Uclés, S., Belmonte, N., Mezcua, M., Martı́nez, A.B.,
control and method validation procedures for pesticide residues Martinez-Bueno, M.J., Gamón, M., & Fernández-Alba, A.R.
analysis in food and feed, http://ec.europa.eu/food/plant/docs/ (2014) J. Environ. Sci. Health B 49, 557–568. doi:10.1080/
plant_pesticides_mrl_guidelines_wrkdoc_11945_en.pdf 03601234.2014.911566