Analytica Chimica Acta 965 (2017) 72e82

Contents lists available at ScienceDirect

Analytica Chimica Acta

journal homepage: www.elsevier.com/locate/aca

Comprehensive characterization of ethoxyquin transformation

products in sh feed by traveling-wave ion mobility spectrometry
coupled to quadrupole time-of-ight mass spectrometry
Noelia Negreira a, b, *, Jorge Regueiro a, Stig Valdersnes a, Marc H.G. Berntssen a,
Robin rnsrud a
a
National Institute of Nutrition and Seafood Research (NIFES), PO Box 2029 Nordnes, N-5817 Bergen, Norway
b
Institute for Food Analysis and Research (IIAA), Department of Analytical Chemistry, Nutrition and Food Sciences, University of Santiago de Compostela,
Constantino Candeira S/N, 15782 Santiago de Compostela, Spain

h i g h l i g h t s g r a p h i c a l a b s t r a c t

Screening of ethoxyquin TPs in sh

feed by UHPLC-TWIMS-QTOFMS.

TWIMS highly improved method
selectivity.

21 TPs identied for the rst time in
sh feed.

High dimerization of ethoxyquin
with other compounds.

A comprehensive database for
ethoxquin TPs is provided.

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

Article history: Feed additives are typically used in intensive farming production over long periods, and hence, they can
Received 24 November 2016 accumulate in farmed animal tissues. Concerns regarding the use of ethoxyquin as an antioxidant feed
Received in revised form additive, have recently arisen due to its potential conversion into a series of transformation products
27 January 2017
(TPs).
Accepted 3 February 2017
Available online 20 February 2017
The aim of this work was to characterize the TPs of ethoxyquin in sh feed by a novel approach based
on the use of traveling-wave ion mobility spectrometry (TWIMS) coupled to high-resolution quadrupole
time-of-ight mass spectrometry (QTOFMS). First, ethoxyquin was oxidized under controlled conditions
Keywords:
Feed additives
and the generated TPs were added to a comprehensive database. Atlantic salmon feeds were then
High-resolution mass spectrometry screened for ethoxyquin TPs using both targeted and untargeted approaches.
Oxidation products Twenty-seven TPs were tentatively identied during the oxidation experiments, fteen of them also
Screening being present in the feed samples. In addition, ten other potential TPs were detected in sh feed
Traveling-wave ion mobility spectrometry following the untargeted approach. Thirty-one of these TPs have been reported for the rst time in this
Untargeted analysis work through the oxidation experiments and the feed samples. Therefore, this study provides valuable
information on the oxidative fate of ethoxyquin in feed, which can be used for future evaluations of
potential risk related to this additive.
2017 Elsevier B.V. All rights reserved.

* Corresponding author. National Institute of Nutrition and Seafood Research 1. Introduction

(NIFES), PO Box 2029 Nordnes, N-5817 Bergen, Norway.
E-mail address: noelia.negreira@outlook.com (N. Negreira). Ethoxyquin (EQ) is a quinoline-based synthetic antioxidant
URL: http://www.nifes.no

http://dx.doi.org/10.1016/j.aca.2017.02.021
0003-2670/ 2017 Elsevier B.V. All rights reserved.
 N. Negreira et al. / Analytica Chimica Acta 965 (2017) 72e82
globally used in animal feeds to prevent lipid peroxidation and to
stabilize fat-soluble vitamins, but also as a preservative in dried
forage crops, as an antiscald agent in pears and apples, and as a
color preservative in some spices [1]. In addition, it is extensively
used in sh meal, a major ingredient of many sh feed formula-
tions, to prevent self-ignition during shipping and storage [2].
In the EU, EQ is authorized in feed applications with a maximum
content of 150 mg/kg complete feed for all farmed animal species
[3]. However, its use has been shown to lead to residues in products
of animal origin intended for human consumption such as poultry
meat, eggs and farmed sh [4e7]. EQ concentrations up to 0.17 mg/
kg have been reported by Lundebye et al. [7] in farmed Atlantic
salmon llets.
Because of its antioxidant nature, EQ can be oxidized into a
series of transformation products (TPs), some of which also exhibit
antioxidant properties themselves [8]. Among them, the EQ 1,80 -
dimer (1,80 -EQDM) has been reported as the major TP of EQ in sh
muscle, being present usually at much higher concentrations than
the parent compound [6,7]. However, other TPs such as EQ
quinone-imine (EQI) have also been reported in sh [6,7]. These TPs
can also be found in sh meal and sh feed [1,9]. Additional TPs,
including demethylethoxyquin (DMEQ), dehydrodemethylethox-
yquin (DHMEQ), dihydroethoxyquin (DHEQ) and the EQ N,N-dimer
(N,N0 -EQDM) have been reported in pears following the postharvest
application of EQ [10].
Some investigations indicated that excessive intake of EQ may
cause negative health effects, including liver and kidney damage, in
animal experiments dealing with toxic effects of EQ [11,12]. In
addition, in vitro studies have shown that EQ induces chromosomal
aberrations in human lymphocytes [13]. According to a recent Eu-
ropean Food Safety Authority (EFSA) scientic opinion [14], EQI
shows structural alerts for mutagenicity, carcinogenicity and DNA
binding, whereas the toxicological prole of 1,80 -EQDM is consid-
ered to reect that of EQ [15].
As most of marine sh farms culture the sh in sea cages, any
feed additive may be discharged into the open water and sediments
in the vicinity of the farms [16]. The EU regulations on authoriza-
tion of feed additives states that degradation and decomposition
products of feed additives should be characterized and the poten-
tial risk for the consumers of food containing residues of the ad-
ditive or its metabolites should be investigated [17]. In addition, the
recent EFSA opinion also highlighted the need for data to evaluate
the environmental impact of EQ and its TPs from sh feed and feces
[14].
Although some work has been done to characterize the TPs of
EQ, it seems very likely that unreported TPs may be formed. For
instance, He and Ackman [18] observed that the sum of EQI and
1,80 -EQDM did not explain the EQ decrease in sh meals and sh
feeds during storage, suggesting the existence of unknown degra-
dation routes.
In this regard, the recent hyphenation of traveling-wave ion
mobility spectrometry (TWIMS) with high-resolution mass spec-
trometry (HRMS) has arisen as a promising tool for both targeted
and untargeted screening approaches [19e21]. On one hand, HRMS
instruments operating in full-scan mode allow screening for a
virtually unlimited number of compounds, providing information
on their molecular formulae due to its high resolving power and
mass accuracy [22e24]. In addition, hybrid MS analyzers such as
the quadrupole-time-of-ight (QTOF) allow obtaining fragment ion
information, which can be used for structural elucidation of the
detected species. On the other hand, TWIMS separates ionized
molecules based on their mobilities in the gas phase under the
inuence of a weak electric eld. A TWIMS device comprises a
series of ring electrodes lled with an inert gas such as nitrogen.
Radio-frequency (RF) voltages of opposite phases are applied to
73
adjacent ring electrodes to provide radial ion connement, while a
transient direct current (DC) voltage is successively applied to the
electrodes creating a traveling-wave upon which ions surf and
traverse the mobility cell [20,25]. Higher-mobility ions are carried
by the wave, whereas ions of lower mobility roll over the wave,
thus taking longer to move through the gas-lled mobility cell.
As a result, ions are separated based on their mobility through
the gas. From the time an ion takes to travel through the TWIMS cell
(i.e., the drift time), its collision cross section (CCS) can be derived
[26]. CCS is a unique physicochemical parameter related to the ion
size, shape and charge (z), and unlike retention time, it is inde-
pendent of the used chromatographic conditions and not affected
by the sample matrices [27]. Consequently, the use of CCS in
addition to retention time and HRMS information offers the op-
portunity to further improve structural elucidation [28,29], in-
crease peak capacity [26,30] and separate isomeric species [21,31].
In the present work, the oxidative fate of EQ was thoroughly
investigated by a novel approach based on ultra-high performance
liquid chromatography (UHPLC) coupled to hybrid TWIMS-
QTOFMS. In a rst stage, EQ TPs were generated by two common
in vitro antioxidant activity tests, namely the hydrogen peroxide
(H2O2) assay and the free radical 2,2-diphenyl-1-picrylhydrazyl
(DPPH) assay. The resulting TPs were characterized by UHPLC-
TWIMS-QTOFMS, and a database including retention times, accu-
rate masses, characteristic fragment ions and CCS values was
established. The presence of many of these TPs in Atlantic salmon
feeds was conrmed using a targeted screening approach. In
addition, several TPs not formed during the antioxidant activity
assays were detected in feed following an untargeted screening
strategy.
2. Experimental section
2.1. Reagents, standards and materials
HPLC grade acetonitrile for sample extraction was obtained from
Sigma-Aldrich (Steinheim, Germany), whereas LCMS grade meth-
anol and acetonitrile were purchased from Merck (Darmstadt,
Germany). Ultrapure water (18.2 MU cm) was produced using a
Milli-Q Gradient water purication system from Merck. LCMS grade
ammonium acetate and acetic acid were obtained from VWR In-
ternational (Oslo, Norway), and formic acid (98e100%) was pur-
chased from Merck. DPPH (95%) was acquired from Alfa Aesar
(Karlsruhe, Germany) and H2O2 solution (30%, w/w) was obtained
from Merck.
EQ (97.8%) was purchased from Sigma-Aldrich, whereas 1,80 -
EQDM (99.2%) and EQI (98.6%) were obtained from HPC Standards
(Cunnersdorf, Germany). 6-Ethoxy-2,2,4-trimethyl-quinolin-8-one
was kindly provided by Dr. F. Thrun from BASF (Ludwigshafen,
Germany). Individual stock solutions (ca. 1000 mg L1) were pre-
pared in acetonitrile and stored in amber glass vials at 25  C.
Different working standard solutions were freshly prepared on the
day of the experiments by appropriate dilution in methanol and/or
ultrapure water. Leucine-enkephalin, used as lock mass standard,
and the instrument calibration solution (CCS Major Mix) were
purchased from Waters (Manchester, UK).
Marvin software was used for drawing and characterizing
chemical structures (v. 16.5.2.0, ChemAxon, Budapest, Hungary).
2.2. Oxidation assays
Oxidation of EQ was performed by two in vitro antioxidant ac-
tivity tests, namely the H2O2 assay [32] and the stable free radical
DPPH assay [33]. For the H2O2 oxidation, 10 mL of 5 M H2O2 aqueous
solution were added to an autosampler vial containing 1 mL of
74 N. Negreira et al. / Analytica Chimica Acta 965 (2017) 72e82
5 mg L1 EQ (23 mM) in water at 25  C. For the DPPH assay, 10 mL of
5 mM DPPH solution in methanol were added to an autosampler
vial containing 1 mL of 5 mg L1 EQ (23 mM) in methanol at 25  C.
Aliquots of both reaction mixtures were injected into the UHPLC-
TWIMS-QTOFMS system at different reaction times to follow the
degradation of EQ and the generation of the corresponding TPs. As a
control, the stability of EQ in water and in methanol at 25  C was
investigated up to 24 h.
2.3. Samples
Commercial Atlantic salmon feed samples were collected by
inspectors from the Norwegian Food Safety Authority (NFSA) from
different feed factories and analyzed by NIFES in the Norwegian
National surveillance program on sh feed conducted by NIFES on
behalf of the NFSA. Samples were collected between July and
December 2015 and then stored at 4  C, protected from light, until
their analysis in July 2016.
2.4. Sample preparation
Feed samples were homogenized at 2000 RPM using a stainless
steel blender GRINDOMIX GM 300 from Retsch (Haan, Germany)
and then extracted following a previously reported procedure
aimed at compounds with a broad range of polarities [21]. Briey, a
sample aliquot (2.5 g) was extracted with 5 mL of acetonitrile/
water/formic acid (75:24:1, v/v/v) by vortex shaking (1 h, 2500
RPM) in a multi-tube vortexer BenchMixer XL (Benchmark Scien-
tic, NJ, USA). The extract was centrifuged for 5 min at 3000 RCF
(Eppendorf 5810R, Eppendorf, Hamburg, Germany). The superna-
tant was collected and stored at 25  C for 3 h in order to precip-
itate lipids and other macromolecules. An aliquot (0.5 mL) was
transferred into a 2 mL polypropylene microcentrifuge tube and
diluted with 0.5 mL of 0.1% formic acid in water (v/v). After vortex
shaking for 20 s, the tube was centrifuged for 5 min at 18000 RCF
(Eppendorf 5427R, Eppendorf) and the supernatant was ltered by
a regenerated cellulose membrane syringe lters (13 mm, 0.2 mm)
from Whatman (Maidstone, UK).
2.5. UHPLC separation
UHPLC was performed on ACQUITY UPLC I-Class system from
Waters, consisting of a binary pump, a vacuum degasser, an auto-
sampler and a column oven. EQ and its TPs were chromatograph-
ically separated using a column ACQUITY UPLC BEH C18
(100  2.1 mm, 1.7 mm) also from Waters, maintained at 45  C. A
linear binary gradient of water (mobile phase A) and methanol
(mobile phase B), both containing ammonium acetate (10 mM, pH
5.0), was used. The mobile phase composition was changed during
the run as follows: 0 min, 2% B; 0.25 min, 2% B; 12.25 min, 99% B;
13.00 min, 99% B; 13.01 min, 2% B, and 17.00 min, 2% B. The ow rate
was set to 0.45 mL min1 and the injection volume was 5 mL.
2.6. TWIMS-QTOFMS analysis
The UHPLC system was coupled to a hybrid mass spectrometer
Vion IMS QTOF from Waters, equipped with a LockSpray ion source
and a novel TWIMS separation cell [21]. The ion source was oper-
ated in positive electrospray ionization (ESI) mode under the
following specic conditions: capillary voltage, 0.45 kV; reference
capillary voltage, 3.00 kV; cone voltage, 10 V; source offset, 80 V;
source temperature, 110  C; desolvation gas temperature, 450  C;
desolvation gas ow, 900 L/h, and cone gas ow, 40 L/h. Nitrogen
(>99.5%) was employed as desolvation and cone gas.
For the TWIMS separation, the following settings were applied:
trap bias, 40 V; stopper height, 40 V; gate height, 40 V; trap wave
velocity, 100 m/s; trap pulse height A, 20 V; trap pulse height B, 5 V,
IMS wave velocity, 250 m/s; IMS wave height, 45V; gate release,
2 ms, and nitrogen (>99.5%) as trap and IMS buffer gas at
1.6 L min1 and 25 mL min1, respectively. CCS calibration was
performed regularly using a commercial mixture of calibrants
prepared in acetonitrile/water/formic acid (50:49.9:0.1, v/v/v).
Data were acquired in high-denition MSE (HDMSE) mode in the
range m/z 50e1000 at 0.2 s/scan. Thus, two independent scans with
different collision energies (CE) were alternatively acquired during
the run: a low-energy (LE) scan at a xed CE of 4 eV, and a high-
energy (HE) scan where the CE was ramped from 8 eV to 45 eV.
Targeted MS/MS experiments were also performed on some
selected precursor ions using different CE (10e45 eV).
Argon (99.999%) was used as collision-induced dissociation
(CID) gas. The TOF was operated in the sensitivity mode, providing
an average resolving power (m/Dm) of approximately 40000 full
width at half maximum (FWHM) in the scanned m/z range. Cali-
bration of the TOF analyzer was carried out regularly by infusion
(10 mL min1) of the corresponding calibration solution according
to the manufacturer's guidelines. For real time accurate-mass
recalibration, a 50 mg L1 standard solution of leucine-enkephalin
in acetonitrile/water/formic acid (50:49.9:0.1, v/v/v) was continu-
ously infused (5 mL min1) through the reference probe and scan-
ned every 30 s. Dual point correction upon leucine-enkephalin CID
fragmentation at 25 eV was conducted using the product ions at m/
z 397.1870 (C21H25N4O
4 ) and m/z 120.0808 (C8H10N ). Data acqui-
sition and evaluation was performed with UNIFI software (v. 1.8,
Waters). The instrumental limit of detection for EQ, dened for a
signal-to-noise ratio of three, was 0.0015 mg L1. Relative standard
deviation (RSD) for EQ ranged from 3.6% at 0.005 mg L1 to 1.3% at
5 mg L1.
3. Results and discussion
3.1. Oxidative degradation of EQ
Due to the antioxidant nature of EQ, an approach based on the
use of scavenging assays was employed to force the degradation of
EQ and generate a series of oxidation TPs. To this end, two common
in vitro antioxidant capacity tests were selected, namely the H2O2
assay and the DPPH assay. The former is normally used to deter-
mine non-radical scavenging activities, whereas the DPPH assay is
widely used to evaluate the capacity to scavenge free radicals [33].
Similar approaches have been used previously to generate oxida-
tion TPs of natural antioxidants [34,35].
The concentrations of H2O2 and DPPH to be used were rst
investigated aiming at the complete degradation of EQ. Thus, the
oxidation reaction was performed at different concentrations of
H2O2 (0.5, 5, 15 and 50 mM) and DPPH (25, 50 and 100 mM) while
keeping the initial EQ concentration (23 mM) constant. Aliquots of
the reaction mixtures were analyzed by UHPLC-TWIMS-QTOFMS at
different reaction times (0, 0.7, 5, 20, 45 and 60 min) in order to
obtain the time courses of the oxidation with these reagents. The
remaining EQ was expressed as the percentage ratio between the
compound peak area at each reaction time and the peak area at
zero time. To verify that the concentration of EQ did not decrease
due to other factors, control experiments were carried out without
adding H2O2 nor DPPH. As shown in Table S-1 (Supporting infor-
mation), EQ was completely degraded at 50 mM of H2O2 after
45 min. When the oxidation was performed with the free radical
DPPH, the degradation was much faster, and EQ completely dis-
appeared after just 5 min at 50 mM. Therefore, 50 mM of H2O2 and
50 mM of DPPH were nally selected as the optimal concentrations
for subsequent experiments.
 N. Negreira et al. / Analytica Chimica Acta 965 (2017) 72e82 75

3.2. Identication of ethoxyquin transformation products
Once the degradation of EQ under the oxidation conditions was
proven, our work focused on the screening and identication of all
potential TPs generated in these reactions. For this purpose, the use
of UHPLC-TWIMS-QTOFMS highly increases the selectivity of the
analysis over traditional MS-based approaches, improving the
condence in the identication. Thus, prior to detection, com-
pounds were separated by retention time, drift time and m/z,
yielding four-dimensional (4D) data for all analytes (retention time,
drift time, m/z and intensity). Furthermore, information on frag-
ment ions was obtained from the HE scans and from the MS/MS
experiments.
UNIFI Software's binary comparison feature was rst used in
order to automatically compare the chromatographic and spectral
data of the oxidation experiments at different reaction times with
the control samples at zero time. Signicant differences were easily
observed using the base peak ion (BPI) chromatogram compare
functionality. In addition, the software generated a list of 4D-peaks
that were present uniquely in the reaction mixtures with H2O2 and
DPPH. Each compound was then individually investigated using all
the available information in an attempt to assign its structural
identity.
Following this approach, twenty-seven potential EQ TPs were
detected in H2O2 and/or DPPH. Their molecular formulae were
proposed based on the accurate mass measurements and the
observed isotopic patterns. Corresponding retention times, accu-
rate measured m/z, proposed molecular formulae, relative mass
measurement errors (Dm), characteristic fragment ions, drift times
and CCS values for all the potential TPs are summarized in Table 1.
As shown, relative mass measurement errors for all the proposed
molecular formulae remained below 3 ppm.
Among them, the identities of four compounds were consistent
with previously reported EQ TPs and/or metabolites, namely EQI,
deethylated EQ (DEQ), DHMEQ and 1,80 -EQDM. EQI and 1,80 -EQDM
were positively conrmed by injection of the corresponding com-
mercial standards. DEQ and DHMEQ were identied on the basis of
their accurate masses and fragment ions, which were in agreement
with those previously reported in the literature [1,10].
Tentative structures were postulated for all the other TPs based
on their molecular formulae and the interpretation of the fragment
ions from HE scans. In this regard, the use of TWIMS in combination
with data independent analysis (DIA) acquisition modes such as
MSE allows the drift time alignment of precursor and product ions,
which highly simplies their correct assignation without the need
for further targeted MS/MS experiments. As the CID fragmentation
occurs after the TWIMS separation, fragment ions should present
the same drift time as the corresponding precursor ion. Thus, low-
and high-energy spectra are both retention time- and drift time-
aligned to lter out fragments that do not match the precursor's
drift time. Very recently, Regueiro et al. [21] demonstrated the
usability of this approach to improve the mass spectral interpre-
tation of pesticides when working with HDMSE mode.
In addition, the MS/MS spectrum of EQ was considered as a basis
for the structural elucidation of its TPs due to possible common
product ions and/or similar fragmentation patterns (Fig. S-1, Sup-
porting information). Thus, EQ showed a [MH] ion at m/z
218.1541, which mainly fragmented into m/z 202.1225, m/z
188.1070, m/z 160.0757 and m/z 148.0757, as previously described
by Pico et al. [10]. The fragment ion at m/z 202.1225 corresponded
to a demethylation and subsequent formation of a double bond,
whereas the ion at m/z 188.1070 might be derived from the loss of a
methyl radical from the former ion. The fragment at m/z 160.0757
would be the result of the loss of an ethyl group and a methyl group
from the fragment ion at m/z 202.1225. The CCS obtained for this
compound was 155.8 2, which is in agreement with previously
reported CCS values [21].
TP-222A (m/z 222.1127), TP-222B (m/z 222.1125) and TP-222C

Table 1
Retention times (tR), accurate m/z, molecular formula, mass measurement errors (Dm), drift time values, collision cross section (CCS) values and characteristic fragment ions of
EQ and its potential TPs generated during the oxidation experiments.

Compound tR (min) Observed m/z Molecular formula Dm (ppm) Drift time (ms) CCS (2) Characteristic fragment ions (m/z)

EQ 8.99 218.1541 [C14H20NO] 0.74 4.48 155.8 204.1383, 202.1225, 190.1228, 188.1070, 174.0912, 160.0757, 148.0757
TP-222A 3.05 222.1127 [C12H16NO3] 0.97 4.24 149.7 204.1022, 190.0867, 150.0917
TP-222B 3.82 222.1125 [C12H16NO3] 0.35 4.19 148.7 204.1018, 190.0868, 150.0911
TP-222C 3.97 222.1125 [C12H16NO3] 0.35 4.17 148.3 150.0550
TP-234A 5.60 234.1491 [C14H20NO2] 0.89 4.56 157.2 216.1383, 188.1072, 173.0832, 148.0754
TP-234B 5.86 234.1492 [C14H20NO2] 1.32 4.57 157.3 216.1380, 188.1069, 173.0829, 148.0753
TP-204 6.06 204.1024 [C12H14NO2] 2.43 3.97 144.0 186.0921, 172.0762, 160.0762
TP-206 6.12 206.1174 [C12H16NO2] 0.90 3.99 144.3 NA
TP-232A 6.49 232.1332 [C14H18NO2] 2.4 4.52 156.4 204.1379, 176.1071, 160.0734
TP-218A 6.55 218.1177 [C13H16NO2] 0.62 4.28 151.1 203.0942, 160.0761
TP-236 6.68 236.1284 [C13H18NO3] 1.02 4.44 154.4 NA
EQI 6.73 188.1074 [C12H14NO] 2.04 3.80 140.6 173.0832, 145.0880, 130.0652
DEQ 6.74 190.1228 [C12H16NO] 0.60 3.98 144.7 188.1071, 174.0900, 145.0880
TP-250 6.86 250.1439 [C14H20NO3] 0.45 4.78 162.2 162.0914
TP-218B 6.83 218.1179 [C13H16NO2] 1.39 4.24 150.0 202.0863
TP-218C 7.26 218.1177 [C13H16NO2] 0.59 4.15 147.7 175.0985, 146.0959
DHMEQ 8.08 202.1230 [C13H16NO] 1.96 4.05 146.3 174.0915, 159.0681, 145.0885, 130.0653
TP-232B 8.17 232.1334 [C14H18NO2] 0.73 4.42 154.0 204.1019, 189.0783, 176.1069, 161.0827, 134.0963, 132.0806
TP-234C 8.28 234.1489 [C14H20NO2] 0.11 4.56 156.9 190.0862, 188.1076, 176.0708
TP-419A 8.49 419.2332 [C26H31N2O3] 0.58 6.45 203.0 333.1593, 305.1279, 276.1251
TP-419B 8.69 419.2323 [C26H31N2O3] 1.59 6.30 198.8 NA
TP-447 8.85 447.2634 [C28H35N2O3] 1.79 7.18 222.9 389.2214, 373.1901, 345.1591, 317.1277, 289.1330, 273.1020, 261.1018
TP-465 9.60 465.2745 [C28H37N2O4] 0.68 6.86 214.0 NA
TP-467A 10.94 467.2901 [C28H39N2O4] 0.97 7.05 219.3 NA
TP-403 11.07 403.2376 [C26H31N2O2] 1.12 6.49 204.3 373.1911, 345.1596, 329.1280, 315.1492, 188.1074
TP-467B 11.09 467.2911 [C28H39N2O4] 1.46 7.09 220.3 373.1909, 359.1742, 345.1594, 329.1283
TP-433 11.47 433.2845 [C28H37N2O2] 1.09 6.57 203.6 NA
1,80 -EQDM 12.23 433.2855 [C28H37N2O2] 1.36 6.95 216.6 375.2065, 347.1743, 333.1596, 216.13791, 188.1069

NA Not available.
76 N. Negreira et al. / Analytica Chimica Acta 965 (2017) 72e82
(m/z 222.1125), eluting at 3.05 min, 3.81 min and 3.97 min,
respectively, were assigned as three isomers with molecular for-
mula [C12H16NO3]. Therefore, these TPs presented two carbon
atoms less, four hydrogen atoms less and two oxygen atoms more
than EQ, suggesting its O-deethylation and two hydroxylations.
These isomers presented very similar CCS values (Table 1), sug-
gesting that the differences between them do not represent major
structural changes. This is a common observation for positional
isomers of small molecules due to their molecular rigidness and
their large similarity in terms of structure [21].
Another three isomeric TPs were found at m/z 234.1489 eluting
at 5.60, 5.86 and 8.28 min. They matched the molecular formula
[C14H20NO2], indicating the possible hydroxylation in three
different positions of the EQ molecule. The two TPs eluting earlier
showed very similar fragmentation patterns and the only difference
with EQ was the presence of an ion at m/z 216.1383, which corre-
sponded to a loss of water from the parent compound. TP-234C
eluted much later than the other two and the fragmentation
pattern did not allow for elucidation of the position of the hydroxyl
group (Fig. S-1, Supporting information).
Two isomers were detected at m/z 232.1338 matching the mo-
lecular formula [C14H18NO2]. TP-232A eluted at 6.49 min and
showed an intense in-source fragment ion at m/z 204.1379, which
was common to one of the major fragments ions of EQ. Considering
the structure of this fragment (Fig. S-1, Supporting information), it
was hypothesized that the oxygen atom might be located on the
nitrogen atom, which would be consistent with the high lability of
the bond and the higher polarity of this TP, as deducted from the
earlier retention time. Therefore, TP-232A was assigned as the N-
oxyl derivative of EQ. On the other hand, TP-232B eluted at 8.17 min
and showed a major fragment ion at m/z 204.1019 corresponding
with [C14H18NO2], which suggested the presence of a ketone
group due to the oxidation of a hydroxyl moiety. This TP could be
positively identied as 6-ethoxy-2,2,4-trimethyl-quinolin-8-one by
analysis of the corresponding standard. Pico et al. [10] also reported
the presence in pears of a TP at m/z 232.1329 eluting, as in our case,
at a lower retention time than EQ and presenting the same frag-
ment ions that we observed for TP-232B. However, these authors
identied this TP as the N-oxyl derivative of EQ, but they could not
conrm it with any standard. Thus, our results seem to indicate that
their tentative identication was incorrect and the TP that they
observed was most likely 6-ethoxy-2,2,4-trimethyl-quinolin-8-
one.
TP-204 (m/z 204.1024) was in agreement with the O-deethyla-
tion of EQ plus an additional hydroxylation and oxidation. This TP
presented the same m/z as one fragment ion of TP-232B that cor-
responded to its O-deethylation. Therefore, TP-204 was tentatively
assigned as the O-deethylated derivative of TP-232B. TP-206 (m/z
206.1174) with a difference of only two hydrogen atoms more than
TP-204 matched with the hydroxylation of DEQ. An additional
oxidation of TP-206 would also lead to TP-204.
Three isomers were detected at m/z 218.1177, m/z 218.1179 and
m/z 218.1177 eluting at 6.55, 6.83 and 7.26 min, respectively.
Because of their molecular formula, they were assumed as the
result of a demethylation and an additional hydroxylation. The
fragmentation experiments did not allow for conrmation of their
structures, but it was hypothesized that the hydroxylation in
different positions could yield these three isomers. It should be
remarked that HRMS is particularly useful since such a small dif-
ference (0.036 u) regarding the exact mass of EQ would not be
observable in a low-resolution MS instrument, which could lead to
their wrong identication as EQ isomers.
TP-236 (m/z 236.1284) with molecular formula [C13H18NO3]
presented one carbon atom and two hydrogen atoms less and two
oxygen atoms more than EQ, suggesting a demethylation and a
dihydroxylation. Though the position of the hydroxyl groups could
not be established, this TP might be derived from TP-218 after hy-
droxylation or from TP-232B after hydroxylation and
demethylation.
TP-250 (m/z 250.1439) eluted at 6.86 min and matched the
molecular formula [C14H20NO3], suggesting that it might be
derived from TP-232B after a hydration of the double bond.
As can be observed, 1,80 -EQDM eluted at 12.23 min and pre-
sented characteristic fragments ions at m/z 375.2065, m/z 347.1743,
m/z 333.1596, m/z 216.13791 and m/z 188.1069. The fragment ions at
m/z 375, 347 and 333 corresponded to successive losses of an ethyl
group, a methyl group and another methyl group, respectively
(Fig. S-1, Supporting information). The fragment at m/z 216 corre-
sponded to the radical ion obtained after the homolytic cleavage of
the C-N bond of the dimer. The ion at m/z 188 was already described
for EQ.
Besides 1,80 -EQDM, another eight dimeric TPs were detected
during the oxidation experiments, namely TP-433, TP-403, TP-
419A, TP-419B, TP-447, TP-465, TP-467A and TP-467B (Table 1).
An isomer of 1,80 -EQDM, TP-433 (m/z 433.2376), was found at
11.47 min. This TP presented a smaller CCS (203.6 2) than 1,80 -
EQDM (216.6 2), which might indicate the existence of signicant
structural differences between both species. Although no frag-
ments were available due to its low intensity, it was hypothesized
that TP-433 might be the N,N'-dimer of EQ that has been previously
reported in pears [1,36].
TP-403 (m/z 403.2376) corresponded to the loss of two methyl
groups from 1,80 -EQDM suggesting the dimerization of the two
molecules of DHMEQ. Its fragmentation followed the same pattern
as 1,80 -EQDM (Fig. S-1, Supporting information).
Two isomers, TP-419A and TP-419B, matched the molecular
formula [C26H31N2O3], which might correspond to the dimeriza-
tion of DHMEQ and TP-218. Major fragment ions for TP-419A were
obtained at m/z 333.1593, already observed for 1,80 -EQDM, and at
m/z 305.1279, assigned to the loss of two methyl groups from the
former ion (Fig. S-1, Supporting information). No fragment ions
were available for TP-419B, but considering that three isomers were
identied for TP-218, this TP could be the result of the dimerization
DHMEQ with a different isomer of TP-218.
TP-447 (m/z 447.2634) matched the molecular formula
[C28H35N2O3] suggesting the hydroxylation followed by oxidation
of 1,80 -EQDM. Therefore, it might be the result of the dimerization
of EQ and TP-232B. This TP showed the same fragmentation pattern
as 1,80 -EQDM, but it was not possible to determine the exact po-
sition of the oxygen atom.
TP-465 (m/z 465.2745) presented two oxygen atoms more than
1,80 -EQDM, and therefore, it might correspond to the dimerization
of a molecule of C14H18NO2 and TP-234. TP-467 differed from TP-
465 in the addition of two hydrogen atoms, i.e. one unsaturation
less. Therefore, it could be due to the hydration of the double bond
of the nitrogen ring. Two isomers, TP-467A and TP-467B, were
detected at 10.94 min and 11.09 min, respectively. For 467A, no
fragments were available. However, for TP-467B a very intense in-
source fragment ion at m/z 449.2798 was observed, which corre-
sponded to a water loss. This loss suggested that the hydroxylation
might have occurred in the double bond of the nitrogen ring.
In order to compare the TPs generated with H2O2 and DPPH, the
responses obtained for the different TPs after 60 min of reaction
were plotted (Fig. 1). Responses are expressed as the percent ratio
of the peak area of each TP to the total peak area of all TPs in the
corresponding experiment. As shown, seven TPs were formed with
both H2O2 and DPPH, whereas seventeen were only observed in the
reaction with H2O2 and three only with DPPH. 1,80 -EQDM and EQI
were the major TPs obtained during the reaction with DPPH, ac-
counting for 70% and 12%, respectively. When the reaction was
 N. Negreira et al. / Analytica Chimica Acta 965 (2017) 72e82 77

Fig. 1. Relative responses for the generated EQ TPs during the oxidation experiments. Reaction time, 60 min.

performed with H2O2, TP-232B and TP-447 were the most intense
TPs with 46% and 14% of the total peak area, respectively. These
differences in the obtained proles might be attributed to the
different reaction mechanisms involved. When the reaction is
carried out with DPPH, both the electron transfer mechanism and
the hydrogen atom transfer mechanism applies [37], which results
in a much faster reaction kinetics. For this reason, the most abun-
dant TPs were those which are supposed to be more stable ac-
cording to the literature, such as 1,8-EQDM and EQI. On the
contrary, the reaction with H2O2 undergoes only through the
hydrogen atom transfer mechanism, and therefore, it takes longer
to yield the most stable TPs. This might explain why TP-232B, an
intermediate TP (Fig. 2), is more abundant with H2O2 at this reac-
tion time. A higher amount of TP-232B would lead also to a higher
amount of TP-447 since this compound is formed by dimerization
of TP-232B with EQ.
These results demonstrate the suitability of the proposed
approach to generate and characterize both known and unreported
TPs of EQ. On the basis of the identied TPs, we propose a tentative
reaction pathway for EQ (Fig. 2).
3.3. Screening of EQ TPs in sh feed
In order to assess the occurrence of the EQ TPs in real matrices,
18 Atlantic salmon feed samples were extracted following the
procedure described in the experimental section and analyzed by
UHPLC-TWIMS-QTOFMS. The selected workow involved an
untargeted data acquisition by HDMSE and the subsequent data

Fig. 2. Proposed pathway for the degradation of EQ. DHEQ was detected only in sh feed.
78 N. Negreira et al. / Analytica Chimica Acta 965 (2017) 72e82

processing using both targeted and untargeted approaches. occurrence of EQ TPs in the salmon feed samples, the same data
were further investigated for the presence of unexpected TP, i.e.
those not included in the list of the TPs generated during the
3.3.1. Targeted screening of TPs
oxidation experiments. Therefore, an untargeted processing
All the samples were rst screened against a target list of
approach was performed. To this end, the UNIFI software applies a
compounds containing all the information in Table 1, using the
series of algorithms (peak-picking, isotope clustering, etc.) to
following maximum tolerances: retention time, 0.1 min; mass
automatically generate a list of unknowns including retention
accuracy, 5 ppm, and CCS difference (DCCS), 2%. Additionally, the
times, drift times, m/z and intensities, that can be further eluci-
presence of at least one characteristic fragment ion in the high-
dated. Examination of the HE spectra allowed to narrow down to
energy spectrum was required to give a positive detection. This
those peaks showing the typical fragmentation pattern of EQ and
targeted screening allowed detecting EQ and 15 of its TPs in the
its TPs. Indeed, as previously observed during the targeted
investigated samples (Table 2). As can be observed, EQ was found in
approach, most of TPs have in common some characteristic frag-
17 samples, which is in agreement with the high detection fre-
ment ions with EQ. In total, 10 different compounds presented
quency previously reported in sh feed samples [21,38]. Among the
fragmentation patterns that pointed to EQ TPs (Table 3). Molecular
TPs, DHMEQ, 1,80 -EQDM, TP-232B and TP-403 were the most
formulae for each of these unknowns were calculated based on the
frequently detected compounds, being present in at least 89% of the
accurate mass measurements and the isotopic patterns. As shown,
samples. Indeed, DHMEQ and TP-232B were found in all of the
mass accuracies for the proposed formulae were below 3 ppm in all
analyzed samples. The high detection frequency observed for the
cases. Among them, the compound at m/z 220.1696 was tentatively
dimer 1,80 -EQDM is consistent with previous results showing this
identied as DHEQ, a TP that has been previously reported in pears
compound as the one of the major TPs of EQ [7,9]. Among the other
[1].
dimeric TPs in our target list, TP-447, TP-433 and TP-465 could also
During the course of these experiments, it was recurrently
be identied in several samples. Other TPs such as TP-234B, TP-
observed the presence of up to four peaks eluting at different
234C and TP-218A were present in more than half of the feed
retention times and presenting the exact mass of EQ within a
samples.
5 ppm window (Fig. 4). Their targeted MS/MS spectra showed the
In an attempt to compare the abundances of the different TPs in
same fragmentation pattern as EQ (Fig. S-2, Supporting informa-
the feed samples, the relative responses of EQ and its TPs were
tion) and their CCS values were also matching that of EQ within 2%
investigated (Fig. 3). Due to the lack of commercial standards for
tolerance. These results suggested either the existence of unre-
most of the TPs, the responses were expressed as the percent ratio
ported EQ isomers or most likely the in-source fragmentation of
of the peak area of each compound to the total peak are of the
labile EQ TPs. Similarly, the EU Reference Laboratory for Residues of
identied compounds in that sample. As shown, 1,80 -EQDM was the
Pesticides (EURL-SRM) reported the presence in salmon of an un-
most abundant TP in most of the samples, presenting in general a
known compound eluting earlier than EQ and presenting the same
higher response than the parent compound EQ. DHMEQ, TP-232B
accurate mass and MS/MS transitions as EQ [39]. However, they did
and TP-403 also showed considerable responses in many samples,
not address the identication of these species.
whereas the rest of TPs presented lower signal intensities. Although
In our case, no potential precursor ions were found at the cor-
this representation may be useful for discussion purposes, it is
responding retention times in the LE spectra. Therefore, in order to
noteworthy that a higher response does not always guarantee a
test our hypothesis, the samples were re-analyzed using softer
higher concentration of the TP as different ESI ionization ef-
conditions during the ESI process: the desolvation gas temperature
ciencies and matrix effects may be expected for different TPs.
was set to 200  C, the cone voltage was lowered to 0 V and the ion
Therefore, these data should be taken only as indicative, and be
source temperature was set to 100  C. Under these conditions, it
treated cautiously until the availability of the corresponding stan-
was possible to detect several ions whose fragmentation might lead
dards allows a proper quantication.
to the observed ions at m/z 218.1539. To conrm them as the pre-
cursor ions, they were subjected to targeted MS/MS experiments
3.3.2. Untargeted screening of TPs using a CE of 15 eV. As can be observed in Fig. 5, all of the selected
In an effort to provide a more comprehensive insight into the precursor ions fragmented yielding fragment ions at m/z
218.1539 5 ppm. These compounds, namely unknown-276 (m/z
276.1961), unknown-246 (m/z 246.1493), unknown-435A (m/z
Table 2
TPs identied in salmon feed samples by the proposed targeted screening. Retention 435.3011) and unknown-435C (m/z 435.3010), were considered as
time shift (DtR), mass measurement errors (Dm) and percent difference of collision precursors ions undergoing signicant in-source fragmentation
cross section (DCCS). under normal ESI operation conditions.
Compound DtR (min) Dm (ppm) DCCS (%) No. positives* When screening for unknown-435A and unknown-435C in the
samples, another isomer at m/z 435.3011 (unknown-435B) was also
EQ 0.03 1.22 1.57 17
TP-222B 0.01 1.71 0.97 1 detected at an intermediate retention time (11.22 min). As the other
TP-234B 0.02 3.71 1.60 10 two, this compound showed a MS/MS spectrum yielding an ion at
TP-204 0.01 2.54 0.72 1 m/z 218.1531 as the base peak (Fig. 5). However, unknown-435B did
TP-232A 0.02 2.70 1.24 2 not present this ion in the LE spectrum, suggesting a higher stability
TP-218A 0.1 2.21 1.70 13
EQI 0.01 2.71 1.03 7
during the ionization process. These three isomers matched the
TP-218C 0.1 3.50 2.00 7 molecular formula [C28H39N2O2], that was tentatively assigned to
DHMEQ 0.01 1.47 1.39 18 a dimeric TP showing one unsaturation less than 1,80 -EQDM.
TP-232B 0.01 2.92 1.62 18 Considering the lability of unknown-435A and unknown-435C, it
TP-234C 0.08 2.67 1.01 10
was postulated that these TPs might be the result of the bond be-
TP-447 0.02 1.06 0.26 2
TP-465 0.05 1.70 1.01 5 tween the nitrogen atom of one molecule of EQ and the alkene of
TP-403 0.02 2.73 1.10 17 another EQ molecule, 1,30 -EQDM and 1,40 -EQDM (Fig. 5). Instead,
TP-433 0.07 2.31 1.88 4 unknown-435B might be derived from 1,80 -EQDM after the hy-
1,80 -EQDM 0.02 2.45 1.27 17 drogenation of the double bond in the nitrogen ring or due to the
*Total number of samples analyzed 18. dimerization of EQ with DHEQ.
 N. Negreira et al. / Analytica Chimica Acta 965 (2017) 72e82 79

Fig. 3. Relative responses for the TPs detected in salmon feed following the targeted screening approach.

Table 3
Retention times (tR), accurate m/z, molecular formula, mass measurement errors (Dm), drift time values, collision cross section (CCS) values and characteristic fragment ions of
potential EQ TPs and number of positives detected by untargeted screening.

Compound tR Observed m/ Molecular Dm Drift time CCS Characteristic fragment ions (m/z) No.
(min) z formula (ppm) (ms) (2) Positivesc

Unknown-414 5.43 414.2275 [C24H32NO5] 0.1 6.37 200.98 396.2174, 356.1854, 340.1539, 228.13823, 188.10694, 18
160.0756

Unknown-396A 7.91 396.2165 [C24H30NO4] 1.1 6.13 194.75 353.1619, 352.1549, 228.1383 18
Unknown-396B 8.14 396.2166 [C24H30NO4] 0.8 6.54 205.75 202.12248, 174.09126, 216.13807 14
Unknown-396C 8.22 396.2169 [C24H30NO4] 0.1 6.25 197.87 228.1383, 202.1228, 200.1063, 340.1550, 167.0707 9
Unknown-276a 8.25 276.1961 [C17H26NO2] 0.9 5.05 168.49 218.1542, 202.1220, 190.1256, 178.1228, 162.0924 18
Unknown-246a 8.72 246.1493 [C15H20NO2] 0.4 4.69 160.14 228.1382, 218.1542, 202.1221, 188.1073, 174.0909, 162.0537 3
DHEQ 9.03 220.1696 [C14H22NO] 2.4 4.50 156.15 204.1383, 202.1225, 190.1228, 188.1070, 160.0757, 148.0757 5
Unknown- 10.79 435.3011 [C28H39N2O2] 0.1 4.43b 154.64b 218.1532 14
435Aa
Unknown-435B 11.22 435.3011 [C28H39N2O2] 0.1 6.94 216.49 218.1548, 178.1218, 150.0931 7
Unknown- 11.59 435.3010 [C28H39N2O2] 0.4 4.44b 154.82b 420.2762, 218.1531, 188.1064, 148.0779 15
435Ca
a
Compounds yielding in-source fragment ions at m/z 218.1539 5 ppm.
b
Corresponding to the in-source fragment ion at m/z 218.1539.
c
Total number of samples analyzed 18.

Unknown-276 matched the molecular formula [C17H26NO2]
and experienced a loss corresponding to C3H6O (58.042 u) upon
mild fragmentation (Fig. 5). This suggested the presence of such a
moiety on the nitrogen atom of EQ molecule. On the other hand,
unknown-246 matched the formula [C15H20NO2], showing the
loss of CO (27.995 u) to yield m/z 218.1539. Therefore, this com-
pound was tentatively identied as the N-formyl derivative of EQ
(Fig. 5), which would explain the high lability observed for this
compound during the ESI process. To discard that this TP was an
artifact formed during the extraction procedure, several samples
were analyzed without the use of formic acid. These results
conrmed the presence of the unknown-246 in the sh feed
samples.
The rest of unknowns included in Table 3 presented MS/MS
spectra that highly resembled the fragmentation pattern of EQ
(Fig. S-3, Supporting information). Some of them, like unknown-
396A, unknown-396C and unknown-414, were highly abundant
in the analyzed feed samples being among the major peaks in the
total ion chromatograms in many samples (Fig. S-4, Supporting
information). Although the corresponding molecular formula were
estimated on the basis of their accurate mass and isotopic pattern,
the enormous number of structural possibilities did not allow
providing tentative identications for these species. Therefore,
further studies using additional characterization techniques such as
nuclear magnetic resonance (NMR) will be necessary to address the
elucidation of these complex TPs. As can be observed in Table 3,
unknown-414, unknown-396A and unknown-276 were the most
frequently detected compounds, being present in all the feed
80 N. Negreira et al. / Analytica Chimica Acta 965 (2017) 72e82

Fig. 4. Extracted ion chromatograms for the [MH] ion at m/z 218.1539 and its main fragment ions at m/z 160.0757 and m/z 188.1070 in a salmon feed sample.

samples. The unknown-396B and the dimeric TPs, unknown-435A
and unknown-435C, presented also a high detection frequency
(78e83%) in the analyzed feed samples.
4. Conclusions
In the current work, a novel approach based on UHPLC-TWIMS-
QTOFMS has been used for the rst time to conduct a compre-
hensive identication of EQ TPs in sh feed samples. The employed
approach involved an untargeted data acquisition taking advantage
of the HDMSE mode, followed by the processing of data using both
targeted and untargeted strategies.
The study revealed the presence of several previously unre-
ported EQ TPs in commercial sh feed, some of them with re-
sponses higher than the parent compound EQ. The high
dimerization of EQ with some of its TPs as well as with other matrix
components of the feed is remarkable.
Further studies are still necessary to (1) quantify the occurrence
 N. Negreira et al. / Analytica Chimica Acta 965 (2017) 72e82
Fig. 5. Accurate MS/MS spectra and proposed chemical structures for the four unknowns yielding m/z 218.1539 5 ppm upon in-source fragmentation and for unknown-435B.
of these new compounds in feed, sh meal, and sh samples, (2)
evaluate their potential toxic effects on consumers and farmed
animals and (3) study their potential release into the aquatic
environment from sea cage farms. In this way, the presented results
are highly valuable since they provide information on molecular
formulae, accurate masses and CCS values of main EQ TPs that
might be used for future monitoring programmes.
Acknowledgments
This research was funded by the Norwegian Ministry of Trade,
Industry and Fisheries. N.N. acknowledges the Ministry of Culture,
Education and University Planning of the Xunta de Galicia for her
postdoctoral contract. The authors acknowledge Dr. F. Thrun (BASF)
for kindly supplying a standard of 6-ethoxy-2,2,4-trimethyl-qui-
nolin-8-one.
81
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.aca.2017.02.021.
References
[1] A. Baszczyk, A. Augustyniak, J. Skolimowski, Ethoxyquin: an antioxidant used
in animal feed, Int. J. food Sci. (2013) 2013.
[2] IMO, (International Maritime Organisation), International Maritime
Dangerous Goods Code, IMO Publishing, London, United Kingdom, 2014.
[3] EC, List of the Authorised Additives in Feedingstuffs Published in Application
of Article 9t (b) of Council Directive 70/524/EEC Concerning Additives in
Feedingstuffs (2004/C 50/01), 2004.
[4] Y. Aoki, et al., Determination of ethoxyquin by high-performance liquid
chromatography with uorescence detection and its application to the survey
of residues in food products of animal origin, J. AOAC Int. 93 (1) (2010)
277e283.
[5] A. Hobson-Frohock, Residues of ethoxyquin in poultry tissues and eggs, J. Sci.
82 N. Negreira et al. / Analytica Chimica Acta 965 (2017) 72e82
Food Agric. 33 (12) (1982) 1269e1274.
[6] V.J.B. Bohne, A.-K. Lundebye, K. Hamre, Accumulation and depuration of the
synthetic antioxidant ethoxyquin in the muscle of Atlantic salmon (Salmo
salar L.), Food Chem. Toxicol. 46 (5) (2008) 1834e1843.
[7] A.-K. Lundebye, et al., Levels of synthetic antioxidants (ethoxyquin, butylated
hydroxytoluene and butylated hydroxyanisole) in sh feed and commercially
farmed sh, Food Addit. Contam. Part A 27 (12) (2010) 1652e1657.
[8] A.J. de Koning, The antioxidant ethoxyquin and its analogues: a review, Int. J.
Food Prop. 5 (2) (2002) 451e461.
[9] S. Thorisson, F.D. Gunstone, R. Hardy, Some oxidation products of ethoxyquin
including those found in autoxidising systems, Chem. Phys. lipids 60 (3)
(1992) 263e271.
[10] Y. Pico , et al., Proling of compounds and degradation products from the
postharvest treatment of pears and apples by ultra-high pressure liquid
chromatography quadrupole-time-of-ight mass spectrometry, Talanta 81 (1)
(2010) 281e293.
[11] V.Y.-M. Leong, T. Brown, Toxicosis in broiler chicks due to excess dietary
ethoxyquin, Avian Dis. (1992) 1102e1106.
[12] M. Manson, J. Green, H. Driver, Ethoxyquin alone induces preneoplastic
changes in rat kidney whilst preventing induction of such lesions in liver by
aatoxin B1, Carcinogenesis 8 (5) (1987) 723e728.
[13] A. Baszczyk, R. Osiecka, J. Skolimowski, Induction of chromosome aberrations
in cultured human lymphocytes treated with ethoxyquin, Mutat. Research/
Genetic Toxicol. Environ. Mutagen. 542 (1) (2003) 117e128.
[14] FEEDAP, EFSA Panel on Additives and Products or Substances used in Animal
Feed. Safety and efcacy of ethoxyquin (6-ethoxy-1,2-dihydro-2,2,4-
trimethylquinoline) for all animal species, EFSA J. 13 (11) (2015) 4272.
[15] R. rnsrud, et al., Investigations on the metabolism and potentially adverse
effects of ethoxyquin dimer, a major metabolite of the synthetic antioxidant
ethoxyquin in salmon muscle, J. Food Protection 74 (9) (2011) 1574e1580.
[16] K. Haya, L. Burridge, B. Chang, Environmental impact of chemical wastes
produced by the salmon aquaculture industry, ICES J. Mar. Sci. J. du Conseil 58
(2) (2001) 492e496.
[17] EU, Commission regulation (EC) No 429/2008 of 25 April 2008 on detailed
rules for the implementation of Regulation (EC) No 1831/2003 of the Euro-
pean Parliament and of the Council as regards the preparation and the pre-
sentation of applications and the assessment and the authorisation of feed
additives, Ofcial J. Eur. Union (2008) 1e65. L133/1.
[18] P. He, R.G. Ackman, HPLC determination of ethoxyquin and its major oxida-
tion products in fresh and stored sh meals and sh feeds, J. Sci. Food Agric.
80 (1) (2000) 10e16.
[19] J.C. May, J.A. McLean, Ion mobility-mass spectrometry: time-dispersive
instrumentation, Anal. Chem. 87 (3) (2015) 1422e1436.
[20] F. Lanucara, et al., The power of ion mobility-mass spectrometry for structural
characterization and the study of conformational dynamics, Nat. Chem. 6 (4)
(2014) 281e294.
[21] J. Regueiro, N. Negreira, M.H.G. Berntssen, Ion mobility-derived collision cross
section as an additional identication point for multi-residue screening of
pesticides in sh feed, Anal. Chem. 88 (2016) 11169e11177.
[22] . Rajski, M. del Mar Go  mez-Ramos, A.R. Ferna ndez-Alba, Large pesticide
multiresidue screening method by liquid chromatography-Orbitrap mass
spectrometry in full scan mode applied to fruit and vegetables, J. Chromatogr.
A 1360 (2014) 119e127.
[23] N. Negreira, et al., Reactivity of vinca alkaloids during water chlorination
processes: identication of their disinfection by-products by high-resolution
quadrupole-Orbitrap mass spectrometry, Sci. Total Environ. 544 (2016)
635e644.
[24] N. Negreira, et al., Transformation of tamoxifen and its major metabolites
during water chlorination: identication and in silico toxicity assessment of
their disinfection byproducts, Water Res. 85 (2015) 199e207.
[25] M.F. Bush, I.D. Campuzano, C.V. Robinson, Ion mobility mass spectrometry of
peptide ions: effects of drift gas and calibration strategies, Anal. Chem. 84 (16)
(2012) 7124e7130.
[26] T. Pacini, et al., Multidimensional analytical approach based on UHPLC-UV-ion
mobility-MS for the screening of natural pigments, Anal. Chem. 87 (5) (2015)
2593e2599.
[27] L. Fiebig, R. Laux, A collision cross section and exact ion mass database of the
formulation constituents polyethylene glycol 400 and polysorbate 80, Int. J.
Ion. Mobil. Spectrom. (2016) 1e7.
[28] M. Carevic, et al., Structural elucidation of enzymatically synthesized galacto-
oligosaccharides using ion-mobility spectrometry tandem mass spectrometry,
J. Agric. Food Chem. 64 (18) (2016) 3609e3615.
[29] A. Troc, et al., Structural elucidation of -lactam diastereoisomers through ion
mobility mass spectrometry studies and theoretical calculations, J. Mass
Spectrom. 51 (4) (2016) 282e290.
[30] V. Shah, et al., Enhanced data-independent analysis of lipids using ion
mobility-TOFMSE to unravel quantitative and qualitative information in hu-
man plasma, Rapid Commun. Mass Spectrom. 27 (19) (2013) 2195e2200.
[31] J. Hofmann, et al., Identication of carbohydrate anomers using ion mobility-
mass spectrometry, Nature 526 (7572) (2015) 241e244.
[32] R.J. Ruch, S.-j. Cheng, J.E. Klaunig, Prevention of cytotoxicity and inhibition of
intercellular communication by antioxidant catechins isolated from Chinese
green tea, Carcinogenesis 10 (6) (1989) 1003e1008.
[33] K. Mishra, H. Ojha, N.K. Chaudhury, Estimation of antiradical properties of
antioxidants using DPPH assay: a critical review and results, Food Chem. 130
(4) (2012) 1036e1043.
[34] Z. NanQun, et al., Identication of oxidation products of (-)-epigallocatechin
gallate and (-)-epigallocatechin with H2O2, J. Agric. Food Chem. 48 (4) (2000)
979e981.
[35] A.I. Calderon, et al., Screening antioxidants using LC-MS: case study with
cocoa, J. Agric. Food Chem. 57 (13) (2009) 5693e5699.
[36] FAO, Pesticide Residues in Food 1999 Evaluations: Residues, Part 1, 2000.
[37] J. Xie, K. Schaich, Re-evaluation of the 2, 2-diphenyl-1-picrylhydrazyl free
radical (DPPH) assay for antioxidant activity, J. Agric. Food Chem. 62 (19)
(2014) 4251e4260.
[38] J. Nacher-Mestre, et al., Screening of pesticides and polycyclic aromatic hy-
drocarbons in feeds and sh tissues by gas chromatography coupled to high-
resolution mass spectrometry using atmospheric pressure chemical ioniza-
tion, J. Agric. Food Chem. 62 (10) (2014) 2165e2174.
[39] EURL-SRM, Analysis of Ethoxyquin and its Metabolites in Fish Using the
QuEChERS Method, Analytical Observations Reports SRM-24/(V1)/17.05.2016,
2016.