Analysis of steroid estrogens in water using liquid chromatography/tandem mass spectrometry with...
-
Upload
ying-hsuan-lin -
Category
Documents
-
view
221 -
download
4
Transcript of Analysis of steroid estrogens in water using liquid chromatography/tandem mass spectrometry with...
RAPID COMMUNICATIONS IN MASS SPECTROMETRY
Rapid Commun. Mass Spectrom. 2007; 21: 1973–1983
) DOI: 10.1002/rcm.3050
Published online in Wiley InterScience (www.interscience.wiley.comAnalysis of steroid estrogens in water using liquid
chromatography/tandem mass spectrometry with
chemical derivatizations
Ying-Hsuan Lin1, Chia-Yang Chen1* and Gen-Shuh Wang1,2
1Institute of Environmental Health, College of Public Health, National Taiwan University, Taipei 10055, Taiwan2Department of Public Health, College of Public Health, National Taiwan University, Taipei 10055, Taiwan
Received 28 February 2007; Revised 21 April 2007; Accepted 25 April 2007
*CorrespoHealth, CTaipei 10E-mail: dContract/tract/gra
Even in trace amounts, estrogens such as 17b-estradiol (E2), estrone (E1), estriol (E3), and 17a-ethinyl
estradiol (EE2) may have adverse effects on humans and the aquatic ecosystem. Therefore, it is
essential to be able to measure trace amounts of steroid estrogens in water. To date, most instruments
are not sensitive enough to detect these chemicals in small samples of water. Sensitivity, however,
may be improved by using appropriate derivatization reagents to modify the structures of these
estrogens so that their ionization efficiency is increased, making them more detectable by liquid
chromatography/mass spectrometry (LC/MS). This study uses dansyl chloride, 2-fluoro-1-methyl-
pyridinium p-toluenesulfonate (FMPTS), and pentafluorobenzyl bromide (PFBBr) as derivatization
reagents to react with the phenolic estrogens tomake themmore detectable inwater.We also test how
environmental matrices (wastewater effluent, river water, and drinking water) influence the detect-
ability of these estrogens. Both qualitative and semi-quantitative comparisons of these derivatization
methods were made. We found that dansyl chloride derivatives created signal intensities one or two
orders of magnitude greater than those normally found in underivatized estrogen standards. The
signals derived by FMPTS were analyte-dependent, and the products derived from E1, E2, and EE2
produced 2.19 to 12.1 times the signal intensity of underivatized E1, E2, and EE2. The product derived
from E3 produced weaker signals than that produced by underivatized E3. The PFBBr derivatives
produced signals that were as much as 5.8 times those found in the underivatized estrogens. When
these derivatizationmethodswere applied to river water, drinkingwater and effluents from a sewage
treatment plant (STP), the different matrices were found to significantly suppress the signals if we
used electrospray ionization, though this influence became less significant if we used atmospheric
pressure chemical ionization. This study suggests that PFBBr derivatization can best be used for the
detection of these estrogens in complex environmental matrices such as river water and STP effluents
and that the dansyl chloride derivatization is best used for clean samples such as drinking water.
Copyright # 2007 John Wiley & Sons, Ltd.
Over the last two decades concern has grown over the issue
of environmental pollution by endocrine disruptors. Some
feminizing compounds are thought to have many different
adverse effects, not only in humans where they suspected to
affect the production of sperm, but also in the environment
where, for example, sex ratios in fish can be reversed. Of
these feminizing compounds, steroid estrogens have been
found to have much greater estrogenic potency.1 Natural
steroid estrogens and synthetic steroid estrogens used in
medicine and veterinary medicine most often find their way
into the environment through human and animal urine.
Although both estrogens are mainly excreted in the form of
biologically inactive conjugates, these conjugates are likely to
ndence to: C.-Y. Chen, Institute of Environmentalollege of Public Health, National Taiwan University,055, [email protected] sponsor: National Science Council, Taiwan; con-nt number: NSC94-2314-B002-278.
be deconjugated and reverted to active free steroids by the
enzymatic reaction of microorganisms in sewage treatment
systems and some other natural environments.2
Trace concentrations of these chemicals in water have been
found to trigger adverse biological effects, particularly in an
aquatic environment.3,4 However, due to the limitations
associated with the analytical techniques used to detect these
estrogens, it is difficult to measure them and assess their
impact on the environment, particularly in complex environ-
mental matrices. Many biological assays and methods of
chemical analysis have been used to measure steroid estro-
gens in different types of water samples. However, the esti-
mates of biological assays, such as immunoassays, often are
confoundedby cross-interactions.1,5 Chemical analysis, includ-
ing gas chromatography/mass spectrometry (GC/MS) or
Copyright # 2007 John Wiley & Sons, Ltd.
1974 Y.-H. Lin, C.-Y. Chen and G.-S. Wang
gas chromatography/tandem mass spectrometry (GC/MS/
MS),6–10 has its own set of limitations, particularly the need
for labor-intensive sample preparation and pretreatment.
Although liquid chromatography (LC)-based methods
provide another choice, they may not be sensitive enough
to measure estrogens in complex environmental matrices.11
Complex environmental matrices not only come with
increased background noise, but also may contain molecules
that could suppress ionization of analytes in the atmospheric
pressure ionization (API) source, especially when electro-
spray ionization (ESI) is used.12
Many factors can affect the generation of ions and affect the
intensity of signals in LC/API-MS. Recently, some research-
ers have modified structures into more ionizable derivatives
to improve the detection of poorly or non-ionizable com-
pounds.13 Because steroid estrogens all contain a phenolic
group (Fig. 1), this group can be used to connect
API-responsive moieties. Most derivatization techniques
for ESI couple a readily ionized group or a permanently
charged moiety with analytes; on the other hand, the
derivatization techniques for atmospheric pressure chemical
ionization (APCI) usually couple moieties that either have
high proton affinities or electron affinities with analytes.14
In this study, we experiment with three derivatization
reagents: dansyl chloride, 2-fluoro-1-methylpyridinium
p-toluenesulfonate (FMPTS), and pentafluorobenzyl bro-
mide (PFBBr). Dansyl chloride, because it has a secondary
amine moiety, is the derivatization reagent commonly used
to improve the detection of steroids with phenolic groups in
the ESI mode.15 FMPTS connects a permanently charged
pyridinium moiety to alcohols through nucleophilic substi-
tution.16 Both derivatization methods may render steroid
estrogens readily detectable in ESI-MS. Another reagent,
PFBBr, can react with the phenolic group of the steroid
estrogens and enhance their electron-capturing capability in
the APCI interface.17
When LC/API-MS is used in the quantitative analysis of
complex mixtures (e.g. environmental samples or biological
fluids), matrix effects must be evaluated. MS may respond
differently to analytes in a complex matrix than it does to
those in a standard solution. In LC/ESI-MS, co-eluting
Figure 1. Structures of the steroid estrogens.
Copyright # 2007 John Wiley & Sons, Ltd.
matrix components can compete with the target analytes to
access the droplet surface for forming gas-phase ions in the
ionization interface and thus reduce the ionization efficiency
of target compounds.18 In addition to their ability to suppress
ionization, matrix components may also present added
background noise and deteriorate detection limits. Several
approaches have been used to evaluate matrix effects.19–22
Matuszewski et al. used post-extraction addition to deter-
mine ion suppression caused by co-eluting matrix.22 In that
study, they compared signal intensities found in neat
standard solutions with those of standards spiked into
biofluids after extraction. The degree of matrix effects under
chromatographic conditions can be assessed by using the
ratio of the responses of analytes in the samples spiked after
extraction to those of pure standards.22
The objectives of this study were to use three chemical
derivatization methods (dansyl chloride, FMPTS, and
PFBBr) with LC/API-MS/MS to qualitatively and semi-
quantitatively analyze steroid estrogens in drinking water,
river water and effluents from a sewage treatment plant
(STP). Signal intensity or limits of detection (LODs) were
found to be improved by derivatization methods used. By
lowering the detection limits in this way we may be able to
usemuch smaller samples of water than is presently required
to monitor these estrogens in clean waters and in more
complex matrices. By reducing the amount of water to be
analyzed, we may also be able to reduce matrix effects.
This study expands the applications of the three
derivatizations. Previously, dansyl chloride has been
used to measure steroid estrogens (17a-ethinyl estradiol
(EE2),15,23,24 17b-estradiol (E2) and estrone (E1)25) in plasma;
FMPTS has been applied to react with alcohols to form
N-methylpyridyl ether ions,16 which might make it suitable
for derivatizing phenolic estrogens; PFBBr derivatization of
E1 and E2 has only been used in neat standard solutions.17
EXPERIMENTAL
Chemicals and reagentsEstrone (E1), 17b-estradiol (E2), estriol (E3), and 17a-ethinyl
estradiol (EE2) were obtained from Sigma (St. Louis, MO,
USA; purity >98%). Dansyl chloride (5-(dimethylamino)
naphthalene-1-sulfonyl chloride, �95% in purity), FMPTS
(2-fluoro-1-methylpyridinium p-toluenesulfonate, technical
grade, �90%), pentafluorobenzyl bromide (PFBBr, purity
>99%), 4-methylmorpholine (purity >99.5%), triethylamine,
formaldehyde, sodium hydrogen carbonate, and potassium
hydroxide were purchased from Sigma-Aldrich (St. Louis,
MO, USA), and formic acid from J. T. Baker (Phillipsburg, NJ,
USA). Solvents, including methanol, acetone, heptane,
acetonitrile and dichloromethane, were all HPLC grade
and purchased from J.T. Baker.
Synthesis of estrogen derivatives
Derivatization with dansyl chlorideThe method of derivatization with dansyl chloride was
modified from Anari et al.15 and Nelson et al.25 A volume of
20mL of the estrogen mixture in acetone (2.5 ng/mL) was
vortexed with 50mL of sodium bicarbonate buffer (100mM,
Rapid Commun. Mass Spectrom. 2007; 21: 1973–1983
DOI: 10.1002/rcm
Analysis of steroid estrogens in water by LC/MS/MS 1975
pH adjusted with NaOH (aq) to 10.5) for 1min. Then 50mL of
1mg/mL dansyl chloride solution was added, and the
mixture was vortexed for another 1min. It was then
incubated at 608C for 3min and then allowed to cool down
to room temperature. The solution was evaporated to
dryness using a SpeedVac concentrator (Thermo Savant
SPD 1010, Holbrook, NY, USA), and the residue was
reconstituted with 20mL of methanol. A volume of 4mL of
the reconstituted solution (representing 10 ng of the original
analyte) was injected onto the LC/MS/MS system for
analysis.
Derivatization with FMPTSThe derivatization with FMPTS was adopted from
Mukaiyama et al.26 and Bald.27 The estrogenmixture solution
in acetonitrile (20mL; 2.5 ng/mL) was vortexed with 50mL of
0.1M triethylamine in acetonitrile for 1min. Then, 50mL of
0.1M FMPTS solution in acetonitrile was added and the
mixture was vortexed for another 1min. It was then shaken
in a reciprocal water bath (RB-120; Cherng Huei, Taiwan) at
room temperature for 1 h (110 rpm). The mixed solution was
evaporated to dryness using the SpeedVac concentrator, and
the residue was redissolved with 20mL of 50:50 (v/v)
methanol/water solutions. A volume of 4mL of the
reconstituted solution was injected onto the LC/MS/MS
system for analysis.
Derivatization with PFBBrThe derivatization with PFBBr was performed according to a
method used by Singh et al.17 The estrogen mixture in
acetonitrile (50mL; 1.0 ng/mL) was vortexed with 50mL of
potassium hydroxide in anhydrous ethanol (8:1000, w/v)
for 1min. Then, 50mL of PFBBr in acetonitrile (1:19, v/v)
was added. It was vortexed for another 1min. The solution
was incubated at 608C for 30min, and then evaporated to
dryness using the SpeedVac concentrator. The residue was
reconstituted with 50mL of methanol. A volume of 10mL of
the reconstituted solution (representing 10 ng of the analyte)
was injected onto the LC/MS/MS system for analysis.
ChromatographyA Waters 616 LC system equipped with an ECO-1 column
oven (Analab, Taipei, Taiwan) and a DU2001 on-line
degasser (Sanwa Tsusho, Tokyo, Japan) was used for liquid
chromatography. Five chromatographic conditions were
optimized for different estrogen derivatives and under-
ivatized parent compounds. A total of 10 ng of each analyte,
both derivatized and underivatized, was injected onto the LC
system for both qualitative and quantitative analysis.
Steroid estrogens without derivatization in ESI(�)Underivatized steroid estrogens in negative ESI mode were
separated using a BetaBasic C18 column (150� 2.1mm, 3mm;
Hypersil-Keystone, Bellefonte, PA, USA). Column tempera-
ture was maintained at 408C. We used a binary gradient
consisting of 10mM 4-methylmorpholine (aq) (pH 9.5–9.6)
(A) and 100% acetonitrile (B) at a flow rate of 0.2mL/min.
Gradient was increased from 10% to 50% of solvent B in
2min, from 50% to 100% in another 11.5min, and decreased
Copyright # 2007 John Wiley & Sons, Ltd.
back to its initial composition in 2.5min. The system was
re-equilibrated for 6min before the next injection.
Steroid estrogens without derivatization in APCI(�)Underivatized estrogens in negative APCI mode were
separated using a BetaBasic C18 column (150� 4.6mm,
3mm; Hypersil-Keystone). Column temperature was main-
tained at 408C. Water (A) and methanol (B) were used to
perform a gradient elution at a flow rate of 1.0mL/min.
Gradient was increased from 10% to 50% of solvent B in
3min, from 50% to 70% in another 12min, and decreased
back to its initial composition in 3min. The system was
re-equilibrated for 4min before the next injection.
Dansyl-estrogens in ESI(þ)Dansyl chloride derivatives were separated using a BetaBasic
C18 column (150� 2.1mm, 3mm; Hypersil-Keystone). Col-
umn temperature was maintained at 408C. Ten mM formic
acid (pH 2.9–3.0) (A) and 100% acetonitrile (B) were used as
the mobile phase at a flow rate of 0.2mL/min. Gradient was
increased from 50% to 80% of solvent B in 2min, from 80% to
100% in another 11.5min, and decreased back to its initial
composition in 2.5min. The system was re-equilibrated for
6min before the next injection.
FMP-estrogens in ESI(þ)FMPTS derivatives were separated using a polystyrene-
divinylbenzene (SDB) PRP-1 column (150� 2.1mm, 3mm;
Hamilton, Reno, Nevada, USA). Column temperature was
maintained at 408C. The mobile phase contained 10mM
formic acid (pH 2.9–3.0) (A) and 100% acetonitrile (B) at a
flow rate of 0.2mL/min. Gradient was increased from 5% to
40% of solvent B in 3min, from 40% to 95% in another 12min,
held at 95% for 2min, and then decreased back to its initial
composition in 2min. The system was re-equilibrated for
3min before the next injection.
PFB estrogens in APCI(þ)PFBBr derivatives were separated using a BetaBasic C18
column (150� 4.6mm, 3mm; Hypersil-Keystone). Column
temperature was maintained at 408C. Water (A) and 100%
methanol (B) was used as the mobile phase at a flow rate of
1.0mL/min. Gradient wasmaintained at 60% of solvent B for
2min, increased from 60% to 100% in 10min, held at 100% for
another 1min and then decreased back to its initial
composition in 2min. The system was re-equilibrated for
3 minu before the next injection.
Mass spectrometryWe used a Finnigan TSQ 7000 triple-quadrupole mass
spectrometer (Finnigan MAT, San Jose, CA, USA) controlled
by the XcaliburTM Home Page version 1.1, which we
operated in the selected reaction monitoring (SRM) mode.
Collision-induced dissociation (CID) was performed using
argon as the collision gas at a pressure between 2.9 to
3.1mTorr. The operation parameters are summarized in
detail in Table 1.
Rapid Commun. Mass Spectrom. 2007; 21: 1973–1983
DOI: 10.1002/rcm
Table 1. Instrumental parameters of the mass spectrometer
Ionsource
ESI sprayvoltage
Corona dischargecurrent
Vaporizertemperature
Heatedcapillary
Sheathgas
Auxiliarygasa
Underivatized estrogens ESI 4.5 kV — — 2508C 70 psi 20Dansyl-estrogens ESI — —FMP-estrogens ESI — —Underivatized estrogens APCI — 5mA 5008C 2008C 80 psi offPFB-estrogens APCI — 10mA 5008C 2008C 40 psi 10
aAuxiliary gas (nitrogen) is presented in arbitrary units.
Figure 2. Sample preparation procedures for water samples.
1976 Y.-H. Lin, C.-Y. Chen and G.-S. Wang
Water sampling and analysisThree types of water samples (river water, STP effluents and
drinking water) were analyzed to evaluate the feasibility of
these derivatization methods. The river water was sampled
from the Danshui River, one of three major rivers in Taiwan.
That river flows through the Taipei metropolitan area,
where it was sampled at Taipei Bridge (E 121830.433 and
N 25803.829). Effluents were taken from Neihu STP, where
they had gone through a secondary treatment process using
the activated sludge units. The drinking water we used was
Taipei tap water. All samples were collected in March 2006.
Samples were first collected in a 20-L pre-cleaned tank at the
sampling site, mixed well, and then subsampled into amber
glass jars with Teflon-liner caps in the laboratory. Formal-
dehyde (1%, v/v) was added to prevent biological degra-
dation and samples were stored at 48C until analysis.
The sample extraction procedures were adopted from the
work of Chen et al.28 and are illustrated in Fig. 2. Before
extraction, 2-L samples were filtered using 90-mm PVDF
(polyvinylidene fluoride) membranes with a pore size of
0.45mm to remove suspended solids. Soluble matrices were
extracted with 50-mm Bakerbond PolarPlus C18 Speedisks
(J.T. Baker). After extraction, disks were washed with 40%
MeOH/60% H2O (v/v). The disks were moved to a vacuum
manifold (J.T. Baker) and dried for 10min under a vacuum of
around �25 kPa before being eluted with three portions of
5mL 50% methanol/50% dichloromethane. The collected
eluates were filtered through 25mm PTFE syringe filters
(pore size 0.2mm) to remove particulates and then divided
into five aliquots. Eluates were concentrated to dryness at
458C in a 1.8-mL vial by the SpeedVac concentrator and were
reconstituted by appropriate solvents, with or without
spiking of 50 ng steroid estrogens, and reacted with the
derivatization reagents.
Evaluation of method performanceAbsolute peak heights (as signal intensity) and signal-
to-noise (S/N) ratios (as relative sensitivities) were used to
evaluate the performance of the three derivatization methods.
Steroid estrogens in three types of environmental mat-
rices were detected using appropriate LC and ionization
methods. Deionized/distilled water from a Milli-Q unit
(Millipore, Bedford, MA, USA) was used as the matrix-free
blank for comparison. Eluates from extraction disks were
equally divided into five aliquots to ensure each aliquot
contained the same amount of endogenous estrogens and
matrix residues. We evaluated the matrix effect on each
method by analyzing estrogens in the drinking water, river
water and STP effluents. A matrix effect factor (MEF) was
defined as the quotient of averaged S/N value between
Copyright # 2007 John Wiley & Sons, Ltd.
environmental matrices and the Milli-Q (MQ) water
(i.e. MEF¼B/A, A: averaged S/N value of analytes in
MQ water, B: averaged S/N value of analytes in drinking
water, river water or STP effluents) expressed by a
percentage, while MQ water was used as a matrix-free
reference. Low MEF values indicated serious matrix effects.
Rapid Commun. Mass Spectrom. 2007; 21: 1973–1983
DOI: 10.1002/rcm
Analysis of steroid estrogens in water by LC/MS/MS 1977
Each analysis was conducted with four duplicates and
consisted of two sets. In the first set, 50 ng of each steroid
estrogen was spiked into the concentrated extracts of water
samples to simulate derivatizations of real samples. The
other set, which was not spiked, was taken as environmental
backgrounds containing endogenous analytes and matrices.
The environmental backgrounds were deducted from the
responses of the first set before the comparisons. Reagent
blanks were included in each batch for analysis. No
experimental contamination was observed.
Data acquisition and analysisMass spectrawere obtainedusingXcaliburHomePage version
1.1 software (Finnigan MAT). Further data processing was
completed with Microsoft Excel 2003. Matrix effects among
various waters, signal intensity and sensitivity enhancement
between different analytical methods were compared statisti-
cally with one-way analysis of variance (ANOVA) followed by
Tukey’s test (p< 0.05) using SPSS for Windows 11.5.0.
Figure 3. Scheme of selected derivatization reactions: (a) d
(c) PFBBr derivatization.
Copyright # 2007 John Wiley & Sons, Ltd.
RESULTS AND DISCUSSION
Qualitative analysis of derivatized productsDansyl chloride, FMPTS, and PFBBr react with the phenolic
group of steroid estrogens, resulting in the formation of
ethers (Fig. 3). The amount of reagents in excess of those
theoretically needed to react with analyte standard compen-
sated for matrix components and ensured the completion of
reactions. The derivatives of estrogen standards were
injected into the mass spectrometer by direct infusion to
optimize the sensitivity of the instrument. Following The
Commission of the European Communities guidelines
suggesting four identification points to improve data
reliability, we monitored one precursor and two product
ions for each analyte.29 The most intense product ion was
used for quantification, and the second most abundant
product ion was used to confirm the analyte identity. The
liquid chromatograms of the five analytical methods are
shown in Fig. 4.
ansyl chloride derivatization, (b) FMPTS derivatization
Rapid Commun. Mass Spectrom. 2007; 21: 1973–1983
DOI: 10.1002/rcm
Figure
4.Liquid
chromatogramsoffive
analyticalsystems:(a)native
estrogensunderESI(�);(b)dansyl-estrogensunderESI(þ);(c)FMP-estrogensunderESI(þ);
(d)native
estrogensunderAPCI(�);and(e)PFB-estrogensunderAPCI(�);10ngofeachanalyte
wasinjected.
Copyright # 2007 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2007; 21: 1973–1
DOI: 10.1002/
1978 Y.-H. Lin, C.-Y. Chen and G.-S. Wang
983
rcm
Analysis of steroid estrogens in water by LC/MS/MS 1979
Analysis of dansyl-estrogen derivatives
Derivatization processAlthough the derivatization efficiencies were not determined
in this study because authentic derivative standardswere not
available, previous reports30,31 have shown them to be close
to 90%, and the variations in yield could be corrected if stable
isotope labeled internal standards were used. The overall
reaction was carried out in acetone; dried residues were
reconstituted with methanol. Practically, the excess reagents
could not be removed after derivatization, so the recon-
stituted solution was used to extract the derivatives from the
reaction mixture. Although derivatized products reconsti-
tuted with the initial mobile phase, 50% ACN/
50% H2O, produced higher analyte signals, the precipitation
of the redissolved bicarbonate salts during the gradient
elution blocked LC tubing. However, when CH3OH was
used for reconstitution, most of the excess reagent, especially
the inorganic slats, was left over.
Product ion characterizationThe four dansyl-estrogens had the same product ions and
similar optimized collision energies. [Mþ233.8]þ was found
to be the most suitable precursor ion, and yielded an intense
product ion m/z 171þ. Another product ion, m/z 156þ,
supposedly resulting from loss of one methyl group from the
product ion ofm/z 171þ, was selected as the confirmatory ion.
The best collision energy was determined to be 42 eV for
dansyl-E1, 40 eV for dansyl-E2, 44 eV for dansyl-E3 and 43 eV
for dansyl-EE2 (Table 2).
LC/MS/MS conditionsAcidic additives of 10mM formic acid (pH 2.9–3.0) were used
as the aqueous mobile phase to enhance the ionization of
dansyl derivatives.15 The signal intensity of dansyl deriva-
tives in MQ water was significantly higher (38–101 times)
than that produced by underivatized estrogens in ESI(–)
(Table 3).
Table 2. SRM transition of native and derivatized estrogens
Derivatization reagent Target compound Precursor ion (m/z)
— E1 269.4�
E2 271.4�
E3 287.4�
EE2 295.4�
Dansyl chloride Dansyl-E1 504.0þ
Dansyl-E2 506.0þ
Dansyl-E3 522.0þ
Dansyl-EE2 530.0þ
FMPTS FMP-E1 362.0þ
FMP-E2 364.0þ
FMP-E3 380.0þ
FMP-EE2 388.0þ
PFBBr PFB-E1 269.4�
PFB-E2 271.4�
PFB-E3 287.4�
PFB-EE2 295.4�
Copyright # 2007 John Wiley & Sons, Ltd.
Analysis of FMP-estrogen derivatives
Derivatization processWe tried to carry out FMPTS derivatization in dichloro-
methane according to the reported literature26 but the
reaction failed. We observed that FMPTS remained sus-
pended in dichloromethane. Replacing dichloromethane
with acetonitrile completely dissolved both the analytes
and the reagent. For the derivatization, it is critical to ensure
that the reagent, which is hygroscopic,32 stays dry and that a
fresh solution is always used. Besides, the best commercial
grade of this reagent is only 90% pure, and there may be a
source of impurities from itself in derivatization. An intense
signal at m/z 375þ from unknown origins was observed even
in the reagent blank by full-scan mass spectra of derivatives
of FMPTS. Fortunately, this ion did not interfere with our
experiments because it could be filtered out in selected
reaction monitoring (SRM) mode.
Product ion characterizationFMPTS reacted with steroid estrogens and formed ether ions
ofm/z [Mþ92]þ, an ion pattern also reported byQuirke et al.16
The product ion of m/z 128þ was used to quantify FMP-E2,
FMP-E3, and FMP-EE2. Another product ion ofm/z 110þwas
used for confirmation. The quantitative and confirmatory
ions of FMP-E1 were m/z 252þ and m/z 238þ, respectively
(Table 2). When the method was applied to real water
samples, product ions of m/z 110 resulting from the
derivatization reagent seemed susceptible to matrix effects;
higher noise levels and co-eluting impurities were observed
when m/z 110 was monitored. On the other hand, though
another product ion of m/z 128 had a lower intensity,
the lower noise levels led to a higher S/N ratio than that of
m/z 110. FMP-E2, FMP-E3 and FMP-EE2 all showed similar
results.
LC/MS/MS conditionsWe used a SDB polymer column and 10mM formic acid as
the aqueous mobile phase. Although adding 10mM formic
Product ion (m/z)
Collision energy (eV)Quantification Confirmation
143� 145� 65183� 145� 46145� 171� 50145� 159� 40171þ 156þ 42171þ 156þ 40171þ 156þ 44171þ 156þ 43252þ 238þ 44128þ 110þ 52128þ 110þ 50128þ 110þ 50143� 145� 65183� 145� 46145� 171� 50145� 159� 40
Rapid Commun. Mass Spectrom. 2007; 21: 1973–1983
DOI: 10.1002/rcm
Table 3. Signal intensities of different estrogen derivatives (mean�SD, n¼ 4)
Native estrogensin ESI(�) Dansyl-estrogens FMP-estrogens
Native estrogensin APCI(�) PFB-estrogens
MQ waterE1 1061� 181a,b 41150� 11458b,c,d,e 12850� 1201a,c 2990� 281a 6165� 802a
E2 1313� 330a 41650� 8458b,c,d,e 2873� 376a 1475� 121a 4833� 884a
E3 1030� 134a 66050� 19725b,c,d,e 244� 55a 1255� 66a 3515� 402a
EE2 467� 124a 47150� 12445b,c,d,e 1800� 110a 479� 8.1a 2623� 361a
Drinking waterE1 946� 125a 36250� 14769b,c,d,e 12250� 1518a 2568� 204a 8793� 498a
E2 988� 392a 42225� 21855b,c,d,e 2730� 340a 1348� 111a 7065� 559a
E3 671� 224a 52750� 17577b,c,d,e 248� 47a 1280� 167a 4923� 484a
EE2 412� 56a 45700� 20703b,c,d,e 1710� 236a 505� 63a 3933� 297a
River waterE1 162� 60a 18167� 4881b,c,d,e 1755� 441a 2590� 1139a 3080� 370a
E2 11.5� 1.2a 18475� 6236b,c,d,e 355� 114a 1397� 671a 2087� 544a
E3 25.3� 1.5a 29775� 4145b,c,d,e 398� 194a 2343� 373a 2863� 322a
EE2 65.5� 8.3a 22650� 8118b,c,d,e 196� 55a 609� 194a 1303� 115a
STP effluentsE1 518� 46a,b 30425� 2069b,c,d,e 11300� 1100a,c,d 3943� 937a,b 4820� 260a
E2 299� 75a,e 22750� 1690b,c,d,e 1487� 55a 1386� 361a 2288� 302a,c
E3 34.6� 9.3a 27300� 2302b,c,d,e 260� 94a 1563� 192a 2125� 127a
EE2 80.3� 12.9a 27650� 2042b,c,d,e 766� 339a 575� 149a 1428� 276a
Note. Signal intensity is defined as absolute peak height from the detector; 10 ng equivalent of each parent chemical was injected theoretically.a Statistically different from dansyl-estrogens.b, Statistically different from FMP-estrogens.c, Statistically different from native estrogens in ESI(�).d, Statistically different from native estrogens in APCI(�).e Statistically different from PFB-estrogens.
1980 Y.-H. Lin, C.-Y. Chen and G.-S. Wang
acid slightly suppressed analyte signal compared with the
use of pure deionized water, it produced better peak shapes
and reproducible retention times. We found that silica-based
columns were not able to retain cationic FMP-derivatives
well, which is consistent with previous reports.32–34 The
solution used for reconstitution also influenced signal
intensity. Residues of pyridinium salts redissolved more
readily in a 50:50 (v/v) CH3OH/H2O solution and produced
about twice the signal intensity of those redissolved with
pure methanol.
Analysis of PFB-estrogen derivatives
Derivatization processThe reaction of PFBBr derivatization is well described by
Singh et al.,17 and we used it without modification.
Product ion characterizationSingh et al.17 reported that PFB-estrogen derivatives of E1
and E2 could undergo electron-capture dissociation and
generate negative ions through the loss of a PFB radical. We
observed similar results in not only PFB-E1 and PFB-E2 but
also in PFB-E3 and PFB-EE2, and the produced negative
precursor ions were identical to those of underivatized ones.
Therefore, the product ions and collision energies for the
derivatized estrogens were identical to the underivatized
ones (Table 2). The position of the corona discharge needle
and its current were critical in the electron-capture
process;17,35 the intensity increased as the current increased.
However, the maximum discharge current of the mass
spectrometer (TSQ 7000) was 10mA, and the corona probe
position fixed. These two parameters could be further
optimized for other types of mass spectrometers if they
Copyright # 2007 John Wiley & Sons, Ltd.
provid higher corona discharge currents and needle
positions can be adjusted.
LC/MS/MS conditionsChromatographic separation of PFB derivatives carried out
with a column of 4.6mm i.d. at a flow rate of 1.0mL/min
provided 10-fold higher signal intensity than that with a
2.1mm i.d. column at a flow rate of 0.2mL/min. In addition,
derivatized PFB-estrogens became more non-polar than the
parent chemicals, and required higher organic portions in
the mobile phase (60–100% methanol) to elute them out,
while native estrogens were eluted at lower organic fractions
(50–70% methanol).
Evaluation of method performance
Comparison of absolute signal intensities observedby different analytical methodsSignal intensity was defined as the absolute peak height of
analytes on the MS detector. We compared these intensities
under five analytical conditions: (1) dansyl-estrogens in
ESI(þ), (2) FMP-estrogens in ESI(þ), (3) PFB-estrogens in
APCI(�), (4) underivatized estrogens detected under ESI(�),
and (5) underivatized estrogens detected under APCI(�). In
previous studies, ESI(�) has been reported to be the most
common method for analyzing steroid estrogens in water.
Thus, the signal intensity of native estrogens in ESI(�) was
designated as a reference to evaluate the performance among
different methods (Table 3).
In simple matrices such as MQ water and drinking water,
dansyl-estrogens produced the highest signal intensity of all
five methods, reaching as high as 101 times the reference
Rapid Commun. Mass Spectrom. 2007; 21: 1973–1983
DOI: 10.1002/rcm
Analysis of steroid estrogens in water by LC/MS/MS 1981
values in MQ water and 111 times the reference values
in drinking water. FMP-estrogens produced better signals
for FMP-E1, FMP-E2 and FMP-EE2, while FMP-E3 did not.
E3 possesses three hydroxyl groups, and FMPTS may not
react specifically with the phenolic one. The byproducts
might not produce identical fragment ions and, therefore, we
might not have detected them. Underivatized E2, E3 and EE2
detected under APCI(�) showed similar intensities to that of
ESI(�), while E1 had 3 times the intensity. PFB-estrogen
signals were enhanced by 3 to 9 times that of the under-
ivatized analytes using ESI(�).
In complex matrices such as river water and STP effluents,
dansyl-estrogen signals were increased significantly up to
2 or 3 orders of magnitude greater than those of under-
ivatized estrogens using ESI(�). Signal intensity of native
estrogen using APCI(�) and PFB derivatization increased by
1 to 2 orders of magnitude compared with the reference
value. The signal intensity of FMP-estrogens ranged from
3–31 times the reference value.
Relative sensitivity between different methodsDifferent from absolute responses, the noise levels were
considered when comparing relative sensitivities. Within the
samematrix, S/N values of underivatized estrogens detected
under ESI(�) were taken as the reference to compare the
relative sensitivities of different analytical methods (Table 4).
In simple matrices such as MQ water and drinking water,
dansyl-estrogens and PFB-estrogens produced significantly
higher sensitivities than the references. In addition, dansyl-
estrogens provided much better sensitivities than PFB-
estrogens. In MQ water, sensitivities of dansyl-estrogens
Table 4. Comparison of sensitivity in different estrogen derivative
Native estrogensin ESI(�) Dansyl-estrogens
MQ waterE1 199� 24a,b 2269� 506b,c,d,e
E2 257� 56a,b 1963� 404b,c,d,e
E3 281� 21a,b 1780� 393b,c,d,e
EE2 188� 26a,b 1068� 190b,c,d,e
Drinking waterE1 156� 23a,b 1145� 70b,c,d,e
E2 193� 42a,b 1436� 238b,c,d,e
E3 173� 46a,b 1133� 198c,d,e
EE2 149� 27a,b 1423� 380b,c,d,e
River waterE1 43.6� 14.6b,e 144� 47b,e
E2 ND 166� 32E3 5.42� 0.36a,b,e 272� 33c,d,e
EE2 ND 158� 39b,d,e
STP effluentsE1 80.6� 11.0b 171� 14b
E2 31.0� 7.5a,b,e 87.7� 14.7b,c,d
E3 5.23� 1.58a,b,e 104� 5.3b,c,d,e
EE2 8.47� 1.17a,b,e 59.5� 5.7b,c
Note. Sensitivity is defined as the signal-to-noise (S/N) ratio; 10 ng equivbND: not detectable (S/N ratio <3).a, Statistically different from dansyl-estrogens.b, Statistically different from PFB-estrogens.c, Statistically different from native estrogens in ESI(�).d, Statistically different from FMP-estrogens.e Statistically different from native estrogens in APCI(�).
Copyright # 2007 John Wiley & Sons, Ltd.
were 5.7–11.4 times higher than the reference values,
whereas those of PFB-estrogens were 3.3–5.2 times higher.
APCI(�) produced slightly higher intensities (1.1–3.1 times)
than ESI(�), though the difference was not significant.
Except for E3 (only one-tenth of the reference value), the
performance of FMP-estrogens was similar to that of the
references. In drinking water, dansyl, FMP, underivatized in
APCI(�) and PFB increased S/N by 7.3–9.6, 0.1–2.2, 1.8–3.3
and 4.0–5.6 times, respectively, compared to underivatized
estrogens in ESI(�).
Significant enhancement of sensitivity was observed in
dansyl-estrogens, PFB-estrogens and underivatized estro-
gens under APCI(�) compared with the reference levels
measured when the five methods were applied to the
complex matrices. Different from that observed in cleaner
matrices, in most cases the sensitivities of PFB-estrogens in
river water and STP effluents were better than those of
dansyl-estrogens. Because native E2 and EE2 were not
detectable (S/N<3) in river water under ESI(�), we assigned
them a S/N value of 1.5 (half of the detection limit) to
facilitate method comparisons. In river water, compared to
the reference values, the enhanced sensitivities of
PFB-estrogens were 11.5 times for PFB-E1, 202 times for
PFB-E2, 49.2 times for PFB-E3 and 264 times for PFB-EE2. The
performances of dansyl-estrogens were 3.3 times greater for
dansyl-E1, 110 times for dansyl-E2, 50.1 times for dansyl-E3
and 105 times for dansyl-EE2. The relative sensitivities of
FMP-estrogens were 1.4–11.8 times higher than reference
values. Under APCI(�), the sensitivity of underivatized
estrogens increased 13.2 times for E1, 139 times for E2, 32.8
for E3 and 33.7 times for EE2. In STP effluents, the sensitivity
of PFB-estrogens was increased 9.59–31.9 times and that of
s (S/N, mean�SD, n¼ 4)
FMP-estrogensNative estrogens
in APCI(�) PFB-estrogens
354� 61a,b 615� 79a 1030� 206a,c,d
259� 41a,b 489� 23a 890� 171a,c,d
29.5� 6.4a,b 306� 20a,b 918� 95a,c,d,e
222� 24a,b 266� 19a,b 642� 108a,c,d,e
335� 58a 515� 75a 881� 146a,c
161� 12a,b,e 444� 79a,b,d 774� 66a,c,d,e
23.8� 4.0a,b 316� 23a,b 860� 277c,d,e
162� 42a,b 262� 70a 627� 125a,c,d
60.2� 6.7b,e 578� 70a,c,d 501� 179a,c,d
16.9� 3.6b 208� 138 303� 138d
31.3� 10.9a,b,e 178� 24a,b,c,d 267� 78c,d,e
17.6� 5.2a,b 50.6� 16.9a,b 395� 8.9a,d,e
114� 14b 569� 70 976� 523a,c,d
33.0� 4.8a,b,e 116� 6.2b,c,d 298� 44a,c,d,e
16.5� 2.5a,b,e 72.6� 3.6a,b,c,d 129� 10a,c,d,e
33.6� 7.2b 50.6� 16.9b,c 270� 34a,c,d,e
alent of each parent chemical was injected theoretically.
Rapid Commun. Mass Spectrom. 2007; 21: 1973–1983
DOI: 10.1002/rcm
1982 Y.-H. Lin, C.-Y. Chen and G.-S. Wang
dansyl-estrogens 2.13–19.9 times. The increases in sensitivity
of FMP-estrogens and underivatized estrogens under
APCI(�) ranged from 1.06–3.97 times and 3.73–13.9 times
the reference values, respectively.
Matrix effects in analytical methodsEstrogens detected under ESI(�) were all susceptible to
matrix effects. The signals of underivatized analytes were
seriously suppressed in river water and STP effluents. In
river water, the MEFs were below 1% except for E1 (19.4%).
In STP effluents, the MEFs were below 1% for E3, 3.2% for
EE2, 11.6% for E2 and 28.6% for E1. MEFs of analytes in
drinking water ranged from 61.3–79.1% (Table 5).
Although dansyl derivatization gave good signal intensity,
the MEFs were similar to those of underivatized ones. For
dansyl-estrogens, MEFs ranged from 5.85–14.2% in river
water, under 1–5.50% in STP effluents and 50.4–72.9% in
drinking water (except for 30% of signal enhancement in
EE2), respectively. Thematrix effects on FMP-estrogens were
also severe. In river water, MEFs of FMP-estrogens ranged
from 6.46–13.8%, with the exception of FMP-E3, which had
an MEF of 83.6%. In STP effluents, MEF was higher (42.4%)
for FMP-E3, while for the other three analytes the MEFs
ranged between 11.2–17.1%. The reason that the MEFs of
FMP-E3 were higher than those of other three FMP-
derivatives might be ascribed to its much lower S/N ratio
in MQ water, which was the denominator to calculate the
MEFs (Table 4). The MEFs of FMP-E3 in drinking water
ranged from 61.7–94.9% (Table 5).
Table 5. Matrix effect factors (MEFs, %) of the analytical
methods
MQwater
Drinkingwater
Riverwater
STPeffluents
Native ESI(�)E1 100% 78.2% 19.4% 28.6%E2 100% 74.8% 0.59% 11.6%E3 100% 61.3% 0.54% 0.54%EE2 100% 79.1% 0.80% 3.20%
Dansyl chlorideE1 100% 50.4% 5.85% 0.86%E2 100% 72.9% 8.00% 3.96%E3 100% 63.7% 11.5% 5.50%EE2 100% 133% 14.2% 5.27%
FMPTSE1 100% 94.9% 13.8% 17.1%E2 100% 61.7% 5.29% 11.2%E3 100% 84.1% 83.6% 42.4%EE2 100% 72.5% 6.46% 13.7%
Native APCI(�)E1 100% 82.9% 84.7% 61.3%E2 100% 90.9% 41.0% 22.9%E3 100% 103% 50.1% 23.3%EE2 100% 98.6% 16.4% 18.1%
PFBBrE1 100% 86.0% 45.3% 52.9%E2 100% 87.4% 33.5% 32.1%E3 100% 94.1% 25.9% 13.7%EE2 100% 98.2% 61.5% 41.2%
Note. Matrix effect factors (MEFs) are defined as the quotient of theS/N value between environmental matrices and MQ water; n¼ 4;10 ng equivalent of each parent chemical was injected theoretically.
Copyright # 2007 John Wiley & Sons, Ltd.
Analytes detected using the APCI interface suffered less
ion suppression than when using the ESI source. Under-
ivatized estrogens produced MEFs ranging from 16.4–84.7%
in river water, 18.1–61.3% in STP effluents and 82.9–103% in
drinking water. E1 was less susceptible to matrix effects than
the other estrogens; its ion suppression was relatively minor
in river water and STP effluents. PFB-estrogens were also less
sensitive tomatrix components. TheMEFs of analytes ranged
from 25.9–61.5% in river water and from 13.7–52.9% in STP
effluents. Higher MEFs were observed in drinking water,
where they ranged between 86.0–98.2% (Table 5).
Feasibility of derivatization methods to realwater samplesTo evaluate the feasibility of derivatization methods to real
water samples, limits of detection (LODs) (based on S/N¼ 3)
were estimated under the assumption that the signal
responses were linear up to the concentrations we used.
Amounts of 50 ng of steroid estrogen standards were spiked
in real water samples (from 2-L samples that were equally
divided into five aliquots for testing five analytical methods).
Each 400-mL sample was equal to a concentration of 125 ng/
L of each analyte. The estimated LODs of PFB-estrogenswere
0.38–2.9 ng/L in STP effluents, 0.75–1.4 ng/L in river water
and 0.43–0.60 ng/L in drinking water. These detection limits
are low enough that they may be used to measure typical
concentrations of estrogens in environmental waters. The
estimated LODs of dansyl-estrogens were 2.2–6.3 ng/L in
STP effluents, 1.4–2.6 ng/L in river water and 0.26–0.33 ng/L
in drinking water. Without complex matrices, dansyl
derivatization may provide low LODs, which is useful for
monitoring human exposure to estrogens in drinking water.
The estimated LODs of FMP-estrogens were 3.3–23 ng/L in
STP effluents, 6.2–22 ng/L in river water and 1.1–16 ng/L in
drinking water. Reactivity of FMPTS to hydroxyl groups
other than those specific to phenolic ones would be the
primary cause of higher LODs of steroid estrogens. Chemical
derivatization can reduce the required sample volume,
which will not only facilitate the extraction of phenolic
chemicals, but also mitigate the matrix effect.
CONCLUSIONS
This study demonstrates that derivatization with dansyl
chloride and PFBBr improves signal intensity in mass
spectrometry and can be used to access exposure to estrogens
in environmental water samples. While dansyl chloride
derivatization produced more intense signals in a simple
matrix such as drinking water, significant ion suppression
occurred when it was used to measure estrogens in river
water and STP effluents. APCI-based PFBBr derivatization
would be more suitable in the detection of estrogens in
complex matrices because they are less affected by their
complexity. Although FMP-derivatives can be used to
enhance ionization in the ESI source, the reagent did not
exclusively react with the phenolic group,whichmight lower
derivatization yields or possibly increase background noises
in complex environmental matrices.
PFB-estrogens possess both qualitative and quantitative
benefits in the detection of estrogens because their product
Rapid Commun. Mass Spectrom. 2007; 21: 1973–1983
DOI: 10.1002/rcm
Analysis of steroid estrogens in water by LC/MS/MS 1983
ions originate from the fragmentation of parent steroid
estrogens. They were also found to be more sensitive than
native estrogens. Although dansyl chloride derivatization
significantly enhanced signal intensity, the product ions of
dansyl-estrogens originate from the bonded derivatization
reagents, making it possible that false positive results would
occur when there are phenols with identical molecular
weights to target steroid estrogens, since all dansyl-
derivatized precursors will produce intense product ions of
m/z 171þ. Therefore, in this case, better sample preparation to
remove interferences or optimized LC conditions to avoid
co-eluting components will be needed. Based on our results,
PFBBr derivatization under APCI(�) with methanol and
water as its mobile phase composition can best be used to
detect phenolic chemicals in complex water samples, such as
river water and STP effluents. Dansyl chloride derivatization
under ESI(þ) with 10mM formic acid and acetonitrile as its
mobile phase can best be used to detect them in simple
matrices, such as MQwater and drinking water. However, if
dansyl chloride is used in complex matrices, then the sample
would need more cleanup to mitigate the matrix effect.
This study describes in detail the derivatization steps,
matrix effect evaluations, chromatographic and mass
spectrometric parameters which can be applied to future
research involving the detection of phenolic chemicals.
Further research should focus on validating this method of
quantifying steroid estrogens and other phenolic pollutants
in different waters.
AcknowledgementsThe authors are grateful for kind comments from Drs. Guor-
Rong Her and Pao-Chi Liao. The LC/MS/MS system
was provided by the College of Medicine, National Taiwan
University. This work is supported by the National Science
Council, Taiwan (NSC94-2314-B002-278).
REFERENCES
1. Gomes RL, Scrimshaw MD, Lester JN. Trends Anal. Chem.2003; 22: 697. DOI: 10.1016/S0165-9936(03)01010-0.
2. D’Ascenzo G, Di Corcia A, Gentili A, Mancini R, Mastro-pasqua R, Nazzari M, Samperi R. Sci. Total Environ. 2003;302: 199. DOI: 10.1016/S0048-9697(02)00342-X.
3. Lai KM, Scrimshaw MD, Lester JN. Crit. Rev. Toxicol. 2002;32: 113.
4. Mills LJ, Chichester C. Sci. Total Environ. 2005; 343: 1. DOI:10.1016/j.scitotenv.2004.12.070.
5. Matsui S, Takigami H, Matsuda T, Taniguchi N, Adachi J,Kawami H, Shimizu Y. Water Sci. Technol. 2000; 42: 173.
6. Ternes TA, Andersen H, Gilberg D, Bonerz M. Anal. Chem.2002; 74: 3498. DOI: 10.1021/ac015717z.
Copyright # 2007 John Wiley & Sons, Ltd.
7. Mol HGJ, Sunarto S, Steijger OM. J. Chromatogr. A 2000; 879:97. DOI: 10.1016/S0021-9673(00)00124-2.
8. Kelly C. J. Chromatogr. A 2000; 872: 309. DOI: 10.1016/S0021-9673(99)01261-3.
9. Fine DD, Breidenbach GP, Price TL,Hutchins SR. J. Chroma-togr. A 2003; 1017: 167. DOI: 10.1016/j.chroma.2003.08.021.
10. Mouatassim-Souali A, Tamisier-Karolak SL, Perdiz D,Cargouet M, Levi Y. J. Sep. Sci. 2003; 26: 105. DOI:10.1002/jssc.200390000.
11. Diaz-Cruz MS, de Alda MJL, Lopez R, Barcelo D. J. MassSpectrom. 2003; 38: 917. DOI: 10.1002/jms.529.
12. Mei H, Hsieh YS, Nardo C, Xu XY, Wang SY, Ng K,Korfmacher WA.Rapid Commun.Mass Spectrom. 2003; 17: 97.DOI: 10.1002/rcm.876.
13. Gao SM, Zhang ZP, Karnes HT. J. Chromatogr. B 2005; 825:98. DOI: 10.1016/j.jchromb.2005.04.021.
14. Higashi T, Shimada K. Anal. Bioanal. Chem. 2004; 378: 875.DOI: 10.1007/s00216-003-2252-z.
15. Anari MR, Bakhtiar R, Zhu B, Huskey S, Franklin RB,Evans DC. Anal. Chem. 2002; 74: 4136. DOI: 10.1021/ac025712h.
16. Quirke JME, Adams CL, Van Berkel GJ. Anal. Chem. 1994;66: 1302. DOI: 10.1021/ac00080a016.
17. Singh G, Gutierrez A, Xu KY, Blair IA.Anal. Chem. 2000; 72:3007. DOI: 10.1021/ac000374a.
18. Rogatsky E, Stein D. J. Am. Soc. Mass Spectrom. 2005; 16:1757. DOI: 10.1016/j.jasms.2005.07.012.
19. Stuber M, Reemtsma T. Anal. Bioanal. Chem. 2004; 378: 910.DOI: 10.1007/s00216-003-2442-8.
20. Taylor PJ. Clin. Biochem. 2005; 38: 328. DOI: 10.1016/j.clin-biochem.2004.11.007.
21. Fernandez-Figares I, Rodriguez LC, Gonzalez-Casado A.J. Chromatogr. B 2004; 799: 73.DOI: 10.1016/j.jchromb.2003.10.012.
22. Matuszewski BK, Constanzer ML, Chavez-Eng CM. Anal.Chem. 2003; 75: 3019. DOI: 10.1021/ac020361s.
23. Li WK, Li YH, Li AC, Zhou SL, Weng ND. J. Chromatogr. B2005; 825: 223. DOI: 10.1016/j.jchromb.2005.01.012.
24. Shou WZ, Jiang XY, Weng ND. Biomed. Chromatogr. 2004;18: 414. DOI: 10.1002/bmc.329.
25. Nelson RE, Grebe SK, O’Kane DJ, Singh RJ. Clin. Chem.2004; 50: 373. DOI: 10.1373/clinchem.2003.025478.
26. Mukaiyama T, Ikeda S, Kobayashi S. Chem. Lett. 1975: 1159.27. Bald E. J. Chromatogr. 1979; 174: 483. DOI: 10.1016/S0021-
9673(00)86029-X.28. Chen C-Y,Wen T-Y,Wang G-S, Cheng H-W, Lin Y-H, Lien
G-W. Sci. Total. Environ. 2007; DOI: 10.1016/j.scitotenv.2007.02.038.
29. Council Directive 96/23/EC Official Journal of the EuropeanCommunities, 17.8.2002, L221/8.
30. Deruiter C, Otten RR, Brinkman UAT, Frei RW.J. Chromatogr. 1988; 436: 429. DOI: 10.1016/S0021-9673(00)94602-8.
31. Penzes LP, Oertel GW. J. Chromatogr. 1970; 51: 325. DOI:10.1016/S0021-9673(01)96876-1.
32. Dunphy JC, Pessler DG, Morrall SW. Environ. Sci. Technol.2001; 35: 1223. DOI: 10.1021/es001491q.
33. Nakagawa Y, Hashimoto Y. J. Mass Spectrom. Soc. Jpn. 2002;50: 330.
34. Barry SJ, Carr RM, Lane SJ, Leavens WJ, Monte S,Waterhouse I. Rapid Commun. Mass Spectrom. 2003; 17:603. DOI: 10.1002/rcm.957.
35. Sheen JF, Her GR. Anal. Bioanal. Chem. 2004; 380: 891. DOI:10.1007/s00216-004-2877-6.
Rapid Commun. Mass Spectrom. 2007; 21: 1973–1983
DOI: 10.1002/rcm