- Email: [email protected]
Fast and multiresidue determination of twenty glucocorticoids in bovine milk using ultra high performance liquid chromatography–tandem mass spectrometry
Accepted Manuscript Title: Fast and multiresidue determination of twenty glucocorticoids in bovine milk using Ultra High Performance Liquid Chromatography- tandem mass spectrometry Author: Y. Deceuninck E. Bichon F. Monteau G. Dervilly-Pinel J.P. Antignac B. Le Bizec PII: DOI: Reference:
S0021-9673(13)00597-9 http://dx.doi.org/doi:10.1016/j.chroma.2013.04.019 CHROMA 354252
To appear in:
Journal of Chromatography A
Received date: Revised date: Accepted date:
29-1-2013 4-4-2013 8-4-2013
Please cite this article as: Y. Deceuninck, E. Bichon, F. Monteau, G. DervillyPinel, J.P. Antignac, B. Le Bizec, Fast and multiresidue determination of twenty glucocorticoids in bovine milk using Ultra High Performance Liquid Chromatography- tandem mass spectrometry, Journal of Chromatography A (2013), http://dx.doi.org/10.1016/j.chroma.2013.04.019 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
Highlights We developed a rapid 10 min UHPLC-MS/MS method for 20 corticosteroids in milk
3
We validated the method according to Commission Decision 2002/657/EC requirements
4
The methods allow low trace level (sub ppb) corticosteroids detection
5
Efficiency has been assessed on various bovine milks
ip t
2
Ac ce
pt
ed
M
an
us
cr
6
Page 1 of 21
6
Fast and multiresidue determination of twenty glucocorticoids in bovine milk
7
using
8
spectrometry.
Ultra
High Performance
Liquid Chromatography- tandem
mass
9 10 11
1
1
1
1
1,2
Y. Deceuninck *, E. Bichon , F. Monteau , G. Dervilly-Pinel , J.P. Antignac
and B. Le Bizec
1
15
LUNAM Université, Oniris, Laboratoire d’Etude des Résidus et Contaminants dans les Aliments
(LABERCA), Nantes, F-44307, France. 2
INRA, Nantes, F-44307, France.
*
Corresponding author
cr
14
1
us
13
ip t
12
16
18
an
17
Yoann DECEUNINCK, ONIRIS, École nationale vétérinaire, agroalimentaire et de l’alimentation
20
Nantes-Atlantique, Laboratoire d’Etude des Résidus et Contaminants dans les Aliments (LABERCA),
21
Atlanpole - La Chantrerie, BP 40706, Nantes, F-44307, France, tel : +33 2 40 68 78 80, fax : +33 2 40
22
68 78 78, email : [email protected].
ed pt Ac ce
23
M
19
Page 2 of 21
Abstract
24
Glucocorticoids constitute a class of molecules widely used in animal husbandry. Some of these
25
compounds are licensed for veterinary practices while their use for growth promoting purposes is
26
prohibited within the European Union. In order to ensure the respect of the legislation and consumers
27
safety, several methodologies have been proposed to monitor these substances in various products,
28
including edible matrices for which a regulatory limit has been set up (MRL). An extended range of
29
targeted analytes together with reduced time of analysis and cost are however still current challenges
30
regularly revisited according to the continuous technological improvements. In this context, the aim of
31
the present study was to develop and implement a new fast and multi-residue method based on
32
UHPLC-MS/MS for the determination of twenty glucocorticoids in bovine milk, included the screening
33
of the three regulated MRL compounds (dexamethasone, betamethasone and prednisolone). This
34
validated method authorises such multi-analyte measurement within a 10 minutes runtime while the
35
signal specificity is ensured through the SRM acquisition mode. Decision limits and detection
36
capabilities were calculated in the range [0.001 and 0.363] µg L , which allows a very efficient control
37
at low trace level for a potential illegal use of these substances. The performances obtained in terms
38
of application range, selectivity and sensitivity were found significantly improved in comparison to
39
other reported approaches either for screening or confirmation purposes: regarding linearity,
40
correlation coefficients were above 0.98 within the range [0.01-5.0 µg L ], repeatability and
41
reproducibility parameters ranged from 1 to 30 % with the maximum relative standard deviation (RSD)
42
observed for cortisone (30.1%). Stability of the stock solutions and minor changes in the standard
43
operating procedure have been included for the determination of ruggedness of the method.
44
Identification was systematically ensured according to 4 identification points, RSD of transitions ratio
45
(T2/T1) ranged from 3.2% and 19.3% and the RSD of the retention time was lower than 0.25%.
us
cr
ip t
23
-1
46
pt
ed
M
an
-1
Keywords
48
Corticosteroids, milk, UHPLC-MS/MS, Chemical food safety, mass spectrometry, MRL, Multi-residue
49
analysis
50
Ac ce
47
Page 3 of 21
50
1- Introduction
51 Glucocorticoids are a class of anti-inflammatory drugs (AIDs) that are widely used in human and
53
veterinary medicine for their anti-inflammatory, antipyretic, analgesic, anti-allergic and metabolic
54
properties [1]. Nevertheless, in livestock, the administration of those molecules results in a significant
55
gain of weight essentially due to an increase of water retention in tissues. Therefore, the use of
56
corticosteroids has a negative impact on meat quality and induces a decrease of the organoleptic
57
properties. Moreover, these substances may cause adverse effects on human health, including
58
hypertension, obesity or osteoporosis [2].
59
Due to additional properties of glucocorticoids such as the feed conversion rate improvement and their
60
synergetic effects when they are combined with other banned substances like steroids or beta
61
agonists [3-5], the use of these molecules in livestock is regulated in the European Union [6]. Some
62
compounds such as prednisolone, methylprednisolone, betamethasone and dexamethasone are
63
licensed for therapeutic diseases in breeding animals. Maximum Residue Limits (MRLs) have been set
64
for these molecules by the EU in several matrices such as muscle, liver, kidney and milk from different
65
animal species [7-8]. For milk, MRLs have been fixed at 6 µg L -1
an
us
cr
ip t
52
-1
for prednisolone, 2 µg L
-1
for
methylprednisolone and 0.3 µg L
for betamethasone and dexamethasone, whatever the animal
67
specie considered (bovine or ovine). Consequently, in the case of an authorized animal treatment with
68
one of these substances, a withdrawal period has to be respected between the end of treatment and
69
slaughter, while the use of other non-licensed analytes is prohibited. Regarding the control of
70
glucocorticoids misuse in livestock, standard operating procedures for both efficient screening and
71
confirmation purposes have to be implemented to ensure the respect of the legislation and more
72
generally the consumer’s safety.
73
In this context, many different standard operating procedures based on gas chromatography (GC) or
74
liquid chromatography (LC) coupled to mass spectrometry (MS) for analysing anti-inflammatory drugs
75
in animal-food products have been proposed in literature and reviewed by Gentili [9]. GC-MS was first
76
used as an adequate technique for screening and confirming glucocorticoids compounds in complex
77
biological matrices (milk, liver or urine) at trace levels, with LOD established around 0.25 µg L for
78
several analytes (triamcinolone, betamethasone, cortisol, flumethasone, desoxycortisone, 6-
79
methylprednisolone, dexamethasone, prednisolone, isoflupredone) [10]. However, while low limits of
80
detection were obtained, the derivatization step which was necessary to enhance the volatility of the
81
target molecules was considered as time-consuming. A lack of specificity for epimeric compounds
82
such as dexamethasone and betamethasone was also found sometimes an issue [11]. On the other
83
hand, the development of LC-MS and LC-MS
84
atmospheric-pressure ionisation (APCI) or atmospheric pressure photoionization (APPI) interfaces are
85
considered as powerful alternative to the use of GC-MS. The performances of these techniques have
86
demonstrated a high efficiency in terms of specificity and sensitivity. Nowadays, the use of LC-MS/MS
87
has become the analytical technique of choice for monitoring corticosteroids, especially for multi-
88
residue analysis at trace levels in complex matrices. More recently, some authors have reported the
Ac ce
pt
ed
M
66
-1
n
techniques using electrospray ionisation (ESI),
Page 4 of 21
advantages of last generation of ultra high performance chromatography (UHPLC) coupled to tandem
90
mass spectrometry in terms of chromatographic resolution, sensitivity and shortened run times [12-14].
91
Most of these studies have however focused on a limited number of compounds. Regarding milk as
92
matrix of interest, some authors have proposed procedures for the determination of 1 to 17
93
glucocorticoids in milk [11-13, 15-17]. In parallel, other have reported the analysis of a range of 1 to 11
94
molecules in liver and/or edible tissues [14-15, 18-24].
95
Most of these analytical methods have been validated according to the European requirements fixed
96
in the 2002/657/EC decision [25]. The reported decision limits (CC) and detection capabilities (CC)
97
in milk are compounds dependant to a large extend but also depend on the standard operating
98
procedure used. For example, Cui et al. [13] reported limits of quantification (LOQ) of 17
99
corticosteroids in milk and eggs ranging from 0.04 to 1.27 µg kg respectively for flumethasone and
100
fluorometholone. Furthermore, Caretti et al. [16] reported CC for 9 analytes ranging 0.05 to 0.74 µg
101
kg respectively for cortisone acetate and flumethasone.
102
In that context, the aim of the present study was to develop a method for screening and confirming in
103
milk an extended range of glucocorticoids that may be used as growth promoting agents. This method
104
comprises the analysis of 20 molecules: amcinonide, beclomethasone, betamethasone, budesonide,
105
cortisol, cortisone, desoxymethasone, dexamethasone, flumethasone, flunisolide, flurandrenolide,
106
halcinonide,
107
dehydrocortexolone, 16-methylcortexolone, 2-methyl-9-fluorocortisone (table 1). Fludrocortisone
108
and three deuterated internal standards, i.e dexamethasone-d4, prednisolone-d6 and triamcinolone
109
acetonide-d6 were also included for accurate quantification according to the isotope dilution method.
110
The proposed standard operating procedure was based on a purification step previously described by
111
Deceuninck et al. [14] from which an additional step consisting in a protein precipitation with acetone
112
was added. An ultra high performance liquid chromatography coupled to a triple quadrupole mass
113
spectrometer (UHPLC-MS/MS) was used for monitoring targeted compounds, and the method was
114
validated according to the Commission Decision 2002/657 criteria [25]. Through the implementation of
115
an efficient standard operating procedure, the developed analytical method enables a high-throughput
116
analysis of corticosteroids in milk with very low associated decision limits and detection capabilities.
117
This analytical method was successfully implemented to the analysis of milk samples coming from the
118
French national control plan.
120
cr
-1
an
methylprednisolone,
prednisolone,
triamcinolone
acetonide,
1-
pt
ed
M
isoflupredone,
us
-1
Ac ce
119
ip t
89
2- Experimental
121 122
2.1- materials and reagents
123
Analytical grade cyclohexane, ethyl acetate, diethylether, isopropanol, acetone and methanol, HPLC
124
grade acetonitrile, as well as solid-phase extraction cartridges (SPE C18: 2 g and silica: 1g) were
125
purchased from Carlo Erba Reactifs SDS (Val de Reuil, France). Sodium acetate and sodium
Page 5 of 21
126
carbonate were obtained from Sigma (St. Louis, MO, USA) and Merck (Darmstadt, Germany)
127
respectively. Enzymatic preparation was a purified lyophilized extract from Helix pomatia (Sigma, St.
128
Louis, MO, USA), dissolved in water (50,000 IU). Ultrapure water was purified using a Milli-Q osmosis
129
system from Millipore (Milford, MA, USA).
130
2.2- reference substances Amcinonide, beclomethasone, betamethasone, cortisol, cortisone, dexamethasone, fludrocortisone,
132
flumethasone, flunisolide, halcinonide, methylprednisolone, prednisolone and triamcinolone acetonide
133
were purchased from Sigma-Aldrich (St. Louis, MO, USA). Budesonide, desoxymethasone,
134
flurandrenolide,
135
fluorocortisone were obtained from Steraloids Inc. Ltd (London, England). Dexamethasone-d4 and
136
prednisolone-d6 (used as internal standards) were purchased from Cluzeau Info Labo (Courbevoie,
137
France) and triamcinolone acetonide- d6 was obtained from RIKILT (Wagenningen, The Netherlands).
138
Stock standard solutions of each molecule were prepared in methanol at a concentration of 1 mg.mL .
139
Working solutions were obtained by tenfold successive dilution in methanol at concentrations from 100
140
ng µL to 0.1 ng µL . All the standard solutions were stored at -20°C, in the dark.
16-methylcortexolone,
2-methyl-9-
-1
us
cr
1-dehydrocortexolone,
-1
-1
an
141
isoflupredone,
ip t
131
2.3- milk samples
Bovine and goat milk samples used for both method development and validation were purchased from
143
a local supermarket. In this way, skimmed and semi-skimmed milk, as well as whole milk were
144
included in the study. The developed and validated method was then applied to a second set of milk
145
samples originated from the French national control plan.
ed
2.4- standard operating procedure
pt
146
M
142
For each milk sample to be analysed, an initial volume of 10 mL was prepared in a polypropylene
148
tube. A protein precipitation was then performed by adding 8 mL of acetone. After evaporation of the
149
organic layer, an enzymatic hydrolysis was carried out at 50°C, during 4 h, using a purified -
150
glucuronidase, extracted from Helix pomatia, in order to deconjugate phase II metabolites (glucuronide
151
and sulfate forms) that might be present in milk samples coming from treated animals. Then, two
152
successive solid phase extractions steps were performed. The first one was carried out using a non-
153
polar (reverse) stationary phase column (C18) previously activated successively with 10 mL of
154
methanol and 10 mL of water. After loading the sample, the column was washed with 5 mL of water
155
and 5 mL of cyclohexane. Then, the analytes were eluted with 6 mL of diethylether. After evaporating
156
the extracts, a washing step of the aqueous layer was performed using sodium carbonate and
157
diethylether. A second SPE purification step was performed using a polar (normal) stationary phase
158
(SiOH). The column was first conditioned with at least 20 mL of cyclohexane. After loading the extract,
159
the column was rinsed with 5 mL of a mixture of cyclohexane / ethyl acetate (50:50, v/v) before elution
160
with a 10 mL ethyl acetate / isopropanol (90:10, v/v) mixture. The extracts were evaporated under N2
161
(40°C), redissolved in 50 µL of the mixture water/methanol (60:40, v/v) + 0.5% of acetic acidand
162
transferred into chromatographic vials for injection and analysis.
Ac ce
147
Page 6 of 21
163
2.5- chromatographic conditions ®
164
The LC separation was achieved on a Waters Acquity UPLC system (Milford, MA, USA), equipped
165
with an Acquity BEH C18 column (1.7 µm, 2.1×100 mm) maintained at 60°C. The LC mobile phases
166
consisted in 0.5% acetic acid in water (solvent A) and 0.5% acetic acid in acetonitrile (solvent B). A
167
flow rate of 0.6 mL min was applied. The injection volume was 2 µL. 2.6- detection parameters
ip t
168
-1
Detection was carried out using a Xevo TQ MS instrument (Waters, Milford, MA, USA), operating in
170
the negative Electrospray Ionisation mode (ESI-). MS data were acquired using the Selected Reaction
171
Monitoring mode (SRM). Capillary voltage was set at 3 kV, source temperature at 150°C,
172
desolvatation temperature at 500°C, desolvatation gas (N2) at 920 L h and collision gas flow at 0.15
173
mL min . Diagnostic SRM transitions were first generated using Waters’Intellistart
174
parameters were then optimized individually for each diagnostic signal. Data acquisition and data
175
processing were performed using MassLynx, version 4.1 software.
TM
software. All the
us
-1
2.7- method validation
an
176
-1
cr
169
Twenty representative bovine milk samples (non-fat, semi-skimmed and unskimmed milk) were
178
selected as blank materials after a preliminary analysis. In all these samples, internal standards were
179
added at a concentration of 0.3 µg L
180
prednislone-d6 and triamcinolone-d6.
181
The applied method validation guideline fitted with the 2002/657/EC requirements and was based on
182
the protocol previously described by Antignac et al. and Vanden Bussche et al. [19,26]. Several
183
parameters, such as specificity, repeatability, linearity, ruggedness, decision limit and detection
184
capability were assessed. The decision limit (CC) was calculated (equation 1) using the standard
185
deviation of the noise amplitude expressed to the corresponding internal standard (δN) and the slope
186
obtained from the calibration curve (a) established from 8 points ranging from 0.01 to 5.0 µg L . The
187
concentrations taken in account depended on the results obtained for each targeted compounds: for
188
example, the calibration curves of cortisol and isoflupredone were carried out according to
189
concentrations ranging from 0.01 to 0.4 µg L
190
capability (CC) was determined (equation 2) using the standard deviation (δS) of the signal amplitude
191
obtained from 20 spiked blank samples expressed to the decision limit level (CC obtained from
192
equation 1.
193
Equation 1: CC=(2.33×δN)/a
194
Equation 2: CC=CC+1.64×δS.
195
Calculated detection capability (CC) and decision limit (CCare applicable to corticosteroids
196
considered as forbidden substances; for MRL substances, these values must be re-assessed in a
197
different context.
-1
for dexamethasone-d4 and 1.0 µg L
for fludrocortisone,
Ac ce
pt
ed
-1
M
177
-1
-1
-1
and 0.2 to 5.0 µg L , respectively. The detection
198
Page 7 of 21
199 200
3- Results and discussion
3.1
UHPLC conditions and MS detection
Preliminary experiments were performed in order to optimize the chromatographic conditions used for
202
the separation of the twenty glucocorticoids of interest. A first evaluation of several column candidates
203
(Acquity BEH C18, 1.7 µm, 2.1×100 mm, Acquity BEH C18, 1.7 µm, 2.1×50 mm, Acquity BEH Phenyl,
204
1.7 µm, 2.1×100 mm, Acquity BEH RP C18 Shield, 1.7 µm, 2.1×50 mm and Hypersil Gold 1.9 µm,
205
2.1×100 mm) was carried out and the Acquity BEH C18, 1.7 µm, 2.1×100 mm column was finally
206
selected according to its good separation performances obtained for the different molecules,
207
especially for dexamethasone and betamethasone isomers. The gradient using water + 0.5% acetic
208
acid as solvent A and acetonitrile + 0.5 % of acetic acid as solvent B was selected and optimized
209
according to our experience background [27]. The initial conditions were set at 95% of solvent A and
210
5% of solvent B for the first 0.5 min, phase B was increased linearly to 25% in 0.5 min maintained for 4
211
min, then increased to 50% in 2 min and maintained for 1.5 min, then increased to 100% in another
212
0.1 min for 0.4 min, and finally returned to the initial composition in 0.2 min. the column was then
213
equilibrated for 0.8 min before the next injection. Then several levels of solvents were programmed
214
until a level at 100% of acetonitrile + 0.5% of acetic acid at 7 min was reached. The flow rate and the
215
temperature were set to 0.6 mL min and 60°C, respectively. The resulting chromatographic run was
216
10 minutes with a first (more polar) compound (isoflupredone) eluted at 2.60 min and a last (less polar)
217
one (amcinonide) eluted at 8.07 min (Figure 1). These optimized chromatographic conditions were
218
fixed for the method validation considering the adequate separation of the target glucocorticoids and
219
particularly the good resolution of the two isomers dexamethasone and betamethasone (Rs=1.04)
220
which is usually reported as an issue and was one of the objectives of the present method
221
development. Figure 1 shows an example of typical chromatographic traces obtained for a
222
representative milk sample, i.e. corresponding to a pool of the twenty blank samples selected for the
223
validation process, spiked with 1 µg L of each molecule. All these targeted molecules were detected
224
in the range [2.60 – 8.07 min]. The Selected Reaction Monitoring (SRM) was used as the adequate
225
acquisition mode for reaching high confidence level in terms of unambiguous identification of target
226
analytes (i.e. a minimum of 4 identification points according to the 2002/657/EC decision). Thus, four
227
SRM transitions per target compound were monitored and optimized using Water’s IntelliStart system
228
(Table 2). Only two transitions were selected and monitored for internal standard compounds. Details
229
of this acquisition method are reported in Table 2, with the corresponding values of both optimized
230
parameters, i.e. cone voltage and collision energy. The acquisition method was divided in four different
231
time windows in order to optimize the sensitivity of the developed method, especially for the dwell time
232
set up. The different acquisition windows allowed the analysis of the compounds eluted in the range
233
[2-3 min], [3.5-4.3 min], [4.5-5.5] and [5.5-9 min]. The results illustrated in Figure 1 (1 µg L
234
spiked sample) indicated that the identification of all the molecules of interest was easy and
235
unambiguous even if the difference in sensitivity between all the compounds was quite significant,
236
especially for 2-methyl-9-fludrocortisone for which a lower factor of response (10 times) was
237
observed in comparison with flunisolide.
an
us
cr
ip t
201
pt
ed
M
-1
Ac ce
-1
-1
milk
Page 8 of 21
238 3.2
239
Validation
As mentioned above, four diagnostic signals were monitored for each target glucocorticoid for
241
unambiguous identification purpose. Nevertheless, only the two main SRM transitions were used for
242
quantification. A previous step consisted in selecting the most mimetic internal standard for each
243
molecule of interest. This work was carried out by injecting standards at different concentration levels.
244
The signal responses were reported to the different internal standards and the calibration curves were
245
built and compared. Therefore, the quantification of all molecules were performed using the most
246
adequate internal standard defined during the validation process, except for three compounds
247
(dexamethasone, prednisolone and triamcinolone acetonide) which quantification was directly carried
248
out by isotopic dilutions with corresponding ISs. As a result, the signals of the following molecules:
249
amcinonide,
250
isoflupredone, 1-dehydrocortexolone, 16-methylcortexolone and 2-methyl-9-fludrocortisone were
251
reported to fludrocortisone; betamethasone and dexamethasone signals were reported to
252
dexamethasone-d4, cortisol, cortisone, flumethasone, flunisolide, methylprednisolone, prednisolone
253
and prednisone signals were reported to prednisolone-d6. Finally triamcinolone acetonide was
254
reported to its deuterated labelled analogue.
budesonide,
desoxymethasone,
halcinonide
an
M
255 256
flurandrenolide,
us
beclomethasone,
cr
ip t
240
Specificity
The specificity of the method was demonstrated on the basis of the analysis of twenty representative
258
blank milk samples (skimmed, semi-skimmed and full fat). No interference peaks from coeluted or
259
endogenous compounds were observed at the expected retention times of all the 20 compounds of
260
interest, while the added internal standards could be detected. Both endogenous compounds, i.e.
261
cortisol and cortisone, were identified in all samples at various levels of concentrations, nevertheless,
262
the average background concentrations of cortisone and cortisol were evaluated respectively to 0.01
263
and 0.36 µg L and were taken into account during the validation process.
264
The specificity of the method was then calculated according to the determination of the average noise
265
response measured in the range of the expected retention times and the corresponding standard
266
deviation, reported to the signal of the internal standard (table 3). Therefore, the specificity of the
267
method was demonstrated with the absence of any significant signal at the expected retention times
268
for the 18 molecules (except for the two endogenous molecules) in the 20 selected representative milk
269
samples.
Ac ce
-1
pt
ed
257
270 271
Linearity
272
The linearity of the developed method was calculated for each of the 20 glucocorticoids of interest.
273
This parameter was evaluated by performing calibration curves using blank matrix, which consisted in
274
a mixture of the 20 selected milk samples used for the specificity evaluation. The blank samples were
275
fortified at twelve different levels of concentrations within a range [0.01 - 5 µg L ]. The calibration
276
curves were obtained by plotting the relative area ratios (analyte / corresponding internal standard)
277
versus the analyte concentrations. The intercept was forced to the mean of the relative intensity
-1
Page 9 of 21
278
obtained for the blank samples. As a result, all the calculated correlation coefficients for each
279
individual molecule were above 0.98 which was defined as acceptance criteria within the study (Table
280
3). Moreover, this procedure used for the linearity evaluation confirmed the choice of the internal
281
standard selected for quantification purposes.
282 283
Trueness and precision Since no certified materials were available for the determination of the selected glucocorticoids in milk
285
samples, the trueness of the method could not be evaluated. Regarding the precision parameter, both
286
repeatability and within-laboratory reproducibility were determined. For this purpose, the twenty blank
287
milk samples were fortified for each molecule at a level of concentration close to the estimated
288
detection capability. Repeatability was calculated on the basis of the coefficient of variation obtained
289
for the twenty repetitions of the fortified milk samples. The observed repeatability was considered as
290
acceptable in the range 1 to 30% and with a maximum relative standard deviation of 30% for cortisone
291
and 1-dehydrocortexolone. Moreover, the same extractions performed under reproducibility conditions
292
provided a maximum coefficient of 30%. For these series of analyses, the number of identification
293
points was higher than 4 IP, at least in 95% of the extracted milk samples. Moreover, the precision of
294
the method was also evaluated according to the retention times observed. Then, the relative retention
295
times (RTT) were determined by calculating the ratio of the analyte retention time and the retention
296
time of the corresponding internal standard. It should be noted that the relative retention times were
297
very repeatable as the relative standard deviations (RSD) were less than 0.25% for all the
298
glucocorticoids. Therefore, all RSD were lower than the acceptance criteria of 2.5% (EU requirements
299
2002/657/EC [23]).
ed
M
an
us
cr
ip t
284
300 301
Decision limit (CC) and detection capability (CC) The decision limit (CC) and the detection capability (CC) were calculated according to both series of
303
analysis, i.e. the analysis of the 20 blank milk samples and the analysis of the same samples fortified
304
to the estimated detection capability levels. The CC and CC obtained for each glucocorticoid and
305
each diagnostic signals are reported in Table 3. As a result, decision limits and detection capabilities
306
(Table 3) ranged between 0.001 and 0.363 µg L , which allowed a very efficient control at low trace
307
level for a potential illegal use of these substances. An additional step of this validation protocol
308
consisted in checking the ability of the method to detect the molecules of interest when a
309
representative milk sample, i.e. the mixture of the 20 milk samples used for the validation process,
310
was spiked to their respective CC level.
311
Figure 2 illustrates two examples (flunisolide and prednisone) of this additional validation step. Fig. 2a
312
illustrates the example of flunisolide for which CC and CC values have been calculated respectively
313
to 0.007 and 0.015 µg L for the first transition and to 0.07 to 0.020 for the second transition. The ion
314
chromatograms
315
(333.3>299.1)) and the two diagnostic transitions of flunisolide (374.9>184.9 and 374.9>312.9) in the
316
milk sample spiked at 0.007 µg L with flunisolide. The signal/noise ratio is observed in adequation
Ac ce
pt
302
-1
-1
presented
correspond
to
the
internal
standard
transition
(prednisolone-d6
-1
Page 10 of 21
317
with the calculated performances because the spiked milk sample provided a S/N ratio around 3 for
318
the second transition.
319
Identically, the example of prednisone illustrated in Fig. 2b confirmed the performances evaluated
320
during the validation process. For this targeted molecule, CC and CC have been calculated
321
respectively to 0.002 and 0.003 µg L
322
second transition. The ion chromatograms illustrate the signal of the internal standard and both
323
transitions for prednisone (416.8>327.0 and 356.9>327.0). Milk sample has been fortified at a
324
concentration of 0.02 µg L which corresponds to the determination of the validation performances.
325
One again, S/N ratio observed for the second transition is in accordance with calculated method
326
performances and makes it easy the detection of 50% of milk samples spiked to CC levels.
-1
for the first transition and to 0.014 and 0.028 µg L
for the
ip t
-1
cr
327 328
-1
Ruggedness
Possible factors that could influence the results have been identified during the development of the
330
analytical procedure. The stability of standard solutions, the composition of samples (skimmed, semi-
331
skimmed and whole milks), the origin of milk (ovine or bovine) and minor changes in the standard
332
operating procedure (operator, solvent and materials batches, elution volume) have been identified as
333
“critical points”.
an
334
us
329
The stability of analytical standards was checked, especially for both triamcinolone and methylprednisolone reported respectively as thermosensitive and
336
photosensitive [16]. Therefore, methylprednisolone was prepared in an amber glass bottle
337
and stored at -20°C, in dark. The stock solution were prepared and tested at the beginning
338
and at the end of the work. Additional verifications (3) were also preformed during the
339
development and validation process. The resulting variation of stability of this targeted
340
molecule was determined to 10%.
341
Regarding triamcinolone acetonide, no degradation has been observed under the
342
described analytical conditions. Therefore, only one peak was observed for this targeted
343
compound. Moreover, no significant variations have been observed concerning the
345 346 347 348 349
ed
pt
Ac ce
344
M
335
stability of the stock standard solutions of the others 18 glucocorticoids.
Minor changes have been also investigated during the validation process.
Therefore, three different operators have been implicated, different solvent and materials batches have been tested and some variations have been implemented to the method validation, such as a variation of 20% of the elution volume (solid phase extraction). Finally, both composition (skimmed, semi-skimmed and whole milk) and origin (bovine and
350
ovine) of the matrices have been investigated. The obtained results were integrated to the
351
reproducibility parameter which provide a maximum coefficient of 30% (for cortisone and
352
1-dehydrocortexolone), as it was observed for experiments realized under repeatability
353
conditions.
354 355 356
Evaluation of the method / application
Page 11 of 21
357
The method was assessed for the analysis of thirty milk samples originating from the national control
358
plan. The 20 validated glucocorticoids were successfully monitored and no suspicious samples could
359
be declared. Moreover, the robustness of the method was achieved insofar as the method provided
360
equivalent results, in terms of internal standard responses (peak intensities, retention times, average
361
noise response), to those obtained for the validation process.
362
Furthermore, the adequacy of the developed method regarding the MRL regulated molecules, i.e.
363
dexamethasone, betamethasone, methylprednisoloneand prednisolone has been checked. The
364
representative milk sample was fortified with 0.1 µg L of the three regulated glucocorticoids, which
365
represented MRL/3 for dexamethasone and betamethasone, MRL/20 for methylprednisolone and
366
MRL/60 for prednisolone. The results presented in Figure 3 illustrate the efficiency and the
367
performance of the method regarding the determination of these compounds at the target regulatory
368
levels of concentrations. Nevertheless, an additional validation process should be carried out for MRL
369
confirmation and quantification purposes.
us
cr
ip t
-1
370 4- Conclusions
an
371
An efficient UHPLC-MS/MS method has been developed and validated according to 2002/657/EC
373
requirements for the screening and the confirmation of an exhaustive list of twenty glucocorticoids in
374
bovine milk samples. The sample preparation combined to the use of a recent and high performance
375
mass spectrometer provided good specificity and sensitivity for all the target compounds. For the four
376
MRL
377
betamethasone, the developed method was able to detect these compounds at a concentration level
378
far below the regulatory limit. Moreover, the use of an UHPLC-MS/MS instrument enabled a short
379
runtime programming (10 minutes) while good peak separation and resolution were observed.
380
Therefore, the evaluated method performances have been determined as highly suitable for the
381
control of a potential illegal use of these molecules and have been successfully implemented on
382
samples originating from a national control plan. A specific validation for quantification purposes of the
383
four regulated MRL compounds in milk (dexamethasone, betamethasone, methylprednisolone and
384
prednisolone) currently being carried out and will be published in a near future.
i.e.
prednisolone,
methylprednisolone,
dexamethasone
and
pt
ed
glucocorticoids,
Ac ce
385
regulated
M
372
Page 12 of 21
385
418 419 420 421
[11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]
ip t
cr
us
[10]
an
[9]
M
[7] [8]
ed
[5] [6]
Council Directive 96/22 of 29 April 1996, Off. J. Eur. Communities L125 (1996) 3. Commission Regulation No. 37/2010, Off. J. Eur. Union No. L15/1 (2010). A. Gentili, TrAC Trends in Analytical Chemistry 26 (2007) 595. P. Delahaut, P. Jacquemin, Y. Colemonts, M. Dubois, J. De Graeve, H. Deluyker, Journal of Chromatography B: Biomedical Sciences and Applications 696 (1997) 203. L. Cun, W. Yinliang, Y. Ting, Z. Yan, Journal of Chromatography A 1217 (2010) 411. E.M. Malone, G. Dowling, C.T. Elliott, D.G. Kennedy, L. Regan, Journal of Chromatography A 1216 (2009) 8132. X. Cui, B. Shao, R. Zhao, Y. Yang, J. Hu, X. Tu, Rapid Commun Mass Spectrom 20 (2006) 2355. Y. Deceuninck, E. Bichon, F. Monteau, J.-P. Antignac, B. Le Bizec, Analytica Chimica Acta 700 (2011) 137. O. Van Den Hauwe, M. Schneider, A. Sahin, C.H. Van Peteghem, H. Naegeli, J. Agric. Food Chem. 51 (2003) 326. F. Caretti, A. Gentili, A. Ambrosi, L.M. Rocca, M. Delfini, M.E. Di Cocco, G. D'Ascenzo, Anal Bioanal Chem 397 (2010) 2477. R. Draisci, C. Marchiafava, L. Palleschi, P. Cammarata, S. Cavalli, Journal of Chromatography B: Biomedical Sciences and Applications 753 (2001) 217. J.P. Antignac, B. Le Bizec, F. Monteau, F. André, Analytica Chimica Acta 483 (2003) 325. M. Cherlet, S. De Baere, P. De Backer, Journal of Chromatography B 805 (2004) 57. C. Baiocchi, M. Brussino, M. Pazzi, C. Medana, C. Marini, E. Genta, Chromatographia 58 (2003) 11. D. Chen, Y. Tao, Z. Liu, Z. Liu, Y. Wang, L. Huang, Z. Yuan, Food Addit Contam 27 (2010) 1363. J.-P. Antignac, B. Le Bizec, F. Monteau, F. Poulain, F. Andre, Journal of Chromatography B: Biomedical Sciences and Applications 757 (2001) 11. J. Chrusch, S. Lee, R. Fedeniuk, J.O. Boison, Food Addit Contam Part A Chem Anal Control Expo Risk Assess 25 (2008) 1482. Commission Decision 2002/657/EC, Off. J. Communities L221 (2002) 8. J.-P. Antignac, B. Le Bizec, F. Monteau, F. Andre, Analytica Chimica Acta 483 (2003) 325. J. Vanden Bussche, L. Vanhaecke, Y. Deceuninck, K. Verheyden, K. Wille, K. Bekaert, B. Le Bizec, H.F. De Brabander, Journal of Chromatography A 1217 (2010) 4285. J.P. Antignac, B. Le Bizec, F. Monteau, F. Poulain, F. Andre, Rapid Commun Mass Spectrom 14 (2000) 33.
pt
[1] [2] [3] [4]
Ac ce
386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417
References
422 423 424 425 426
Page 13 of 21
427
Table and figure captions
428 Table 1: Names, structures and mono isotopic masses of the targeted glucocorticoids
430
Table 2: MS/MS parameters for each glucocorticoids of interest
431
Table 3: Validation parameters: linearity, repeatability, reproducibility, decision limits and detection
432
capabilities
433
Figure 1: LC-MS/MS chromatograms (SRM mode) of a bovine milk sample fortified with 1 µg L of
434
each molecule. A: isoflupredone, B: prednisone, C: prednisolone-d6, D: prednisolone, E:
435
fludrocortisone, F: cortisone, G: cortisol, H: 1-dehydrocortexolone, I: flumethasone, J: betamethasone,
436
K: methylprednisolone, L: dexamethasone-d4, M: dexamethasone, N: beclomethasone, O: 2-methyl-
437
9-fluorocortisone, P: triamcinolone acetonide-d6, Q: triamcinolone acetonide, R: flunisolide, S:
438
flurandrenolide, T: desoxymethasone, U: 16-methylcortexolone, V: budesonide, W: halcinonide, X:
439
amcinonide.
440
Figure 2: LC-MS/MS chromatograms (SRM mode) of a representative milk sample no-fortified (left) or
441
spiked (right) with 0.007 µg L of flunisolide (fig. 2A) or 0.02 µg L of prednisone (fig 2B).
442
Figure 3: LC-MS/MS chromatograms (SRM mode) of a representative milk sample fortified with 0.1
443
µg L of MRL regulated glucocorticoids. A: 363.2>309.1 (dexamethasone-d4), B: 451.2>361.2
444
(betamethasone and dexamethasone), C: 361.2>307.2 (betamethasone and dexamethasone), D:
445
333.3>299.1 (prednisolone-d6), E: 329.3>295.1 (prednisolone), F: 329.3>280.0 (prednisolone).
an
us
cr
-1
-1
M
-1
pt
ed
-1
Ac ce
446
ip t
429
Page 14 of 21
446
Table 1 O R 11
18
CH3
12 11
CH3 19
C
9
13
3
A
10
B R9 6
5
OH
17
14
R 21 R 16
16
15
8 7
Monoisotopic mass (amu)
Doublebond position
R6
R9
R11
ip t
O
20
D
1 2
21
R16
R21
Amcinonide
502.23
1-4
-
-F
-OH
-O-C5H8- O(17)
-OCO-CH3
Beclomethasone
408.17
1-4
-
-Cl
-OH
-CH3 ()
-OH
Betamethasone
392.20
1-4
-
-F
-OH
-CH3 ()
-OH
Budesonide
430.23
1-4
Cortisol
362.20
4
Cortisone
360.19
4
Desoxymethasone*
376.20
1-4
Dexamethasone
392.20
Flumethasone
410.19
Flunisolide
434.21
Flurandrenolide
436.22
Halcinonide
454.19
4
us
an
-OH
-O-C(H-C3H7)O(17)
-OH
-
-
-OH
-
-OH
-
-
=O
-
-OH
-
-F
-OH
-CH3 ()
-OH
1-4
-
-F
-OH
-CH3 ()
-OH
1-4
-F
-F
-OH
-CH3 ()
-OH
1-4
-F
-
-OH
-O-C(CH3)2-O(17)
-OH
4
-F
-
-OH
-O-C(CH3)2-O(17)
-OH
4
-
-F
-OH
-O-C(CH3)2-O(17)
-Cl
378.18
1-4
-
-F
-OH
-
-OH
Methylprednisolone
374.21
1-4
-CH3 ()
-
-OH
-
-OH
Prednisolone
360.19
1-4
-
-
-OH
-
-OH
358.17
1-4
-
-
=O
-
-OH
434.21
1-4
-
-F
-OH
-O-C(CH3)2-O(17)
-OH
1-dehydrocortexolone
344.20
1-4
-
-
-
-
-OH
16a-methylcortexolone
360.23
4
-
-
-
-CH3 ()
-OH
2a-methyl-9afludrocortisone**
392.20
4
-
-F
=O
-
-OH
Prednisone Triamcinolone acetonide
448 449
Ac ce
Isoflupredone
M
-
pt
-
ed
Molecule
cr
R6
447
* no OH group in C17 position ** CH3 in position 2.
Page 15 of 21
Table 2
Cortisol
Cortisone
Desoxymethasone
Dexamethasone
Flumethasone
20 20 20 20 14 14 14 14 20 40 40 20 20 20 20 20 16 16 22 16 12 26 26 12 24 24 24 24 20 40 40 20 16 24 16 16 28 28 28 28 20 20 20 20 16 16 36 36 18 18 18 18 46 20
Flunisolide
Flurandrenolide
Halcinonide
Isoflupredone Methylprednisolone
Collision energy (eV) 16 18 42 12 14 26 36 20 18 18 20 30 12 20 24 28 16 32 12 40 14 10 14 8 16 12 26 28 18 18 20 30 18 14 38 30 18 18 28 30 14 18 28 26 32 14 28 24 16 30 38 30 18 18
cr
ip t
560.9>357.0 560.9>481.0 560.9>341.0 560.9>441.1 466.8>377.0 466.8>297.0 466.8>137.0 466.8>341.0 451.2>361.2 361.2>307.2 361.2>325.1 451.2>307.2 489.0>357.0 489.0>339.0 489.0>187.0 489.0>295.0 420.9>331.1 420.9>297.0 360.9>331.1 420.9>283.9 418.9>329.0 358.8>329.0 358.8>301.0 418.8>359.0 434.8>355.0 434.8>375.0 374.8>121.0 434.8>121.0 451.2>361.2 361.2>307.2 361.2>325.1 451.2>307.2 468.8>379.0 408.8>379.0 468.8>305.0 468.8>325.0 374.9>184.9 374.9>312.9 374.9>134.9 374.9>157.0 495.0>377.0 495.0>359.0 495.0>187.0 495.0>315.0 512.9>433.0 512.9>453.0 452.9>309.0 452.9>397.0 436.9>347.0 436.9>293.0 436.9>277.9 436.9>311.0 343.2>309.2 433.3>343.2
us
Budesonide
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2
an
Betamethasone
Cone voltage (V)
M
Beclomethasone
Transition
ed
Amcinonide
Signal
pt
Analytes
Ac ce
451 452
Page 16 of 21
1-dehydrocortexolone
16-methylcortexolone
2-methyl-9fludrocortisone
453
Fludrocortisone Prednisolone-d6
455 456
1 2 1 2 1 2 1 2 1
363.2>309.1 363.2>327.1 349.2>313.1 349.2>295.1 333.3>299.1 425.4>333.3 498.9>419.0 498.9>375.0 355.2>255.1
40 40 40 40 40 18 24 24 34
M
Cone voltage (V)
Collision energy (eV) 20 18 20 22 20 20 20 14 14
Ac ce
Fluorometholone
454
Transition
pt
Triamcinolone acetonide-d6
Signal
ed
internal and external standards Dexamethasone-d4
24 32 24 20 24 18 14 8 14 6 16 14 22 18 14 36 38 8 12 8 46 38 8 16 32 10
ip t
Triamcinolone acetonide
cr
Prednisone
46 20 46 46 20 20 14 28 28 14 22 20 22 22 14 14 14 14 14 14 14 14 30 30 30 30
us
Prednisolone
343.2>294.1 433.3>309.1 329.3>280.0 329.3>295.1 419.2>295.1 419.2>329.2 416.8>327.0 356.9>327.0 356.9>299.0 416.8>357.0 492.9>413.0 492.9>375.0 492.9>337.0 492.9>357.0 402.8>313.1 402.8>297.0 402.8>282.0 402.8>343.1 418.8>329.1 418.8>359.0 418.8>285.0 418.8>313.0 390.8>361.0 390.8>313.0 390.8>137.0 390.8>333.0
an
3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4
Page 17 of 21
ip t R
2
Specificity
Slope (a)
Intercept (b)
Noise variability RSD%
Retention time
RTT average
CV%
CV% relative amplitude signal 1
us
Linearity
cr
Table 3
Identification Ratio average (T2-/T1)
Signal 1 RSD% of the ratio T2/T1
Signal 2
CC (µg L-1)
CC (µg L-1)
CC (µg L-1)
CC (µg L-1)
0.990
0.208
0.001
56.1
2.930
0.20
17.4
0.14
15.1
0.001
0.003
0.006
0.015
Beclomethasone
0.997
0.301
0.001
59.6
1.672
0.18
24.2
0.78
10.4
0.001
0.003
0.005
0.026
Betamethasone
0.994
4.234
0.007
52.5
0.975
0.12
10.1
0.74
11.0
0.001
0.001
0.002
0.003
Budesonide
0.999
0.184
0.001
105.6
2.556
0.18
21.4
0.43
17.2
0.003
0.005
0.019
0.037
Cortisol
0.990
1.838
0.557
21.2
1.053
0.00
16.5
0.48
3.2
0.172
0.360
0.179
0.381
Cortisone
0.993
0.983
0.006
72.3
1.014
0.19
14.4
0.26
13.0
0.011
0.017
0.014
0.024
Desoxymethasone
0.981
0.540
0.001
117.5
2.239
0.19
15.1
0.24
13.1
0.001
0.002
0.026
0.048
Dexamethasone
0.988
2.590
0.007
51.0
1.012
0.15
8.0
0.93
6.9
0.002
0.002
0.003
0.004
Flumethasone
0.999
1.212
0.000
101.3
1.490
0.13
14.4
0.23
9.1
0.001
0.001
0.003
0.005
Flunisolide
0.996
0.102
0.001
98.2
1.986
0.20
18.2
0.57
12.7
0.007
0.012
0.007
0.015
Flurandrenolide
0.993
0.660
0.000
106.2
2.106
0.18
16.5
0.47
18.0
0.001
0.002
0.002
0.004
Halcinonide
0.992
0.123
0.001
55.2
2.872
0.18
24.5
1.48
11.9
0.001
0.003
0.025
0.007
M
ep te
Ac c
Isoflupredone
an
Amcinonide
d
457
0.999
1.693
0.001
82.8
0.947
0.19
12.5
0.45
11.7
0.001
0.002
0.001
0.002
0.992
0.716
0.001
123.4
1.508
0.21
15.9
0.97
12.2
0.001
0.002
0.002
0.003
0.990
0.707
0.005
76.4
1.015
0.00
8.8
0.90
12.9
0.012
0.015
0.015
0.021
0.994
0.342
0.001
70.8
0.985
0.00
17.0
0.17
19.3
0.002
0.003
0.014
0.028
Triamcinolone acetonide
0.997
0.877
0.001
111.7
1.017
0.16
12.0
0.88
12.0
0.002
0.003
0.003
0.004
1-dehydrocortexolone
0.983
1.134
0.000
145.6
1.395
0.25
13.7
0.34
14.7
0.001
0.001
0.001
0.001
16-methylcortexolone
0.999
1.019
0.001
114.4
2.435
0.18
24.9
0.42
10.8
0.001
0.002
0.138
0.363
2-methyl-9-fludrocortisone
0.990
0.026
0.001
43.1
1.743
0.15
21.0
1.54
12.5
0.003
0.005
0.004
0.007
Methylprednisolone Prednisolone Prednisone
458 Page 18 of 21
458
Figure 1
459
251658240 A H B
D J
M
E K
cr
F
ip t
I
C
G
us
L
S
N Q R
O
M
P
an
U T
V
X W
463
pt
462
Ac ce
461
ed
460
Page 19 of 21
463 464
Figure 2
ed
467
pt
466
251658240
Ac ce
465
B
M
an
us
cr
ip t
A
Page 20 of 21
467 468
Figure 3
A
ip t
B
C
cr
D
us
E
F
251658240
an
469
Ac ce
pt
ed
M
470
Page 21 of 21
S0021-9673(13)00597-9 http://dx.doi.org/doi:10.1016/j.chroma.2013.04.019 CHROMA 354252
To appear in:
Journal of Chromatography A
Received date: Revised date: Accepted date:
29-1-2013 4-4-2013 8-4-2013
Please cite this article as: Y. Deceuninck, E. Bichon, F. Monteau, G. DervillyPinel, J.P. Antignac, B. Le Bizec, Fast and multiresidue determination of twenty glucocorticoids in bovine milk using Ultra High Performance Liquid Chromatography- tandem mass spectrometry, Journal of Chromatography A (2013), http://dx.doi.org/10.1016/j.chroma.2013.04.019 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
Highlights We developed a rapid 10 min UHPLC-MS/MS method for 20 corticosteroids in milk
3
We validated the method according to Commission Decision 2002/657/EC requirements
4
The methods allow low trace level (sub ppb) corticosteroids detection
5
Efficiency has been assessed on various bovine milks
ip t
2
Ac ce
pt
ed
M
an
us
cr
6
Page 1 of 21
6
Fast and multiresidue determination of twenty glucocorticoids in bovine milk
7
using
8
spectrometry.
Ultra
High Performance
Liquid Chromatography- tandem
mass
9 10 11
1
1
1
1
1,2
Y. Deceuninck *, E. Bichon , F. Monteau , G. Dervilly-Pinel , J.P. Antignac
and B. Le Bizec
1
15
LUNAM Université, Oniris, Laboratoire d’Etude des Résidus et Contaminants dans les Aliments
(LABERCA), Nantes, F-44307, France. 2
INRA, Nantes, F-44307, France.
*
Corresponding author
cr
14
1
us
13
ip t
12
16
18
an
17
Yoann DECEUNINCK, ONIRIS, École nationale vétérinaire, agroalimentaire et de l’alimentation
20
Nantes-Atlantique, Laboratoire d’Etude des Résidus et Contaminants dans les Aliments (LABERCA),
21
Atlanpole - La Chantrerie, BP 40706, Nantes, F-44307, France, tel : +33 2 40 68 78 80, fax : +33 2 40
22
68 78 78, email : [email protected].
ed pt Ac ce
23
M
19
Page 2 of 21
Abstract
24
Glucocorticoids constitute a class of molecules widely used in animal husbandry. Some of these
25
compounds are licensed for veterinary practices while their use for growth promoting purposes is
26
prohibited within the European Union. In order to ensure the respect of the legislation and consumers
27
safety, several methodologies have been proposed to monitor these substances in various products,
28
including edible matrices for which a regulatory limit has been set up (MRL). An extended range of
29
targeted analytes together with reduced time of analysis and cost are however still current challenges
30
regularly revisited according to the continuous technological improvements. In this context, the aim of
31
the present study was to develop and implement a new fast and multi-residue method based on
32
UHPLC-MS/MS for the determination of twenty glucocorticoids in bovine milk, included the screening
33
of the three regulated MRL compounds (dexamethasone, betamethasone and prednisolone). This
34
validated method authorises such multi-analyte measurement within a 10 minutes runtime while the
35
signal specificity is ensured through the SRM acquisition mode. Decision limits and detection
36
capabilities were calculated in the range [0.001 and 0.363] µg L , which allows a very efficient control
37
at low trace level for a potential illegal use of these substances. The performances obtained in terms
38
of application range, selectivity and sensitivity were found significantly improved in comparison to
39
other reported approaches either for screening or confirmation purposes: regarding linearity,
40
correlation coefficients were above 0.98 within the range [0.01-5.0 µg L ], repeatability and
41
reproducibility parameters ranged from 1 to 30 % with the maximum relative standard deviation (RSD)
42
observed for cortisone (30.1%). Stability of the stock solutions and minor changes in the standard
43
operating procedure have been included for the determination of ruggedness of the method.
44
Identification was systematically ensured according to 4 identification points, RSD of transitions ratio
45
(T2/T1) ranged from 3.2% and 19.3% and the RSD of the retention time was lower than 0.25%.
us
cr
ip t
23
-1
46
pt
ed
M
an
-1
Keywords
48
Corticosteroids, milk, UHPLC-MS/MS, Chemical food safety, mass spectrometry, MRL, Multi-residue
49
analysis
50
Ac ce
47
Page 3 of 21
50
1- Introduction
51 Glucocorticoids are a class of anti-inflammatory drugs (AIDs) that are widely used in human and
53
veterinary medicine for their anti-inflammatory, antipyretic, analgesic, anti-allergic and metabolic
54
properties [1]. Nevertheless, in livestock, the administration of those molecules results in a significant
55
gain of weight essentially due to an increase of water retention in tissues. Therefore, the use of
56
corticosteroids has a negative impact on meat quality and induces a decrease of the organoleptic
57
properties. Moreover, these substances may cause adverse effects on human health, including
58
hypertension, obesity or osteoporosis [2].
59
Due to additional properties of glucocorticoids such as the feed conversion rate improvement and their
60
synergetic effects when they are combined with other banned substances like steroids or beta
61
agonists [3-5], the use of these molecules in livestock is regulated in the European Union [6]. Some
62
compounds such as prednisolone, methylprednisolone, betamethasone and dexamethasone are
63
licensed for therapeutic diseases in breeding animals. Maximum Residue Limits (MRLs) have been set
64
for these molecules by the EU in several matrices such as muscle, liver, kidney and milk from different
65
animal species [7-8]. For milk, MRLs have been fixed at 6 µg L -1
an
us
cr
ip t
52
-1
for prednisolone, 2 µg L
-1
for
methylprednisolone and 0.3 µg L
for betamethasone and dexamethasone, whatever the animal
67
specie considered (bovine or ovine). Consequently, in the case of an authorized animal treatment with
68
one of these substances, a withdrawal period has to be respected between the end of treatment and
69
slaughter, while the use of other non-licensed analytes is prohibited. Regarding the control of
70
glucocorticoids misuse in livestock, standard operating procedures for both efficient screening and
71
confirmation purposes have to be implemented to ensure the respect of the legislation and more
72
generally the consumer’s safety.
73
In this context, many different standard operating procedures based on gas chromatography (GC) or
74
liquid chromatography (LC) coupled to mass spectrometry (MS) for analysing anti-inflammatory drugs
75
in animal-food products have been proposed in literature and reviewed by Gentili [9]. GC-MS was first
76
used as an adequate technique for screening and confirming glucocorticoids compounds in complex
77
biological matrices (milk, liver or urine) at trace levels, with LOD established around 0.25 µg L for
78
several analytes (triamcinolone, betamethasone, cortisol, flumethasone, desoxycortisone, 6-
79
methylprednisolone, dexamethasone, prednisolone, isoflupredone) [10]. However, while low limits of
80
detection were obtained, the derivatization step which was necessary to enhance the volatility of the
81
target molecules was considered as time-consuming. A lack of specificity for epimeric compounds
82
such as dexamethasone and betamethasone was also found sometimes an issue [11]. On the other
83
hand, the development of LC-MS and LC-MS
84
atmospheric-pressure ionisation (APCI) or atmospheric pressure photoionization (APPI) interfaces are
85
considered as powerful alternative to the use of GC-MS. The performances of these techniques have
86
demonstrated a high efficiency in terms of specificity and sensitivity. Nowadays, the use of LC-MS/MS
87
has become the analytical technique of choice for monitoring corticosteroids, especially for multi-
88
residue analysis at trace levels in complex matrices. More recently, some authors have reported the
Ac ce
pt
ed
M
66
-1
n
techniques using electrospray ionisation (ESI),
Page 4 of 21
advantages of last generation of ultra high performance chromatography (UHPLC) coupled to tandem
90
mass spectrometry in terms of chromatographic resolution, sensitivity and shortened run times [12-14].
91
Most of these studies have however focused on a limited number of compounds. Regarding milk as
92
matrix of interest, some authors have proposed procedures for the determination of 1 to 17
93
glucocorticoids in milk [11-13, 15-17]. In parallel, other have reported the analysis of a range of 1 to 11
94
molecules in liver and/or edible tissues [14-15, 18-24].
95
Most of these analytical methods have been validated according to the European requirements fixed
96
in the 2002/657/EC decision [25]. The reported decision limits (CC) and detection capabilities (CC)
97
in milk are compounds dependant to a large extend but also depend on the standard operating
98
procedure used. For example, Cui et al. [13] reported limits of quantification (LOQ) of 17
99
corticosteroids in milk and eggs ranging from 0.04 to 1.27 µg kg respectively for flumethasone and
100
fluorometholone. Furthermore, Caretti et al. [16] reported CC for 9 analytes ranging 0.05 to 0.74 µg
101
kg respectively for cortisone acetate and flumethasone.
102
In that context, the aim of the present study was to develop a method for screening and confirming in
103
milk an extended range of glucocorticoids that may be used as growth promoting agents. This method
104
comprises the analysis of 20 molecules: amcinonide, beclomethasone, betamethasone, budesonide,
105
cortisol, cortisone, desoxymethasone, dexamethasone, flumethasone, flunisolide, flurandrenolide,
106
halcinonide,
107
dehydrocortexolone, 16-methylcortexolone, 2-methyl-9-fluorocortisone (table 1). Fludrocortisone
108
and three deuterated internal standards, i.e dexamethasone-d4, prednisolone-d6 and triamcinolone
109
acetonide-d6 were also included for accurate quantification according to the isotope dilution method.
110
The proposed standard operating procedure was based on a purification step previously described by
111
Deceuninck et al. [14] from which an additional step consisting in a protein precipitation with acetone
112
was added. An ultra high performance liquid chromatography coupled to a triple quadrupole mass
113
spectrometer (UHPLC-MS/MS) was used for monitoring targeted compounds, and the method was
114
validated according to the Commission Decision 2002/657 criteria [25]. Through the implementation of
115
an efficient standard operating procedure, the developed analytical method enables a high-throughput
116
analysis of corticosteroids in milk with very low associated decision limits and detection capabilities.
117
This analytical method was successfully implemented to the analysis of milk samples coming from the
118
French national control plan.
120
cr
-1
an
methylprednisolone,
prednisolone,
triamcinolone
acetonide,
1-
pt
ed
M
isoflupredone,
us
-1
Ac ce
119
ip t
89
2- Experimental
121 122
2.1- materials and reagents
123
Analytical grade cyclohexane, ethyl acetate, diethylether, isopropanol, acetone and methanol, HPLC
124
grade acetonitrile, as well as solid-phase extraction cartridges (SPE C18: 2 g and silica: 1g) were
125
purchased from Carlo Erba Reactifs SDS (Val de Reuil, France). Sodium acetate and sodium
Page 5 of 21
126
carbonate were obtained from Sigma (St. Louis, MO, USA) and Merck (Darmstadt, Germany)
127
respectively. Enzymatic preparation was a purified lyophilized extract from Helix pomatia (Sigma, St.
128
Louis, MO, USA), dissolved in water (50,000 IU). Ultrapure water was purified using a Milli-Q osmosis
129
system from Millipore (Milford, MA, USA).
130
2.2- reference substances Amcinonide, beclomethasone, betamethasone, cortisol, cortisone, dexamethasone, fludrocortisone,
132
flumethasone, flunisolide, halcinonide, methylprednisolone, prednisolone and triamcinolone acetonide
133
were purchased from Sigma-Aldrich (St. Louis, MO, USA). Budesonide, desoxymethasone,
134
flurandrenolide,
135
fluorocortisone were obtained from Steraloids Inc. Ltd (London, England). Dexamethasone-d4 and
136
prednisolone-d6 (used as internal standards) were purchased from Cluzeau Info Labo (Courbevoie,
137
France) and triamcinolone acetonide- d6 was obtained from RIKILT (Wagenningen, The Netherlands).
138
Stock standard solutions of each molecule were prepared in methanol at a concentration of 1 mg.mL .
139
Working solutions were obtained by tenfold successive dilution in methanol at concentrations from 100
140
ng µL to 0.1 ng µL . All the standard solutions were stored at -20°C, in the dark.
16-methylcortexolone,
2-methyl-9-
-1
us
cr
1-dehydrocortexolone,
-1
-1
an
141
isoflupredone,
ip t
131
2.3- milk samples
Bovine and goat milk samples used for both method development and validation were purchased from
143
a local supermarket. In this way, skimmed and semi-skimmed milk, as well as whole milk were
144
included in the study. The developed and validated method was then applied to a second set of milk
145
samples originated from the French national control plan.
ed
2.4- standard operating procedure
pt
146
M
142
For each milk sample to be analysed, an initial volume of 10 mL was prepared in a polypropylene
148
tube. A protein precipitation was then performed by adding 8 mL of acetone. After evaporation of the
149
organic layer, an enzymatic hydrolysis was carried out at 50°C, during 4 h, using a purified -
150
glucuronidase, extracted from Helix pomatia, in order to deconjugate phase II metabolites (glucuronide
151
and sulfate forms) that might be present in milk samples coming from treated animals. Then, two
152
successive solid phase extractions steps were performed. The first one was carried out using a non-
153
polar (reverse) stationary phase column (C18) previously activated successively with 10 mL of
154
methanol and 10 mL of water. After loading the sample, the column was washed with 5 mL of water
155
and 5 mL of cyclohexane. Then, the analytes were eluted with 6 mL of diethylether. After evaporating
156
the extracts, a washing step of the aqueous layer was performed using sodium carbonate and
157
diethylether. A second SPE purification step was performed using a polar (normal) stationary phase
158
(SiOH). The column was first conditioned with at least 20 mL of cyclohexane. After loading the extract,
159
the column was rinsed with 5 mL of a mixture of cyclohexane / ethyl acetate (50:50, v/v) before elution
160
with a 10 mL ethyl acetate / isopropanol (90:10, v/v) mixture. The extracts were evaporated under N2
161
(40°C), redissolved in 50 µL of the mixture water/methanol (60:40, v/v) + 0.5% of acetic acidand
162
transferred into chromatographic vials for injection and analysis.
Ac ce
147
Page 6 of 21
163
2.5- chromatographic conditions ®
164
The LC separation was achieved on a Waters Acquity UPLC system (Milford, MA, USA), equipped
165
with an Acquity BEH C18 column (1.7 µm, 2.1×100 mm) maintained at 60°C. The LC mobile phases
166
consisted in 0.5% acetic acid in water (solvent A) and 0.5% acetic acid in acetonitrile (solvent B). A
167
flow rate of 0.6 mL min was applied. The injection volume was 2 µL. 2.6- detection parameters
ip t
168
-1
Detection was carried out using a Xevo TQ MS instrument (Waters, Milford, MA, USA), operating in
170
the negative Electrospray Ionisation mode (ESI-). MS data were acquired using the Selected Reaction
171
Monitoring mode (SRM). Capillary voltage was set at 3 kV, source temperature at 150°C,
172
desolvatation temperature at 500°C, desolvatation gas (N2) at 920 L h and collision gas flow at 0.15
173
mL min . Diagnostic SRM transitions were first generated using Waters’Intellistart
174
parameters were then optimized individually for each diagnostic signal. Data acquisition and data
175
processing were performed using MassLynx, version 4.1 software.
TM
software. All the
us
-1
2.7- method validation
an
176
-1
cr
169
Twenty representative bovine milk samples (non-fat, semi-skimmed and unskimmed milk) were
178
selected as blank materials after a preliminary analysis. In all these samples, internal standards were
179
added at a concentration of 0.3 µg L
180
prednislone-d6 and triamcinolone-d6.
181
The applied method validation guideline fitted with the 2002/657/EC requirements and was based on
182
the protocol previously described by Antignac et al. and Vanden Bussche et al. [19,26]. Several
183
parameters, such as specificity, repeatability, linearity, ruggedness, decision limit and detection
184
capability were assessed. The decision limit (CC) was calculated (equation 1) using the standard
185
deviation of the noise amplitude expressed to the corresponding internal standard (δN) and the slope
186
obtained from the calibration curve (a) established from 8 points ranging from 0.01 to 5.0 µg L . The
187
concentrations taken in account depended on the results obtained for each targeted compounds: for
188
example, the calibration curves of cortisol and isoflupredone were carried out according to
189
concentrations ranging from 0.01 to 0.4 µg L
190
capability (CC) was determined (equation 2) using the standard deviation (δS) of the signal amplitude
191
obtained from 20 spiked blank samples expressed to the decision limit level (CC obtained from
192
equation 1.
193
Equation 1: CC=(2.33×δN)/a
194
Equation 2: CC=CC+1.64×δS.
195
Calculated detection capability (CC) and decision limit (CCare applicable to corticosteroids
196
considered as forbidden substances; for MRL substances, these values must be re-assessed in a
197
different context.
-1
for dexamethasone-d4 and 1.0 µg L
for fludrocortisone,
Ac ce
pt
ed
-1
M
177
-1
-1
-1
and 0.2 to 5.0 µg L , respectively. The detection
198
Page 7 of 21
199 200
3- Results and discussion
3.1
UHPLC conditions and MS detection
Preliminary experiments were performed in order to optimize the chromatographic conditions used for
202
the separation of the twenty glucocorticoids of interest. A first evaluation of several column candidates
203
(Acquity BEH C18, 1.7 µm, 2.1×100 mm, Acquity BEH C18, 1.7 µm, 2.1×50 mm, Acquity BEH Phenyl,
204
1.7 µm, 2.1×100 mm, Acquity BEH RP C18 Shield, 1.7 µm, 2.1×50 mm and Hypersil Gold 1.9 µm,
205
2.1×100 mm) was carried out and the Acquity BEH C18, 1.7 µm, 2.1×100 mm column was finally
206
selected according to its good separation performances obtained for the different molecules,
207
especially for dexamethasone and betamethasone isomers. The gradient using water + 0.5% acetic
208
acid as solvent A and acetonitrile + 0.5 % of acetic acid as solvent B was selected and optimized
209
according to our experience background [27]. The initial conditions were set at 95% of solvent A and
210
5% of solvent B for the first 0.5 min, phase B was increased linearly to 25% in 0.5 min maintained for 4
211
min, then increased to 50% in 2 min and maintained for 1.5 min, then increased to 100% in another
212
0.1 min for 0.4 min, and finally returned to the initial composition in 0.2 min. the column was then
213
equilibrated for 0.8 min before the next injection. Then several levels of solvents were programmed
214
until a level at 100% of acetonitrile + 0.5% of acetic acid at 7 min was reached. The flow rate and the
215
temperature were set to 0.6 mL min and 60°C, respectively. The resulting chromatographic run was
216
10 minutes with a first (more polar) compound (isoflupredone) eluted at 2.60 min and a last (less polar)
217
one (amcinonide) eluted at 8.07 min (Figure 1). These optimized chromatographic conditions were
218
fixed for the method validation considering the adequate separation of the target glucocorticoids and
219
particularly the good resolution of the two isomers dexamethasone and betamethasone (Rs=1.04)
220
which is usually reported as an issue and was one of the objectives of the present method
221
development. Figure 1 shows an example of typical chromatographic traces obtained for a
222
representative milk sample, i.e. corresponding to a pool of the twenty blank samples selected for the
223
validation process, spiked with 1 µg L of each molecule. All these targeted molecules were detected
224
in the range [2.60 – 8.07 min]. The Selected Reaction Monitoring (SRM) was used as the adequate
225
acquisition mode for reaching high confidence level in terms of unambiguous identification of target
226
analytes (i.e. a minimum of 4 identification points according to the 2002/657/EC decision). Thus, four
227
SRM transitions per target compound were monitored and optimized using Water’s IntelliStart system
228
(Table 2). Only two transitions were selected and monitored for internal standard compounds. Details
229
of this acquisition method are reported in Table 2, with the corresponding values of both optimized
230
parameters, i.e. cone voltage and collision energy. The acquisition method was divided in four different
231
time windows in order to optimize the sensitivity of the developed method, especially for the dwell time
232
set up. The different acquisition windows allowed the analysis of the compounds eluted in the range
233
[2-3 min], [3.5-4.3 min], [4.5-5.5] and [5.5-9 min]. The results illustrated in Figure 1 (1 µg L
234
spiked sample) indicated that the identification of all the molecules of interest was easy and
235
unambiguous even if the difference in sensitivity between all the compounds was quite significant,
236
especially for 2-methyl-9-fludrocortisone for which a lower factor of response (10 times) was
237
observed in comparison with flunisolide.
an
us
cr
ip t
201
pt
ed
M
-1
Ac ce
-1
-1
milk
Page 8 of 21
238 3.2
239
Validation
As mentioned above, four diagnostic signals were monitored for each target glucocorticoid for
241
unambiguous identification purpose. Nevertheless, only the two main SRM transitions were used for
242
quantification. A previous step consisted in selecting the most mimetic internal standard for each
243
molecule of interest. This work was carried out by injecting standards at different concentration levels.
244
The signal responses were reported to the different internal standards and the calibration curves were
245
built and compared. Therefore, the quantification of all molecules were performed using the most
246
adequate internal standard defined during the validation process, except for three compounds
247
(dexamethasone, prednisolone and triamcinolone acetonide) which quantification was directly carried
248
out by isotopic dilutions with corresponding ISs. As a result, the signals of the following molecules:
249
amcinonide,
250
isoflupredone, 1-dehydrocortexolone, 16-methylcortexolone and 2-methyl-9-fludrocortisone were
251
reported to fludrocortisone; betamethasone and dexamethasone signals were reported to
252
dexamethasone-d4, cortisol, cortisone, flumethasone, flunisolide, methylprednisolone, prednisolone
253
and prednisone signals were reported to prednisolone-d6. Finally triamcinolone acetonide was
254
reported to its deuterated labelled analogue.
budesonide,
desoxymethasone,
halcinonide
an
M
255 256
flurandrenolide,
us
beclomethasone,
cr
ip t
240
Specificity
The specificity of the method was demonstrated on the basis of the analysis of twenty representative
258
blank milk samples (skimmed, semi-skimmed and full fat). No interference peaks from coeluted or
259
endogenous compounds were observed at the expected retention times of all the 20 compounds of
260
interest, while the added internal standards could be detected. Both endogenous compounds, i.e.
261
cortisol and cortisone, were identified in all samples at various levels of concentrations, nevertheless,
262
the average background concentrations of cortisone and cortisol were evaluated respectively to 0.01
263
and 0.36 µg L and were taken into account during the validation process.
264
The specificity of the method was then calculated according to the determination of the average noise
265
response measured in the range of the expected retention times and the corresponding standard
266
deviation, reported to the signal of the internal standard (table 3). Therefore, the specificity of the
267
method was demonstrated with the absence of any significant signal at the expected retention times
268
for the 18 molecules (except for the two endogenous molecules) in the 20 selected representative milk
269
samples.
Ac ce
-1
pt
ed
257
270 271
Linearity
272
The linearity of the developed method was calculated for each of the 20 glucocorticoids of interest.
273
This parameter was evaluated by performing calibration curves using blank matrix, which consisted in
274
a mixture of the 20 selected milk samples used for the specificity evaluation. The blank samples were
275
fortified at twelve different levels of concentrations within a range [0.01 - 5 µg L ]. The calibration
276
curves were obtained by plotting the relative area ratios (analyte / corresponding internal standard)
277
versus the analyte concentrations. The intercept was forced to the mean of the relative intensity
-1
Page 9 of 21
278
obtained for the blank samples. As a result, all the calculated correlation coefficients for each
279
individual molecule were above 0.98 which was defined as acceptance criteria within the study (Table
280
3). Moreover, this procedure used for the linearity evaluation confirmed the choice of the internal
281
standard selected for quantification purposes.
282 283
Trueness and precision Since no certified materials were available for the determination of the selected glucocorticoids in milk
285
samples, the trueness of the method could not be evaluated. Regarding the precision parameter, both
286
repeatability and within-laboratory reproducibility were determined. For this purpose, the twenty blank
287
milk samples were fortified for each molecule at a level of concentration close to the estimated
288
detection capability. Repeatability was calculated on the basis of the coefficient of variation obtained
289
for the twenty repetitions of the fortified milk samples. The observed repeatability was considered as
290
acceptable in the range 1 to 30% and with a maximum relative standard deviation of 30% for cortisone
291
and 1-dehydrocortexolone. Moreover, the same extractions performed under reproducibility conditions
292
provided a maximum coefficient of 30%. For these series of analyses, the number of identification
293
points was higher than 4 IP, at least in 95% of the extracted milk samples. Moreover, the precision of
294
the method was also evaluated according to the retention times observed. Then, the relative retention
295
times (RTT) were determined by calculating the ratio of the analyte retention time and the retention
296
time of the corresponding internal standard. It should be noted that the relative retention times were
297
very repeatable as the relative standard deviations (RSD) were less than 0.25% for all the
298
glucocorticoids. Therefore, all RSD were lower than the acceptance criteria of 2.5% (EU requirements
299
2002/657/EC [23]).
ed
M
an
us
cr
ip t
284
300 301
Decision limit (CC) and detection capability (CC) The decision limit (CC) and the detection capability (CC) were calculated according to both series of
303
analysis, i.e. the analysis of the 20 blank milk samples and the analysis of the same samples fortified
304
to the estimated detection capability levels. The CC and CC obtained for each glucocorticoid and
305
each diagnostic signals are reported in Table 3. As a result, decision limits and detection capabilities
306
(Table 3) ranged between 0.001 and 0.363 µg L , which allowed a very efficient control at low trace
307
level for a potential illegal use of these substances. An additional step of this validation protocol
308
consisted in checking the ability of the method to detect the molecules of interest when a
309
representative milk sample, i.e. the mixture of the 20 milk samples used for the validation process,
310
was spiked to their respective CC level.
311
Figure 2 illustrates two examples (flunisolide and prednisone) of this additional validation step. Fig. 2a
312
illustrates the example of flunisolide for which CC and CC values have been calculated respectively
313
to 0.007 and 0.015 µg L for the first transition and to 0.07 to 0.020 for the second transition. The ion
314
chromatograms
315
(333.3>299.1)) and the two diagnostic transitions of flunisolide (374.9>184.9 and 374.9>312.9) in the
316
milk sample spiked at 0.007 µg L with flunisolide. The signal/noise ratio is observed in adequation
Ac ce
pt
302
-1
-1
presented
correspond
to
the
internal
standard
transition
(prednisolone-d6
-1
Page 10 of 21
317
with the calculated performances because the spiked milk sample provided a S/N ratio around 3 for
318
the second transition.
319
Identically, the example of prednisone illustrated in Fig. 2b confirmed the performances evaluated
320
during the validation process. For this targeted molecule, CC and CC have been calculated
321
respectively to 0.002 and 0.003 µg L
322
second transition. The ion chromatograms illustrate the signal of the internal standard and both
323
transitions for prednisone (416.8>327.0 and 356.9>327.0). Milk sample has been fortified at a
324
concentration of 0.02 µg L which corresponds to the determination of the validation performances.
325
One again, S/N ratio observed for the second transition is in accordance with calculated method
326
performances and makes it easy the detection of 50% of milk samples spiked to CC levels.
-1
for the first transition and to 0.014 and 0.028 µg L
for the
ip t
-1
cr
327 328
-1
Ruggedness
Possible factors that could influence the results have been identified during the development of the
330
analytical procedure. The stability of standard solutions, the composition of samples (skimmed, semi-
331
skimmed and whole milks), the origin of milk (ovine or bovine) and minor changes in the standard
332
operating procedure (operator, solvent and materials batches, elution volume) have been identified as
333
“critical points”.
an
334
us
329
The stability of analytical standards was checked, especially for both triamcinolone and methylprednisolone reported respectively as thermosensitive and
336
photosensitive [16]. Therefore, methylprednisolone was prepared in an amber glass bottle
337
and stored at -20°C, in dark. The stock solution were prepared and tested at the beginning
338
and at the end of the work. Additional verifications (3) were also preformed during the
339
development and validation process. The resulting variation of stability of this targeted
340
molecule was determined to 10%.
341
Regarding triamcinolone acetonide, no degradation has been observed under the
342
described analytical conditions. Therefore, only one peak was observed for this targeted
343
compound. Moreover, no significant variations have been observed concerning the
345 346 347 348 349
ed
pt
Ac ce
344
M
335
stability of the stock standard solutions of the others 18 glucocorticoids.
Minor changes have been also investigated during the validation process.
Therefore, three different operators have been implicated, different solvent and materials batches have been tested and some variations have been implemented to the method validation, such as a variation of 20% of the elution volume (solid phase extraction). Finally, both composition (skimmed, semi-skimmed and whole milk) and origin (bovine and
350
ovine) of the matrices have been investigated. The obtained results were integrated to the
351
reproducibility parameter which provide a maximum coefficient of 30% (for cortisone and
352
1-dehydrocortexolone), as it was observed for experiments realized under repeatability
353
conditions.
354 355 356
Evaluation of the method / application
Page 11 of 21
357
The method was assessed for the analysis of thirty milk samples originating from the national control
358
plan. The 20 validated glucocorticoids were successfully monitored and no suspicious samples could
359
be declared. Moreover, the robustness of the method was achieved insofar as the method provided
360
equivalent results, in terms of internal standard responses (peak intensities, retention times, average
361
noise response), to those obtained for the validation process.
362
Furthermore, the adequacy of the developed method regarding the MRL regulated molecules, i.e.
363
dexamethasone, betamethasone, methylprednisoloneand prednisolone has been checked. The
364
representative milk sample was fortified with 0.1 µg L of the three regulated glucocorticoids, which
365
represented MRL/3 for dexamethasone and betamethasone, MRL/20 for methylprednisolone and
366
MRL/60 for prednisolone. The results presented in Figure 3 illustrate the efficiency and the
367
performance of the method regarding the determination of these compounds at the target regulatory
368
levels of concentrations. Nevertheless, an additional validation process should be carried out for MRL
369
confirmation and quantification purposes.
us
cr
ip t
-1
370 4- Conclusions
an
371
An efficient UHPLC-MS/MS method has been developed and validated according to 2002/657/EC
373
requirements for the screening and the confirmation of an exhaustive list of twenty glucocorticoids in
374
bovine milk samples. The sample preparation combined to the use of a recent and high performance
375
mass spectrometer provided good specificity and sensitivity for all the target compounds. For the four
376
MRL
377
betamethasone, the developed method was able to detect these compounds at a concentration level
378
far below the regulatory limit. Moreover, the use of an UHPLC-MS/MS instrument enabled a short
379
runtime programming (10 minutes) while good peak separation and resolution were observed.
380
Therefore, the evaluated method performances have been determined as highly suitable for the
381
control of a potential illegal use of these molecules and have been successfully implemented on
382
samples originating from a national control plan. A specific validation for quantification purposes of the
383
four regulated MRL compounds in milk (dexamethasone, betamethasone, methylprednisolone and
384
prednisolone) currently being carried out and will be published in a near future.
i.e.
prednisolone,
methylprednisolone,
dexamethasone
and
pt
ed
glucocorticoids,
Ac ce
385
regulated
M
372
Page 12 of 21
385
418 419 420 421
[11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]
ip t
cr
us
[10]
an
[9]
M
[7] [8]
ed
[5] [6]
Council Directive 96/22 of 29 April 1996, Off. J. Eur. Communities L125 (1996) 3. Commission Regulation No. 37/2010, Off. J. Eur. Union No. L15/1 (2010). A. Gentili, TrAC Trends in Analytical Chemistry 26 (2007) 595. P. Delahaut, P. Jacquemin, Y. Colemonts, M. Dubois, J. De Graeve, H. Deluyker, Journal of Chromatography B: Biomedical Sciences and Applications 696 (1997) 203. L. Cun, W. Yinliang, Y. Ting, Z. Yan, Journal of Chromatography A 1217 (2010) 411. E.M. Malone, G. Dowling, C.T. Elliott, D.G. Kennedy, L. Regan, Journal of Chromatography A 1216 (2009) 8132. X. Cui, B. Shao, R. Zhao, Y. Yang, J. Hu, X. Tu, Rapid Commun Mass Spectrom 20 (2006) 2355. Y. Deceuninck, E. Bichon, F. Monteau, J.-P. Antignac, B. Le Bizec, Analytica Chimica Acta 700 (2011) 137. O. Van Den Hauwe, M. Schneider, A. Sahin, C.H. Van Peteghem, H. Naegeli, J. Agric. Food Chem. 51 (2003) 326. F. Caretti, A. Gentili, A. Ambrosi, L.M. Rocca, M. Delfini, M.E. Di Cocco, G. D'Ascenzo, Anal Bioanal Chem 397 (2010) 2477. R. Draisci, C. Marchiafava, L. Palleschi, P. Cammarata, S. Cavalli, Journal of Chromatography B: Biomedical Sciences and Applications 753 (2001) 217. J.P. Antignac, B. Le Bizec, F. Monteau, F. André, Analytica Chimica Acta 483 (2003) 325. M. Cherlet, S. De Baere, P. De Backer, Journal of Chromatography B 805 (2004) 57. C. Baiocchi, M. Brussino, M. Pazzi, C. Medana, C. Marini, E. Genta, Chromatographia 58 (2003) 11. D. Chen, Y. Tao, Z. Liu, Z. Liu, Y. Wang, L. Huang, Z. Yuan, Food Addit Contam 27 (2010) 1363. J.-P. Antignac, B. Le Bizec, F. Monteau, F. Poulain, F. Andre, Journal of Chromatography B: Biomedical Sciences and Applications 757 (2001) 11. J. Chrusch, S. Lee, R. Fedeniuk, J.O. Boison, Food Addit Contam Part A Chem Anal Control Expo Risk Assess 25 (2008) 1482. Commission Decision 2002/657/EC, Off. J. Communities L221 (2002) 8. J.-P. Antignac, B. Le Bizec, F. Monteau, F. Andre, Analytica Chimica Acta 483 (2003) 325. J. Vanden Bussche, L. Vanhaecke, Y. Deceuninck, K. Verheyden, K. Wille, K. Bekaert, B. Le Bizec, H.F. De Brabander, Journal of Chromatography A 1217 (2010) 4285. J.P. Antignac, B. Le Bizec, F. Monteau, F. Poulain, F. Andre, Rapid Commun Mass Spectrom 14 (2000) 33.
pt
[1] [2] [3] [4]
Ac ce
386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417
References
422 423 424 425 426
Page 13 of 21
427
Table and figure captions
428 Table 1: Names, structures and mono isotopic masses of the targeted glucocorticoids
430
Table 2: MS/MS parameters for each glucocorticoids of interest
431
Table 3: Validation parameters: linearity, repeatability, reproducibility, decision limits and detection
432
capabilities
433
Figure 1: LC-MS/MS chromatograms (SRM mode) of a bovine milk sample fortified with 1 µg L of
434
each molecule. A: isoflupredone, B: prednisone, C: prednisolone-d6, D: prednisolone, E:
435
fludrocortisone, F: cortisone, G: cortisol, H: 1-dehydrocortexolone, I: flumethasone, J: betamethasone,
436
K: methylprednisolone, L: dexamethasone-d4, M: dexamethasone, N: beclomethasone, O: 2-methyl-
437
9-fluorocortisone, P: triamcinolone acetonide-d6, Q: triamcinolone acetonide, R: flunisolide, S:
438
flurandrenolide, T: desoxymethasone, U: 16-methylcortexolone, V: budesonide, W: halcinonide, X:
439
amcinonide.
440
Figure 2: LC-MS/MS chromatograms (SRM mode) of a representative milk sample no-fortified (left) or
441
spiked (right) with 0.007 µg L of flunisolide (fig. 2A) or 0.02 µg L of prednisone (fig 2B).
442
Figure 3: LC-MS/MS chromatograms (SRM mode) of a representative milk sample fortified with 0.1
443
µg L of MRL regulated glucocorticoids. A: 363.2>309.1 (dexamethasone-d4), B: 451.2>361.2
444
(betamethasone and dexamethasone), C: 361.2>307.2 (betamethasone and dexamethasone), D:
445
333.3>299.1 (prednisolone-d6), E: 329.3>295.1 (prednisolone), F: 329.3>280.0 (prednisolone).
an
us
cr
-1
-1
M
-1
pt
ed
-1
Ac ce
446
ip t
429
Page 14 of 21
446
Table 1 O R 11
18
CH3
12 11
CH3 19
C
9
13
3
A
10
B R9 6
5
OH
17
14
R 21 R 16
16
15
8 7
Monoisotopic mass (amu)
Doublebond position
R6
R9
R11
ip t
O
20
D
1 2
21
R16
R21
Amcinonide
502.23
1-4
-
-F
-OH
-O-C5H8- O(17)
-OCO-CH3
Beclomethasone
408.17
1-4
-
-Cl
-OH
-CH3 ()
-OH
Betamethasone
392.20
1-4
-
-F
-OH
-CH3 ()
-OH
Budesonide
430.23
1-4
Cortisol
362.20
4
Cortisone
360.19
4
Desoxymethasone*
376.20
1-4
Dexamethasone
392.20
Flumethasone
410.19
Flunisolide
434.21
Flurandrenolide
436.22
Halcinonide
454.19
4
us
an
-OH
-O-C(H-C3H7)O(17)
-OH
-
-
-OH
-
-OH
-
-
=O
-
-OH
-
-F
-OH
-CH3 ()
-OH
1-4
-
-F
-OH
-CH3 ()
-OH
1-4
-F
-F
-OH
-CH3 ()
-OH
1-4
-F
-
-OH
-O-C(CH3)2-O(17)
-OH
4
-F
-
-OH
-O-C(CH3)2-O(17)
-OH
4
-
-F
-OH
-O-C(CH3)2-O(17)
-Cl
378.18
1-4
-
-F
-OH
-
-OH
Methylprednisolone
374.21
1-4
-CH3 ()
-
-OH
-
-OH
Prednisolone
360.19
1-4
-
-
-OH
-
-OH
358.17
1-4
-
-
=O
-
-OH
434.21
1-4
-
-F
-OH
-O-C(CH3)2-O(17)
-OH
1-dehydrocortexolone
344.20
1-4
-
-
-
-
-OH
16a-methylcortexolone
360.23
4
-
-
-
-CH3 ()
-OH
2a-methyl-9afludrocortisone**
392.20
4
-
-F
=O
-
-OH
Prednisone Triamcinolone acetonide
448 449
Ac ce
Isoflupredone
M
-
pt
-
ed
Molecule
cr
R6
447
* no OH group in C17 position ** CH3 in position 2.
Page 15 of 21
Table 2
Cortisol
Cortisone
Desoxymethasone
Dexamethasone
Flumethasone
20 20 20 20 14 14 14 14 20 40 40 20 20 20 20 20 16 16 22 16 12 26 26 12 24 24 24 24 20 40 40 20 16 24 16 16 28 28 28 28 20 20 20 20 16 16 36 36 18 18 18 18 46 20
Flunisolide
Flurandrenolide
Halcinonide
Isoflupredone Methylprednisolone
Collision energy (eV) 16 18 42 12 14 26 36 20 18 18 20 30 12 20 24 28 16 32 12 40 14 10 14 8 16 12 26 28 18 18 20 30 18 14 38 30 18 18 28 30 14 18 28 26 32 14 28 24 16 30 38 30 18 18
cr
ip t
560.9>357.0 560.9>481.0 560.9>341.0 560.9>441.1 466.8>377.0 466.8>297.0 466.8>137.0 466.8>341.0 451.2>361.2 361.2>307.2 361.2>325.1 451.2>307.2 489.0>357.0 489.0>339.0 489.0>187.0 489.0>295.0 420.9>331.1 420.9>297.0 360.9>331.1 420.9>283.9 418.9>329.0 358.8>329.0 358.8>301.0 418.8>359.0 434.8>355.0 434.8>375.0 374.8>121.0 434.8>121.0 451.2>361.2 361.2>307.2 361.2>325.1 451.2>307.2 468.8>379.0 408.8>379.0 468.8>305.0 468.8>325.0 374.9>184.9 374.9>312.9 374.9>134.9 374.9>157.0 495.0>377.0 495.0>359.0 495.0>187.0 495.0>315.0 512.9>433.0 512.9>453.0 452.9>309.0 452.9>397.0 436.9>347.0 436.9>293.0 436.9>277.9 436.9>311.0 343.2>309.2 433.3>343.2
us
Budesonide
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2
an
Betamethasone
Cone voltage (V)
M
Beclomethasone
Transition
ed
Amcinonide
Signal
pt
Analytes
Ac ce
451 452
Page 16 of 21
1-dehydrocortexolone
16-methylcortexolone
2-methyl-9fludrocortisone
453
Fludrocortisone Prednisolone-d6
455 456
1 2 1 2 1 2 1 2 1
363.2>309.1 363.2>327.1 349.2>313.1 349.2>295.1 333.3>299.1 425.4>333.3 498.9>419.0 498.9>375.0 355.2>255.1
40 40 40 40 40 18 24 24 34
M
Cone voltage (V)
Collision energy (eV) 20 18 20 22 20 20 20 14 14
Ac ce
Fluorometholone
454
Transition
pt
Triamcinolone acetonide-d6
Signal
ed
internal and external standards Dexamethasone-d4
24 32 24 20 24 18 14 8 14 6 16 14 22 18 14 36 38 8 12 8 46 38 8 16 32 10
ip t
Triamcinolone acetonide
cr
Prednisone
46 20 46 46 20 20 14 28 28 14 22 20 22 22 14 14 14 14 14 14 14 14 30 30 30 30
us
Prednisolone
343.2>294.1 433.3>309.1 329.3>280.0 329.3>295.1 419.2>295.1 419.2>329.2 416.8>327.0 356.9>327.0 356.9>299.0 416.8>357.0 492.9>413.0 492.9>375.0 492.9>337.0 492.9>357.0 402.8>313.1 402.8>297.0 402.8>282.0 402.8>343.1 418.8>329.1 418.8>359.0 418.8>285.0 418.8>313.0 390.8>361.0 390.8>313.0 390.8>137.0 390.8>333.0
an
3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4
Page 17 of 21
ip t R
2
Specificity
Slope (a)
Intercept (b)
Noise variability RSD%
Retention time
RTT average
CV%
CV% relative amplitude signal 1
us
Linearity
cr
Table 3
Identification Ratio average (T2-/T1)
Signal 1 RSD% of the ratio T2/T1
Signal 2
CC (µg L-1)
CC (µg L-1)
CC (µg L-1)
CC (µg L-1)
0.990
0.208
0.001
56.1
2.930
0.20
17.4
0.14
15.1
0.001
0.003
0.006
0.015
Beclomethasone
0.997
0.301
0.001
59.6
1.672
0.18
24.2
0.78
10.4
0.001
0.003
0.005
0.026
Betamethasone
0.994
4.234
0.007
52.5
0.975
0.12
10.1
0.74
11.0
0.001
0.001
0.002
0.003
Budesonide
0.999
0.184
0.001
105.6
2.556
0.18
21.4
0.43
17.2
0.003
0.005
0.019
0.037
Cortisol
0.990
1.838
0.557
21.2
1.053
0.00
16.5
0.48
3.2
0.172
0.360
0.179
0.381
Cortisone
0.993
0.983
0.006
72.3
1.014
0.19
14.4
0.26
13.0
0.011
0.017
0.014
0.024
Desoxymethasone
0.981
0.540
0.001
117.5
2.239
0.19
15.1
0.24
13.1
0.001
0.002
0.026
0.048
Dexamethasone
0.988
2.590
0.007
51.0
1.012
0.15
8.0
0.93
6.9
0.002
0.002
0.003
0.004
Flumethasone
0.999
1.212
0.000
101.3
1.490
0.13
14.4
0.23
9.1
0.001
0.001
0.003
0.005
Flunisolide
0.996
0.102
0.001
98.2
1.986
0.20
18.2
0.57
12.7
0.007
0.012
0.007
0.015
Flurandrenolide
0.993
0.660
0.000
106.2
2.106
0.18
16.5
0.47
18.0
0.001
0.002
0.002
0.004
Halcinonide
0.992
0.123
0.001
55.2
2.872
0.18
24.5
1.48
11.9
0.001
0.003
0.025
0.007
M
ep te
Ac c
Isoflupredone
an
Amcinonide
d
457
0.999
1.693
0.001
82.8
0.947
0.19
12.5
0.45
11.7
0.001
0.002
0.001
0.002
0.992
0.716
0.001
123.4
1.508
0.21
15.9
0.97
12.2
0.001
0.002
0.002
0.003
0.990
0.707
0.005
76.4
1.015
0.00
8.8
0.90
12.9
0.012
0.015
0.015
0.021
0.994
0.342
0.001
70.8
0.985
0.00
17.0
0.17
19.3
0.002
0.003
0.014
0.028
Triamcinolone acetonide
0.997
0.877
0.001
111.7
1.017
0.16
12.0
0.88
12.0
0.002
0.003
0.003
0.004
1-dehydrocortexolone
0.983
1.134
0.000
145.6
1.395
0.25
13.7
0.34
14.7
0.001
0.001
0.001
0.001
16-methylcortexolone
0.999
1.019
0.001
114.4
2.435
0.18
24.9
0.42
10.8
0.001
0.002
0.138
0.363
2-methyl-9-fludrocortisone
0.990
0.026
0.001
43.1
1.743
0.15
21.0
1.54
12.5
0.003
0.005
0.004
0.007
Methylprednisolone Prednisolone Prednisone
458 Page 18 of 21
458
Figure 1
459
251658240 A H B
D J
M
E K
cr
F
ip t
I
C
G
us
L
S
N Q R
O
M
P
an
U T
V
X W
463
pt
462
Ac ce
461
ed
460
Page 19 of 21
463 464
Figure 2
ed
467
pt
466
251658240
Ac ce
465
B
M
an
us
cr
ip t
A
Page 20 of 21
467 468
Figure 3
A
ip t
B
C
cr
D
us
E
F
251658240
an
469
Ac ce
pt
ed
M
470
Page 21 of 21