Accepted Manuscript Title: Stir-membrane solid-liquid-liquid microextraction for the determination of parabens in human breast milk samples by ultra high performance liquid chromatography-tandem mass spectrometry Author: Roc´ıo Rodr´ıguez-G´omez Mercedes Rold´an-Piju´an Rafael Lucena Soledad C´ardenas Alberto Zafra-G´omez Oscar Ballesteros Alberto Naval´on Miguel Valc´arcel PII: DOI: Reference:
S0021-9673(14)00856-5 http://dx.doi.org/doi:10.1016/j.chroma.2014.05.071 CHROMA 355463
To appear in:
Journal of Chromatography A
Received date: Revised date: Accepted date:
26-3-2014 19-5-2014 28-5-2014
Please cite this article as: R. Rodr´iguez-G´omez, M. Rold´an-Piju´an, R. Lucena, S. C´ardenas, A. Zafra-G´omez, O. Ballesteros, A. Naval´on, M. Valc´arcel, Stir-membrane solid-liquid-liquid microextraction for the determination of parabens in human breast milk samples by ultra high performance liquid chromatography-tandem mass spectrometry, Journal of Chromatography A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.071 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.
Stir-membrane solid-liquid-liquid microextraction for the determination of parabens in human breast milk samples by ultra high performance liquid chromatography-tandem mass spectrometry
‡
Research Group of Analytical Chemistry and Life Sciences, Department of Analytical Chemistry, Campus of Fuentenueva, University of Granada, E-18071 Granada, Spain. Department of Analytical Chemistry, Institute of Fine Chemistry and Nanochemistry, Marie Curie Building, Campus of Rabanales, University of Córdoba, E-14071 Córdoba, Spain.
te
d
M
an
us
* Corresponding author: (tel/fax) +34-957-218-616; (e-mail)
[email protected] cr
†
ip t
Rocío Rodríguez-Gómez†, Mercedes Roldán-Pijuán‡, Rafael Lucena‡, Soledad Cárdenas‡, Alberto Zafra-Gómez†, Oscar Ballesteros†, Alberto Navalón†, Miguel Valcárcel*‡
Ac ce p
1 2 3 4 5 6 7 8 9 10 11 12 13 14
1
Page 1 of 33
ABSTRACT: In this article, stir-membrane solid-liquid-liquid microextraction (SM-SLLME)
15
is tailored for the analysis of solid matrices and it has been evaluated for the determination of
16
parabens in l breast milk samples. A three-phase microextraction mode was used for the
17
extraction of the target compounds taking advantage of their acid-base properties. The unit
18
allows the simultaneous extraction of the target compounds from the solid sample to an
19
organic media and the subsequent transference of the analytes to an aqueous acceptor phase.
20
The method includes the identification and quantification of the analytes by ultra high
21
performance liquid chromatography coupled to tandem mass spectrometry (UHPLC-MS/MS).
22
All the variables involved in the extraction procedure have been accurately studied and
23
optimized. The analytes were detected and quantified using a triple quadrupole mass
24
spectrometer (QqQ). The selection of two specific fragmentation transitions for each
25
compound allowed simultaneous quantification and identification. The method has been
26
analytically characterized on the basis of its linearity, sensitivity and precision. Limits of
27
detection ranged from 0.1 to 0.2 ng mL-1 with precision better than 8%, (expressed as relative
28
standard deviation). Relative recoveries were in the range from 91 to 106% which
29
demonstrated the applicability of the stir-membrane solid-liquid-liquid microextraction for the
30
proposed analytical problem. Moreover, the method has been satisfactorily applied for the
31
determination of parabens in lyophilized breast milk samples from 10 randomly selected
32
individuals.
33
Keywords: Stir-membrane solid-liquid-liquid microextraction; Lyophilized human milk
34
samples; Parabens, UHPLC-MS/MS.
Ac ce p
te
d
M
an
us
cr
ip t
14
35
2
Page 2 of 33
1. Introduction.
36
The alkyl esters of p-hydroxybenzoic acid (parabens, PBs) are a group of compounds widely
37
used as bactericide and antimicrobial preservatives, especially against mold and yeast in
38
cosmetic products, pharmaceuticals, and in food and beverage processing [1]. The biological
39
activity of PBs is based on their inhibitory effects on membrane transport and mitochondrial
40
function processes. These compounds are present, individually or in combination, in a large
41
amount of commercial formulations. Although PBs have been considered for years to be
42
relatively safe compounds with a low bioaccumulation potential [2], some studies suggest that
43
they present a moderate endocrine disrupting activity and therefore they can cause adverse
44
effects on humans and wildlife. In fact, the ability of PBs to disrupt physiologically important
45
functions in both in vitro systems [3] and in vivo models [4-6] has been demonstrated. As
46
well, the presence of non-metabolized PBs in breast cancer tissues [7] has focused the
47
attention in their potential carcinogenic and toxic nature [2, 6, 8].
48
The human exposure to PBs may occur through ingestion, inhalation or dermal absorption.
49
This exposure, estimated in 76 mg per day, involves different sources such as cosmetics and
50
personal care products (50 mg/day), drugs (25 mg/day) or food (1 mg/day) [1]. After intake,
51
PBs are metabolized by hydrolysis of the ester bond and by glucuronidation [9]. However, the
52
parent compounds (free forms) can still be detected in biological samples such as urine [10],
53
serum and seminal plasma [11] and human milk [12].
54
Since breast milk is the main route of exposure for breastfed infants, its analysis is of special
55
scientific interest. The isolation of the target analytes from this complex biological sample is a
56
critical aspect even when a chromatographic technique is employed due to selectivity and
57
sensitivity issues. Selectivity aspects, including ion suppression, are also critical when mass
58
spectrometry is employed as detection technique. Current sample preparation of breast milk
Ac ce p
te
d
M
an
us
cr
ip t
35
3
Page 3 of 33
typically involves a previous acid treatment and a centrifugation step [13] to release the target
60
analytes from the sample matrix. Classic liquid-liquid extraction [14] and solid phase
61
extraction [12] with intense cleanup procedures have been proposed for the extraction of these
62
compounds from such samples. These classic techniques, which are usually tedious and
63
require high volume of sample and organic solvents, are gradually being replaced by
64
microextraction techniques, which are characterized by their simplicity, green nature and
65
efficiency. Microextraction techniques, especially solid phase microextraction, have been
66
proposed for the isolation of pesticides [15], polychlorinated biphenyl congeners [16] and
67
monocyclic aromatic amines [17] from human milk samples.
68
Stir membrane extraction (SME) [18, 19], which use a polymeric membrane as extracting
69
phase, has been proposed as a novel technique that integrates the extraction and stirring
70
element in the same device. The main shortcoming of SME is the limited adsorption capacity
71
of conventional membranes which can be enhanced changing the nature of the extracting
72
phase. The use of liquid extracting phases has brought the development of the stir-membrane
73
liquid-liquid microextraction (SM-SLLME), both in the two-phase [20] and three-phase
74
modes [21, 22]. However, the design of the unit does not allow the processing of sample
75
volumes lower than 10 mL which is an obvious restriction in bioanalysis. The adaptation of
76
stir-membrane liquid phase microextraction (SM-LPME) for the extraction of volume limited
77
biological samples, such as saliva, has also been recently published [23].
78
In the present work, SM-SLLME is tailored to the analysis of solid samples and it has been
79
evaluated for the determination of PBs in lyophilized human milk samples. The unit allows
80
the extraction of the target compounds from the solid sample to an organic media and the
81
subsequent transference of the analytes to an aqueous acceptor phase which is compatible
82
with ultra high performance liquid chromatography coupled to tandem mass spectrometry
Ac ce p
te
d
M
an
us
cr
ip t
59
4
Page 4 of 33
(UHPLC-MS/MS). The new proposal gives a selectivity and sensitivity enhancement which is
84
critical when biological samples are studied, even if a UHPLC-MS/MS system is employed.
85
In addition, the extraction technique seems to be versatile enough to face up the isolation and
86
preconcentration of hydrophobic ionizable compounds from different solid matrices. The
87
method was validated and satisfactorily applied to determine the free PBs content in samples
88
collected from 10 volunteers.
cr
ip t
83
us
89
2. Experimental Section
91
2.1 Chemicals and materials
92
All reagents were of analytical grade or better. Methylparaben (MPB), ethylparaben (EPB),
93
propylparaben (PPB), butylparaben (BPB) and ethylparaben
94
were supplied by Sigma-Aldrich (Madrid, Spain). Stock standard solutions of each compound
95
were prepared in acetonitrile (Panreac, Barcelona, Spain) at a concentration of 100 mg L-1 and
96
stored at 4 ºC in the dark. Working solutions were prepared by a rigorous dilution of the
97
stocks in human milk samples prior to their lyophilization. Potassium hydroxide, hydrochloric
98
acid, acetonitrile, hexane and dichloromethane, employed in the sample treatment, were
99
purchased from Panreac (Barcelona, Spain). LC-MS grade acetonitrile and water, used as
100
components of the chromatographic mobile phase were also purchased from Sigma-Aldrich.
101
Polytetrafluoroethylene (PTFE) tape (75 µm, 0.5 µm of pore size) from Miarco (Valencia,
102
Spain) and Eppendorf tube (2 mL in volume) were employed in the construction of the
103
extraction device. Sample agitation for the extraction procedure was carried out in an eight-
104
position digital agitator-vibrator purchased from J.P. Selecta (Barcelona, Spain).
13
C6-ring labelled (EPB-13C6)
Ac ce p
te
d
M
an
90
105 106
2.2 Chromatographic analysis
5
Page 5 of 33
Two chromatographic systems, including different detectors, were employed in the
108
development of the present research. The optimization of the extraction procedure was carried
109
out on a Waters-Acquity™ Ultra Performance LC system (Waters, Manchester, UK) using an
110
Acquity UPLC® BEH C18 column (1.7 μm particle size, 2.1 mm × 100 mm) maintained at 45
111
ºC. The mobile phase consisted of 60 % of Milli-Q ultrapure (Millipore Corp, Madrid, Spain)
112
water and 40 % acetonitrile under isocratic conditions. The flow rate was maintained at 0.5
113
mL min-1. The injection volume was 1 μL with partial loop with needle overfill mode. The
114
separated analytes were detected using a PDA eλ (extended wavelength) Detector (Waters) at
115
254 nm. System control was achieved with Empower software.
116
The extraction performance is calculated in relative terms (extraction recovery and
117
enrichment factors) and the obtained values are independent of the instrument employed. For
118
this reason, UPLC-MS/MS was finally used as instrumental technique in order to improve the
119
sensitivity and selectivity of the UV detection. In this sense, method validation and sample
120
analysis were performed on a Waters Acquity UPLC™ H–Class (Waters, Manchester, UK),
121
consisting of an ACQUITY UPLC™ binary solvent manager and an ACQUITY UPLC™
122
sample manager. A Xevo TQS tandem quadrupole mass spectrometer (Waters) equipped with
123
an orthogonal Z–spray™ electrospray ionization (ESI) source was used for PBs detection. An
124
Acquity UPLC® BEH C18 column (1.7 μm particle size, 2.1 mm × 100 mm) was used. The
125
compounds were separated using a gradient mobile phase consisting of LC–MS grade water
126
(solvent A) and LC–MS grade acetonitrile (solvent B). Gradient conditions were: 0.0-2.0 min,
127
40% B; 2.0-5.0 min, 40-90% B; 5.0-5.1 min, 90-100% B; 5.1-8.0 min, 100% B and back to
128
40% in 0.1 min. Flow rate was 0.5 mL min-1. Total run time was 10.0 min. The injection
129
volume was 10 μL and the column temperature was maintained at 40 ºC.
Ac ce p
te
d
M
an
us
cr
ip t
107
6
Page 6 of 33
For mass spectrometric analysis, the tandem mass spectrometer was operated in the selected
131
reaction monitoring (SRM) mode and Q1 and Q3 quadrupoles were set at unit mass
132
resolution. ESI was performed in the negative ion mode. The mass spectrometric conditions
133
were optimized for each compound by continuous infusion of standard solutions (1 mg L-1).
134
The ion source temperature was maintained at 150 ºC. Instrument parameters were as follows:
135
capillary voltage, 0.60 kV; source temperature, 150 ºC; desolvation temperature, 500 ºC; cone
136
gas flow, 150 L h-1; desolvation gas flow, 500 L h-1; collision gas flow, 0.15 mL min-1, and
137
nebulizer gas flow, 7.0 bar. Nitrogen (99.995%) was used as cone and desolvation gas, and
138
argon (99.999%) was used as a collision gas. Dwell time was set at 25 ms. Optimized
139
parameters for each compound are also listed together with the mass transitions in Table 1.
an
us
cr
ip t
130
M
140 141
d
143
.
te
142
Table 1
2.3 Samples collection and storage
145
Human milk samples were obtained from healthy lactating women living in Granada, Spain.
146
It is important to point out that none of them follow a medical treatment. Samples were
147
collected using a breast pump in a 100 mL PTFE flask and immediately cold storage at -20
148
ºC. All volunteers signed their informed consent to participate in the study.
149
Ac ce p
144
150
2.4 Preparation of spiked samples
151
The blank samples were previously analyzed in order to ensure the absence of the analytes or
152
that these were below the limits of detection of the method (LODs). Due to the absence of
153
certified reference materials, for recovery studies, blank samples were spiked at different
7
Page 7 of 33
concentrations by adding 100 µL of spiking standard solutions to 3 mL of human breast milk.
155
The mixtures were vortexed for 2 min and then, left to stand for 24 h at 4 ºC in the dark
156
before analysis. This allows the analytes to interact with the human milk sample. Then,
157
aliquots of 3 mL of spiked samples were placed in an 8 mL glass vial and frozen at -80 ºC for
158
12 h prior being lyophilized.
ip t
154
cr
159
2.5 Extraction unit
161
The design of the extraction device is the consequence of a deep research in stir membrane
162
extraction and related techniques during the last years. The unit is based on our previous
163
adaptation of stir-membrane liquid phase microextraction to the analysis of volume limited
164
biological samples.23 In this case, the unit has been simplified and only two commercial
165
elements are needed for its construction: an Eppendorf tube of 2 mL and a PTFE tape (75 µm
166
in thickness, 0.5 µm of pore size).
an
M
d
te
168
Figure 1
Ac ce p
167
us
160
169
The assembly process generates two different chambers (lower and upper) separated by a
170
polymeric membrane which avoids direct contact and mixing. The upper section is filled with
171
an appropriate organic solvent where the solid sample is dispersed while the lower chamber
172
(ca. 100 µL) is filled with the acceptor alkaline phase. The extraction device can be simply
173
agitated in an eight-position digital agitator-vibrator allowing the solid-liquid-liquid
174
extraction of the analytes. In addition, the sample throughput is enhanced since various units
175
can be built in a reproducible way, thus allowing that eight samples can be extracted at the
176
same time. Moreover, each unit can be re-used several times just replacing the membrane and
177
cleaning the Eppendorf tube with Milli-Q water and methanol.
8
Page 8 of 33
2.6 Extraction procedure
179
Firstly, the cap of the Eppendorf unit is filled with 100 μL of a 10-2 M potassium hydroxide
180
solution, which acts as acceptor solution. Then, the polymeric membrane is conveniently
181
placed over the cap. Later on, 0.1 g of freeze-dried human milk containing the analytes
182
(corresponding to 1 mL of milk) and 1.5 mL of a mixture of hexane:dichloromethane, 80:20
183
(v/v) were added inside the Eppendorf tube. Once ready, the extraction units are placed on the
184
eight-position digital agitator-vibrator and stirred for 45 min at room temperature (25 ºC).
185
After this time, the sample and the solvent are withdrawn from the unit and the acceptor phase
186
is collected using a 100 µL microsyringe and transferred to a chromatographic vial. Finally,
187
40 µL of a mixture of hydrochloric acid:acetonitrile 1:1 (v/v) are added and 10 µL of the final
188
solution are injected into the chromatographic system.
M
an
us
cr
ip t
178
189
3. Results and discussion
191
Different variables, which are summarized in Table 2, may affect the efficiency of the
192
extraction procedure and therefore their effects on the analyte extraction were considered in
193
the optimization of the extraction procedure. Table 2 also reflects their initial values, the
194
studied interval and the final optimum values. The optimization was performed under a
195
univariate approach although two couples of variables were considered jointly since they are
196
directly related. Lyophilized human breast milk containing the four PBs studied, at a
197
concentration of 100 µg L-1, was used in these studies.
198
Ac ce p
te
d
190
Table 2
199 200
3.1 Selection of the extraction solvent and the final acceptor phase.
9
Page 9 of 33
The proposed extraction procedure allows the solid-liquid microextraction of PBs from
202
lyophilized samples and the subsequent liquid-liquid microextraction from the organic solvent
203
to an aqueous phase compatible with UHPLC–MS/MS. Therefore, the appropriate selection
204
of the organic solvent is a critical step as it plays a double role. On the one hand, the organic
205
solvent should be able to extract the analytes from the freeze-dried breast milk and the
206
analytes should have a higher solubility in the final aqueous solution.
207
As target compounds present a marked interaction with the fatty compounds of the sample,
208
we think in the possibility of using hexane as solvent. Nevertheless, the extraction with pure
209
hexane did not provide good results. In fact, the signals obtained for MPB and EPB were too
210
low while no peaks were obtained for the other compounds. In order to enhance the extraction
211
of the compounds, dichloromethane was added as modifier. The hexane:dichloromethane
212
volume ratio was evaluated in the range from 100:0 to 0:100 (v/v). The results showed that
213
the efficiency of the extraction for all analytes increase up to the 80:20 ratio and an intensive
214
decrease being observed for higher values. The use of the 80:20 ratio produces a 3-fold
215
increase of the signals of MPB and EPB compared to that obtained with pure hexane and it
216
also allows the extraction of PPB and BPB. In addition, this ratio gives an average signal
217
improvement of 20% compared with that obtained with pure dichloromethane.
218
This behavior can be ascribed to two different aspects. On the one hand, the analytes present
219
a high solubility in dichloromethane but in real samples the analytes are interacting with the
220
fatty compounds of the sample matrix which are soluble in hexane. A mixture of both
221
solvents provides a balanced media to break the analyte-matrix interaction and to solubilize
222
the analytes. Moreover, the optimum ratio of solvents also influences the final liquid-liquid
223
partition since the solubility of the target in the aqueous phase should be higher than that in
224
the organic donor phase.
Ac ce p
te
d
M
an
us
cr
ip t
201
10
Page 10 of 33
Considering the acid–base properties of PBs, with pKa in the range of 8.1-8.5, their water
226
solubility can be modulated by an effective pH adjustment. This fact has been exploited to
227
recover the target compounds from the organic solvents mixture. The pH of the acceptor
228
phase was evaluated in the range from 9 to 12 by using different potassium hydroxide
229
concentrations. In the pH range studied the results showed an analytical signal increase with
230
the pH. According to the results, pH 12 was selected as the optimum value since it provides a
231
2,9-fold (average value for all the analytes) signals improvement compared to that obtained at
232
pH 9. This fact can be ascribed to the complete ionization of the target compounds at this pH
233
which increases their solubility in the aqueous medium.
234
3.2. Optimization of sample amount and organic solvent volume
235
Sample amount and solvent volume are crucial variables in this extraction procedure.
236
Moreover, both variables are related to each other and when higher sample amounts are
237
processed, higher volumes of solvent are required to effectively leach the analytes. Sample
238
amount and organic solvent volumes were optimized together considering four different
239
sample amounts (0.05, 0.10, 0.20 and 0.30 g) and four different solvent volumes (0.5, 1.0, 1.5,
240
and 2.0 mL). The maximum value of both variables was selected according to practical
241
considerations. Amounts higher than 0.30 g produce the clogging of the membrane while
242
extractant volumes larger than 2.0 mL cannot be used due to the limited device dimensions.
243
Figure 2 shows the bivariant effect on the extraction factors (EFs) for all analytes by means
244
of a contour surface graph. In the light of the results, when an amount of 0.05 g was
245
employed, the EFs were very low regardless the organic solvent volume tested. When 0.10 g
246
of sample was extracted, the EFs increased with the organic solvent volume up to 1.5 mL,
247
decreasing for higher volumes. This behaviour was observed for all analytes except for MPB,
Ac ce p
te
d
M
an
us
cr
ip t
225
11
Page 11 of 33
248
the most polar analyte, which was preferably extracted with sample amount of 0.20 g and 1.0
249
mL. Figure 2
251
It might be expected that when 0.3 g of sample was extracted, the EFs increase. However, the
252
experimental results showed a notable decrease on the extraction efficiency. This could be
253
ascribed to the low organic solvent volume which is insufficient to leach the analytes from the
254
sample. According to the obtained results, a sample amount of 0.10 g and a volume of 1.5 mL
255
of organic solvent were selected as the optimal values to carry out the extraction procedure.
us
cr
ip t
250
an
256
3.3 Optimization of the extraction time
258
Extraction time was investigated in the range from 5 to 60 min and the results, which are
259
depicted in Figure 3, showed two different behaviors depending on the polarity of the
260
analyte. The extraction of the most polar analytes (MPB and EPB) increases markedly and
261
almost linearly with the time up to 45 min, while from 45 to 60 min the increase was less
262
pronounced. On the other hand, the extraction of the most hydrophobic analytes slightly
263
increases up to 15 min and the extraction equilibrium is achieved, remaining almost constant
264
for higher times. Thus, 45 min was selected as the optimum value as a compromise between
265
the sample throughput and sensitivity.
267
d
te
Ac ce p
266
M
257
Figure 3
268 269
3.4 Analytical characterization of the method
270
3.4.1 Calibration curves
12
Page 12 of 33
A seven-concentration calibration curve for UHPLC–MS/MS was built for each compound.
272
Each level of concentration was made in triplicate. Calibration graphs were constructed using
273
analyte/surrogate peak area ratio versus concentration of analyte. Calibration graphs were
274
made using SRM mode. EPB-13C6 (100 ng mL-1, final concentration in milk samples) was
275
used as surrogate. Table 3 shows the analytical parameters obtained.
277
cr
276
ip t
271
us
Table 3
278
In order to estimate the presence/absence of matrix effect, two calibration curves were
280
obtained for each compound, one in the initial mobile phase and the other in blank human
281
milk. The Student’s t-test was applied in order to compare the calibration curves. First, the
282
variances estimated as S2y/x were compared by means of a Snedecor’s F-test. As is shown in
283
Table 3, the Student’s t-test showed statistical differences among slope values for the
284
calibration curves of PPB and BPB and consequently, a significant matrix effect was observed
285
being of a great magnitude in the case of BPB, and the use of matrix-matched calibration was
286
necessary. A possible explanation for this subject could be that the chemical structure and,
287
consequently, the physical and chemical properties of the analyzed compounds are slightly
288
different within the same family of compounds. Therefore, although the compound selected as
289
surrogate has a similar basic structure compared to the analytes, and the use of this compound
290
as internal standard or as surrogate is accepted in scientific literature, it differ slightly due to
291
the presence of different substituents in the molecule. Since it is impossible to have the
292
corresponding isotopically labeled standard for each one of the studied analytes, it was
293
decided to work with matrix-matched calibration in all cases.
Ac ce p
te
d
M
an
279
294
13
Page 13 of 33
3.4.2 Method validation
296
Validation in terms of linearity, sensitivity, accuracy (trueness and precision), and selectivity,
297
was performed according to the US Food and Drugs Administration (FDA) guideline for
298
bioanalytical assay validation [24].
299
Linearity. A concentration range for the minimal quantified amount (LOQ) to 100 ng mL-1
300
PBs was established. Linearity of the calibration graphs was tested using the determination
301
coefficients (% R2) and the P-values (% Plof) of the lack-of-fit test [25]. The values obtained
302
for R2 ranged from 99.7 for EPB, PPB and BPB and 99.8% for MPB, and Plof values were
303
higher than 5% in all cases. These facts indicate a good linearity within the stated ranges.
304
Limits of detection (LOD) and quantification (LOQ). Those are two fundamental parameters
305
that need to be examined in the validation of any analytical method to determine if an analyte
306
is present in a sample. In the present work, these parameters were calculated by taking into
307
consideration the standard deviation of residuals, Sy/x, the slope, b, of the calibration graphs
308
and an estimate s0 obtained by extrapolation of the standard deviation of the blank [26]. The
309
LOD was 3·s0 and the LOQ was 10·s0. Limits of quantification ranged from 0.4 to 0.5 ng mL-
310
1
311
lyophilization, are also summarized in Table 3.
312
Accuracy (precision and trueness). Due to the absence of certified materials, in order to
313
evaluate the trueness and the reproducibility of the method, a study with spiked human milk
314
samples, at three concentrations levels for each compound was performed on 6 consecutive
315
days. Spiked samples were analyzed in triplicate each day in repeatability conditions. A total
316
of 18 measurements for each level were carried out. The concentrations studied were 1, 50
317
and 100 ng mL-1. The precision was expressed as relative standard deviation (RSD) and the
Ac ce p
te
d
M
an
us
cr
ip t
295
. These values, which are referred to the original breast milk samples before their
14
Page 14 of 33
318
trueness was evaluated by a recovery assay. The precision and the trueness of the proposed
319
analytical method are shown in Table 4.
320 321
ip t
Table 4
322
Trueness was evaluated by determining the recovery of known amounts of the tested
324
compounds in human breast milk samples. Recoveries were determined analyzing the spiked
325
samples and the concentration of each compound was determined by interpolation in the
326
standard calibration curve within the linear dynamic range and compared with the amount of
327
analytes previously added to the samples. The results, which are also summarized in Table 4,
328
were between 91 and 106%. Therefore, it can be concluded that the extraction process fulfills
329
the 70-130% recovery criterion [27]. The results show the potential of the proposed stir
330
membrane solid-liquid-liquid microextraction for the extraction of the four PBs from
331
lyophilized human milk samples.
te
d
M
an
us
cr
323
Inter-day precision (expressed as RSD) was lower than 8%. Therefore, all compounds were
333
within the acceptable limits for bioanalytical method validation, which are considered ≤ 15%
334
of the actual value, except at the LOQ, which it should not deviate more than 20%. The data
335
obtained demonstrated that the proposed method is highly reproducible. Precision and
336
trueness data indicate the potential of the proposed stir-membrane solid-liquid-liquid
337
microextraction for the extraction of the PBs from human milk samples.
338
3.5 Application of the method
339
The validated method was applied to the determination of the amounts of EDCs in 10 human
340
breast milk samples. The results obtained as mean of six determinations are summarized in
341
Table 5. Figure 4 shows a chromatogram of a contaminated natural sample.
Ac ce p
332
15
Page 15 of 33
342 343
Table 5, Figure 4
344
As it is shown in the table, PBs were detected and quantified in many of the analyzed
346
samples. MPB was quantified in 80% of samples (8/10) at concentrations ranging from 0.9 to
347
11.3 ng mL-1. EPB was detected in 6 samples and quantified in 4 of them, with concentrations
348
ranging from 0.5 to 13.2 ng mL-1. Among the four PBs analyzed, PPB was the major
349
compound found in most milk samples (9/10), with the higher concentrations (from 0.5 to
350
37.0 ng mL-1). This indicates concurrent exposure to this compound. Regarding to BPB, this
351
compound was also frequently detected (8/10) although the concentrations (0.6–11.3 ng mL-1)
352
were lower than those of PPB. It is important to remark that the concentrations of total PBs in
353
two of the samples (mothers 01 and 07) were significantly higher than those in the other
354
samples. This fact suggests higher exposures of those mothers to PBs, which can be attributed
355
to the abusive usage of products such as personal hygiene products and cosmetics containing
356
these compounds, recognized by two of the mothers in the initial survey.
cr
us
an
M
d
te
Ac ce p
357
ip t
345
358
CONCLUSIONS
359
In the present work, a new extraction procedure evolved from stir membrane extraction
360
technique is proposed for the isolation and preconcentration of four PBs from human breast
361
milk before their final chromatographic analysis. The extraction procedure allows the
362
simultaneous solid-liquid extraction of the PBs from the lyophilized samples and the
363
subsequent liquid-liquid extraction from the organic extract to an aqueous phase.
364
The integration of both extractions allows not only the improvement of basic analytical
365
properties but also the simplification and miniaturization of the sample treatment. In addition,
16
Page 16 of 33
the sample throughput is enhanced since several samples can be extracted at the same time.
367
The new proposal, which potential has been proved, is highly versatile and it can be easily
368
adapted to face up different analytical problems.
369
The use of tandem MS, allows the unequivocal identification and quantification of these
370
compounds in samples. The main parameters affecting the SM-SLLME procedure (pH and
371
extraction time), have been completely characterized studying in depth the influence of these
372
variables on the analytical signal. The nature of the organic solvent employed for the phase
373
extractants; and the optimal instrumental conditions for UHPLC–MS/MS analysis were also
374
evaluated and optimized. As a result, a selective, sensitive and accurate method for
375
determination of four PBs in human breast milk samples has been developed and validated.
376
The method was satisfactorily applied for the determination of target compounds in human
377
milk samples from 10 randomly selected women.
378
The comparison of the proposed method with other counterparts [12, 28-34] for the
379
determination of parabens in biological fluids is presented in Table 6. Compared with those
380
method focused on human milk analysis [12, 32] the new proposal provides the better
381
sensitivity, the narrower interval of relative recovery values and a good precision. However,
382
our method uses a higher sample volume and requires a larger extraction time.
383
The comparison with those methods focused on the analysis of urine [28-30, 33,34] shows a
384
similar sensitivity and precision (reference 30 excepted), but some of the counterparts require
385
larger extraction times. Finally, our proposal provides better LODs that those obtained for the
386
analysis of plasma samples [31].
Ac ce p
te
d
M
an
us
cr
ip t
366
Table 6
387 388
Acknowledgments
17
Page 17 of 33
This study was supported by the Regional Government of Andalusia (Project of Excellence
390
No. P09-CTS-4470) and the Spanish Ministry of Science and Innovation (grant CTQ2011-
391
23790). M. Roldán-Pijuán expresses her gratitude for the predoctoral grant (refs AP2009-
392
3499) from the Spanish Ministry of Education and R. Rodríguez-Gómez to the Regional
393
Government of Andalusia for the fellowship granted to her.
Ac ce p
te
d
M
an
us
cr
394
ip t
389
18
Page 18 of 33
REFERENCES
395
[1] I. Jiménez-Díaz, F. Vela-Soria, A. Zafra-Gómez, A. Navalón, O. Ballesteros, N. Navea,
396
M. F. Fernández, N. Olea, J. L. Vílchez, Talanta 84 (2011) 702.
397
[2] R. Golden, J. Gandy, G. Vollmer, Crit. Rev. Toxicol. 35 (2005) 435.
398
[3] J.A.van Meeuwen, O. van Son, A.H. Piersma, P.C. de Jong, M. van den Berg, Toxicol.
399
Appl. Pharmacol. 230 (2008) 372.
400
[4] M. G. Soni, I. G. Carabin, G. A. Burdock, Food Chem. Toxicol.43 (2005) 985.
401
[5] J. R. Byford, L. E. Shaw, M. G. B. Drew, G. S. Pope, M. J. Sauer, P. D. Darbre, J. Steroid
402
Biochem. Mol. Biol. 80 (2002) 49.
403
[6] J. Boberg, C. Taxvig, S. Christiansen, U. Hass, Reprod. Toxicol. 30 (2010) 301.
404
[7] P. D. Darbre, A. Aljarrah, W. R. Miller, N. G. Coldham, M. J. Sauer, G. S. Pope, J. Appl.
405
Toxicol. 24 (2004) 5.
406
[8] P.D. Darbre, P.W. Harvey, J. Appl. Toxicol. 28 (2008) 561.
407
[9] S. Abbas, H. Greige-Gerges, N. Karam, M. H. Piet, P. Netter, J. Magdalou, Drug Metab.
408
Pharmacokinet 25 (2010) 568.
409
[10] X. Ye, A. M. Bishop, J. A. Reidy, L. L. Needham, A. M. Calafat, Environ. Health Perspect.
410
114 (2006) 1843.
411
[11] H. Frederiksen, N. Jorgensen, A. M. Andersson, J. Expo. Sci. Environ. Epidemiol. 21
412
(2011) 262.
413
[12] X. Ye, A. M. Bishop, L. L. Needham, A. M. Calafat Anal. Chim. Acta 622 (2008)150.
414
[13] A. Bjørhovde, T. G. Halvorsen, K. E. Rasmussen, S. Pedersen-Bjergaard, Anal. Chim.
415
Acta 491 (2003) 155.
416
[14] P. Turnpenny, J. Rawal, T. Schardt, S. Lamoratta, M. W. Mueller, K. Brady, Bioanalysis
417
17 (2011) 1911.
Ac ce p
te
d
M
an
us
cr
ip t
394
19
Page 19 of 33
[15] M. J. González-Rodríguez, F. J. Arrebola Liébanas, A. Garrido Frenich, J. L. Martínez
419
Vidal, F. J. Sánchez López, Anal. Bioanal. Chem. 382 (2005) 164.
420
[16] C. H. Kowalski, G. A. da Silva, R. J. Poppi, H. T. Godoy, F. Augusto, Anal. Chim. Acta
421
585 (2007) 66.
422
[17] L. S. DeBruin, P. D. Josephy, J. Pawliszyn, Anal. Chem. 70(1998) 1986.
423
[18] M. C. Alcudia-León, R. Lucena, S. Cárdenas, M. Valcárcel, Anal. Chem.81 (2009) 8957.
424
[19] M. C. Alcudia-León, B. Lendl, R. Lucena, S. Cárdenas, M. Valcárcel, Anal. Bioanal.
425
Chem. 398 (2010) 1427.
426
[20] M. C. Alcudia-León, R. Lucena, S. Cárdenas, M. Valcárcel, J. Chromatogr. A 1218
427
(2011) 869.
428
[21] M. C. Alcudia-León, R. Lucena, S. Cárdenas, M. Valcárcel, J. Chromatogr. A 1218
429
(2011) 2176.
430
[22] S. Riaño, M. C. Alcudia-León, R. Lucena, S. Cárdenas, M. Valcárcel, Anal. Bioanal.
431
Chem. 403 (2011) 2583.
432
[23] M. Roldán-Pijuán, M. C. Alcudia-León, R. Lucena, S. Cárdenas, M. Valcárcel,
433
Bioanalysis 5 (2013) 307.
434
[24] Guidance for Industry, Bioanalytical Method Validation, U.S. Department of Health and
435
Human Services, Food and Drug Administration, Center for Drug Evaluation and Research
436
(CDER), Center for Veterinary Medicine (CVM), 2001.
437
[25] Analytical Methods Committee Analyst 119 (1994) 2363.
438
[26] L. A. Currie, Anal. Chim. Acta 391 (1999) 127.
439
[27] US Environmental Protection Agency, Office of Solid Waste EPA Method 8330, SW-
440
846 1996. Available from NTIS.
Ac ce p
te
d
M
an
us
cr
ip t
418
20
Page 20 of 33
[28] S. Y. Lee, E. Son, J. Y. Kang, H. S. Lee, M. K. Shin, H. S. Nam, S. Y. Kim, Y. M. Jang,
442
G. S. Rhee, Bull. Korean Chem. Soc. 34 (2013) 1131.
443
[29] L. Dewalque, C. Pirard, N. Dubois, C. Charlier, J. Chromatogr B, 949– 950 (2014) 37.
444
[30] H. Frederiksen, N. Jørgensen, A.M, Andersson, J. Expo. Sci. Environ. Epidemiol. 21
445
(2011) 262.
446
[31] T. M. Sandanger, S. Huber, M. K. Moe, T. Braathen, H. Leknes, E. Lund, J. Expo. Sci.
447
Environ. Epidemiol. 21 (2011) 595.
448
[32] L. P. Melo, M. E. C. Queiroz, Anal. Methods 5 (2013) 3538.
449
[33] A. G. Asimakopoulos, L. Wang, N. S. Thomaidis, K. Kannan, J. Chromatogr. A 1324
450
(2014) 141.
451
[34] F. Vela-Soria, O. Ballesteros, A. Zafra-Gómez, L. Ballesteros, A. Navalón, Anal.
452
Bioanal. Chem. DOI 10.1007/s00216-014-7785-9.
cr
us
an
M
d te Ac ce p
453
ip t
441
21
Page 21 of 33
Figure Captions
454
Figure 1. Description of the extraction device.
455
Figure 2. Bivariant effect of sample amount and organic phase volume on the enrichment
456
factors of the analytes.
457
Figure 3. Effect of the extraction time on the absolute recovery of the analytes.
458
Figure 4. Chromatogram of a contaminated human milk sample (Sample 01).
cr
ip t
453
us
459
Ac ce p
te
d
M
an
460
22
Page 22 of 33
*Highlights (for review)
Highlights
Ac
ce pt
ed
M
an
us
cr
ip t
Stir membrane device is adapted to solid-liquid-liquid extraction This combination is used to the isolation of parabens from lyophilized human milk The extracted analytes are finally determined by UHPLC-MS/MS Limits of detection were in the low ng mL-1 range with precision better than 8% Recoveries varied between 91 and 106% indicating the applicability of the proposal
Page 23 of 33
Ac
ce
pt
ed
M
an
us
cr
i
Figure
Page 24 of 33
Ac
ce
pt
ed
M
an
us
cr
i
Figure 2
Page 25 of 33
Ac
ce
pt
ed
M
an
us
cr
i
Figure 3
Page 26 of 33
Ac
ce
pt
ed
M
an
us
cr
i
Figure 4
Page 27 of 33
Table 1
Table 1. Transitions and optimized potentials for UHPLC–MS/MS analysis
EPB
PPB
151.1 135.8
b
CE
-38
-22
-38
-14
165.1 91.9a
-38
-24
165.1 136.6b
-38
-16
179.1 91.8a
-42
-24
-42
-16
179.1 136.1
a
b
b
Transitions
CV
CE
-42
-24
193.1 136.1
-42
-16
171.1 98.0a
-36
-24
171.1 142.7b
-36
-14
193.1 91.4a
BPB
b
13
EPB- C6
SRM transition used for confirmation; CV, Cone voltage (V); CE,
Ac
ce pt
ed
M
an
us
SRM transition used for quantification; Collision energy (eV)
Compound
ip t
MPB
151.1 91.8a
CV
cr
Compound Transitions
Page 28 of 33
Table 2
Table 2. List of the variables involved in the SM-SLLME extraction process Initial value
Optimization range
Optimum value
Solvent (hexane:dichloromethane)
0:100 (v/v)
0:100-100:0 (v/v)
80:20 (v/v)
Acceptor phase pH
9
9-12
12
Sample amount (g)
0.05
0.05-0.30
0.10
Organic solvent volume (mL)
0.5
0.5-2.0
1.5
Extraction time (min)
5
5-60
ip t
Variable
Ac
ce pt
ed
M
an
us
cr
45
Page 29 of 33
Tables
Table 3 Analytical and statistical parameters. Matrix Matched Calibration (human breast milk) b (mL ng-1)
sb (mL ng-1)
% R2
% PLof
LOD (ng mL-1)
LOQ (ng mL-1)
LDR (ng mL-1)
MPB EPB PPB BPB
1.367·10-2 5.281·10-3 3.080·10-3 1.020·10-3
1.292·10-4 5.825·10-5 2.980·10-5 1.068·10-5
99.8 99.7 99.7 99.7
19.0 33.9 91.6 28.0
0.2 0.1 0.1 0.2
0.5 0.5 0.4 0.5
0.5-100 0.5-100 0.4-100 0.5-100
ip t
Compound
cr
Calibration in solvent b (mL ng-1)
sb (mL ng-1)
% R2
tcalc
Slopes
MPB EPB PPB BPB
1.479·10-2 5.035·10-3 3.398·10-3 1.320·10-3
1.709·10-4 4.588·10-5 3.746·10-5 9.005·10-6
99.9 99.6 99.8 99.5
1.944 0.961 2.399 6.385
Equal Equal Different Different
Matrix effect corrected by the surrogate? Yes Yes No No
an
us
Compound
Ac
ce pt
ed
M
b, slope; sb, slope standard deviation; R2, determination coefficient; LOD, limit of detection; LOQ, limit of quantification; LDR, linear dynamic range; ttab (α=0.05 and 38 f.d) = 2.03
Page 30 of 33
Table 4
Table 4. Recovery assay, precision and trueness of the method
PPB
BPB
ce pt
ed
M
an
Mean of 18 determinations; RSs; RSD, relative standard deviation
Recovery (%) 91 103 96 93 106 98 95 98 103 96 104 103
Ac
a
RSD (%) 7.7 3.1 1.1 8.0 5.0 3.6 7.5 3.9 4.1 6.0 1.8 3.1
ip t
EPB
Founda (ng mL-1) 0.91 51.3 96.4 0.93 53.0 98.1 0.95 49.2 103.0 0.96 51.8 102.8
cr
MPB
Spiked (ng mL-1) 1.0 50.0 100.0 1.0 50.0 100.0 1.0 50.0 100.0 1.0 50.0 100.0
us
Compound
Page 31 of 33
Table 5
Table 5 Application to human breast milk samples. a
BPB 11.3 (4.7) 2.2 (4.5) D ND 3.2 (4.7) 0.6 (4.7) 6.5 (5.7) 1.7 (3.1) ND 1.5 (5.3)
ip t
PPB 37.0 (6.4) 5.9 (6.8) D D 5.9 (4.8) 0.5 (6.0) 27.2 (2.2) 2.6 (6.1) ND 0.6 (5.1)
ce pt
ed
M
an
us
Mean of 6 determinations (ng mL-1); ND, not detected (LOD and 70
LC-MS/MS
0.5-1 (LOQe)
67-126
5.8-18.6
[33]
DLLME
Urine
5 mL
TCM
>20
UHPLC-MS/MS
0.1
94-103
2.0-9.9
[30]
SM-SLLME
Human milk
1 mL
HEX:DCM (4:1) / aqueous phase pH 12
45
LC-MS/MS
0.1-0.2
91-106
1.1-7.7
This method
Ac c
SPE
SPE, solid phase extraction; LLE, liquid liquid extraction; DLLME dispersive liquid liquid microextraction; MISPE, molecularly imprinted solid phase extraction; SM-SLLME, stir membrane solid liquid liquid microextraction. HLB, hydrophilic-lipophilic balance; MIP, molecularly imprinted polymer; EA: ethyl acetate; TCM, trichloromethane; HEX, hexane; DCM, dichloromethane. HPLC, high performance liquid chromatography; MS, mass spectrometry; LC, liquid chromatography; UHPLC, ultra high performance liquid chromatography; UPLC, ultra performance liquid chromatography; TOF, time of flight; FTIR, fourier transform-infrared spectroscopy; UV, ultraviolet. LOD, limit of detection; e LOQ, limit of quantification; f MDL, method detection limit.
Page 33 of 33