Application of complementary mass spectrometric techniques to the identification of ketoprofen phototransformation products

Article abstract of DOI:10.1002/jms.1906

Ketoprofen (KP) is a nonsteroidal anti-inflammatory drug, which during UV irradiation rapidly transforms into benzophenone derivatives. Such transformation products may occur after topical application of KP, which is then exposed to sunlight resulting in a photo-allergic reaction. These reactions are mediated by the benzophenone moiety independently of the amount of allergen. The same reactions will also occur during wastewater or drinking water treatment albeit their effect in the aqueous environment is yet to be ascertained. In addition, only a few such transformation products have been recognised. To enable the detection and structural elucidation of the widest range of KP transformation products, this study applies complementary chromatographic and mass spectrometric techniques including gas chromatography coupled to single quadrupole or ion trap mass spectrometry and liquid chromatography hyphenated with quadrupole-time-of-flight mass spectrometry. Based on structural information gained in tandem and multiple MS experiments, and on highly accurate molecular mass measurements, chemical structures of 22 transformation products are proposed and used to construct an overall breakdown pathway. Among the identified transformation products all but two compounds retained the benzophenone moietya€"a result, which raises important issues concerning the possible toxic synergistic effects of KP and its transformation products. These findings trigger further research into water treatment technologies that would limit their entrance into environmental or drinking waters. Copyright

Full text of DOI:10.1002/jms.1906

Research Article
Received: 17 November 2010
Accepted: 18 February 2011
Published online in Wiley Online Library: 2011
(wileyonlinelibrary.com) DOI 10.1002/jms.1906
Application of complementary mass
spectrometric techniques to the identification
of ketoprofen phototransformation products
∗
ˇ
ˇ
Tina Kosjek, Silva Perko, Ester Heath, Bogdan Kralj and Dusan Zigon
Ketoprofen (KP) is a nonsteroidal anti-inflammatory drug, which during UV irradiation rapidly transforms into benzophenone
derivatives. Such transformation products may occur after topical application of KP, which is then exposed to sunlight resulting
in a photo-allergic reaction. These reactions are mediated by the benzophenone moiety independently of the amount of
allergen. The same reactions will also occur during wastewater or drinking water treatment albeit their effect in the aqueous
environment is yet to be ascertained. In addition, only a few such transformation products have been recognised. To enable
the detection and structural elucidation of the widest range of KP transformation products, this study applies complementary
chromatographic and mass spectrometric techniques including gas chromatography coupled to single quadrupole or ion trap
mass spectrometry and liquid chromatography hyphenated with quadrupole–time-of-flight mass spectrometry. Based on
structural information gained in tandem and multiple MS experiments, and on highly accurate molecular mass measurements,
chemical structures of 22 transformation products are proposed and used to construct an overall breakdown pathway. Among
the identified transformation products all but two compounds retained the benzophenone moiety – a result, which raises
important issues concerning the possible toxic synergistic effects of KP and its transformation products. These findings
trigger further research into water treatment technologies that would limit their entrance into environmental or drinking
c
waters. Copyright ꢀ 2011 John Wiley & Sons, Ltd.
Keywords: ketoprofen; transformation product; mass spectrometry; identification; photodegradation
INTRODUCTION
α-(3-ethylphenyl)phenylmethanol.[10] The presence of several
dimeric photoproducts has also been suggested.[10–12] To date, re-
search has focused on the in vivo[9] and in vitro[8,9,13] phototoxicity
of KP and its benzophenone TPs, and the photochemical mech-
anisms of their degradation routes.[11,12,14] Further breakdown of
the benzophenone ring system including its cleavage would pre-
sumably minimise the skin photosensitisation reactions mediated
by the benzophenone ring system, which to our knowledge is yet
to be reported.
Mass spectrometry is considered a principal tool for identify-
ing TPs, especially since it enables an efficient analysis of trace
amounts of analytes in complex organic mixtures. Such compli-
cated environmental or biological samples require separation of
components prior to mass spectrometric analysis, which justifies
the use of hyphenated techniques such as GC-MS or LC-MS. In
this work, we applied three mass spectrometers, a single-stage
quadrupole (Q) and an ion trap (IT) both hyphenated to a GC,
and a quadrupole–time-of-flight (QqToF) MS coupled to an LC. In
the Q mass analyser, the ions generated in the source undergo
electron impact (EI) fragmentation, which results in complex, am-
biguous spectral data and hence in nonselectivity that is its main
disadvantage.[15] In contrast, the IT mass detector has the unique
ability to isolate and to accumulate ions. By iterating ion trapping
Ketoprofen [2-(3-benzoylphenyl)propanoic acid] (KP) is a common
nonsteroidal anti-inflammatory drug important for its analgesic,
antipyretic and anti-inflammatory activity. Despite its success, it is
associated with certain adverse effects, such as peptic ulcer, acute
renal failure, bronchospasm[1] and skin photosensitivity disorders
including phototoxic, after oral absorption, and photo-allergic af-
ter topical application.[2–4] Whereas phototoxicity is unrelated to
the immunity response, i.e. every person administered the drug
could be affected if exposed to sunlight (UV radiation), the photo-
allergic reaction is a hypersensitivity disorder of the immune
system and symptom severity is neither related to dosage nor
the intensity of the UV radiation.[3,4] Therefore, the photo-allergic
reactions may arise not only from the topical KP formulations
but also from potable and environmental waters, where KP is
commonly found.[5,6] It is considered that both pharmacological
mechanisms correlate to each other in that the generation of
free radicals from the photolabile KP molecule attack human cells
(phototoxicity) and bind to neighbouring skin proteins so that
the antigen is grown (photo-allergy).[7] These photosensitivity
reactions are associated with the benzophenone chromophore
incorporated into the KP structure.[8,9] Besides the parent com-
pound, there are also the radical intermediates[10] and the final
phototransformation products (TPs) formed by the rapid KP
decomposition under UV irradiation that preserve the benzophe-
none chromophore and are responsible for the photosensitisation
observedin vivo,viatheformationofoxidativestress.[4] Thesecom-
pounds include 3-ethylbenzophenone (EtBP),[10] 3-acetylbenzo-
phenone (AcBP),[10,11] 3-(α-hydroxyethyl)-benzophenone[10] and
∗
Correspondence to: Tina Kosjek, Department of Environmental Sciences, Jozˇef
Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia. E-mail: tina.kosjek@ijs.si
Department of Environmental Sciences, Jozˇef Stefan Institute, Ljubljana,
Slovenia
c
J. Mass. Spectrom. 2011, 46, 391–401
Copyright ꢀ 2011 John Wiley & Sons, Ltd.

T. Kosjek et al.
and scanning – it allows the generation of collision-induced dis-
sociation (CID) spectra of the parent and fragment ions (and their
fragmentions), thus increasing the level of confidence in assigning
a particular structure.[15] Alternatively, the hybrid QqToF, in which
the final resolving mass filter of a triple Q is replaced by a ToF
analyser, not only allows MS2 operation but also has the neces-
sary accuracy and resolution to give exact-mass measurements.
Together with MS methods, both chromatographic techniques
complement each other to account for a wide range of polar-
ity, acido-basic characteristics and different functional groups,
which fits our goal of determining the highest possible number of
TPs formed during UV degradation of KP. Accordingly, the main
objective of this work is to demonstrate an application of comple-
mentary chromatographic and mass spectrometric techniques for
identification of the TPs formed during the UV breakdown of KP
and to integrate them into a breakdown pathway.
The SPE cartridges were first conditioned with 3 ml of ethylacetate
and 3 ml of methanol, equilibrated with 3 ml of MilliQ water
(pH 2–3 water in case of acidified sample matrix) and enriched
at a flow-rate of approx. 2 ml min−1. The cartridges were then
dried for 30 min under vacuum and eluted with 1 ml of acetone,
1 ml of ethylacetate/acetone (7 : 3) and 1 ml of ethylacetate. The
combined eluant was evaporated to dryness with nitrogen. The
extracts were then dissolved in either 0.5 ml of ethylacetate for
analysis by gas GC-MS or in 0.5 ml of 1 : 4 acetonitrile/water for
LC-MS analysis. For derivatisation, 30-µl MTBSTFA or MSTFA was
added to the ethylacetate extracts and left to react at 60 ◦C for 1 h.
Two internal standards were spiked into the water samples prior to
the SPE, where ibuprofen-d3 (2.4 nM) was used for quantification
of low-level compounds, and ibuprofen (0.58 µM), which was used
for quantification of high-level compounds.
Instrumental analysis
GC-MS
EXPERIMENTAL
HP 6890 series (Hewlett-Packard, Waldbron, Germany) gas
chromatograph fitted with a single quadrupole mass selective
detector (GC-MSD) was used. The GC oven was programmed
as follows: an initial temperature of 65 ◦C was held for 2 min,
then ramped at 30 ◦C/min to 180 ◦C, at 7 ◦C/min to 280 ◦C, at
40 ◦C/min to 300 ◦C, and finally held for 3 min. The total GC
run time was 22.44 min. A DB-5 MS 30 m × 0.25 mm × 0.25 µm
(Agilent J&W, CA, USA) capillary column was used, with He as
the carrier gas (37 cm s−1). One-microlitre samples were injected
at 250 ◦C in splitless mode, and the transfer line was maintained
at 280 ◦C. The MS was operated in EI ionisation mode at 70 eV.
In SCAN mode, masses from m/z 50 to 550 were scanned, while
in SIM mode, target fragment ions were modified depending on
the target compounds. Where possible, one quantification ion
(normally the base peak fragment ion) and two confirmation ions
per compound were followed. The GC-MSD used Chemstation
software for instrumental control and data processing.
Standards, reagents and chemicals
KP (3-benzoyl-α-methylbenzeneacetic acid, CAS: 22 071-15-4) was
of ≥98% purity and was obtained from Sigma (St. Louis, MO,
USA). 3-AcBP (CAS: 66 067-44-5) was purchased at Synchem
OHG (Felsberg-Altenburg, Germany). 3-Isopropyl-benzophenone
(iPrBP) and 3-ethylbenzophenone (EtBP) were synthesised using
Friedel-CraftsalkylationatFacultyofChemistryandChemicalEngi-
neering, University of Ljubljana (Ljubljana, Slovenia). Two internal
standards, ibuprofen and ibuprofen-d3, were applied. Ibuprofen
[2-(4-(2-methylpropyl)phenyl)propanoic acid, CAS: 15 687-27-1]
was of 98% purity and was obtained from Sigma (St. Louis, MO,
USA), while ibuprofen-d3 (α-methyl-d3) was purchased from Dr
Ehrenstorfer GmbH (Ausburg, Germany). The derivatising agents
were N-methyl-N-[tert-butyldimethylsilyl]trifluoroacetimide (MTB-
STFA), which was purchased at Acros Organics (NJ, USA), and
N-methyl-N-[trimethylsilyl]trifluoroacetamide (MSTFA), obtained
from Sigma Aldrich (St. Louis, MO). All applied solvents (ethy-
lacetate, acetone, acetonitrile, methanol, MilliQ) and chemicals
(hydrochloric acid 37%, formic acid 50% and hydrogen peroxide
30%) were of analytical grade purity.
GC-MS/MS
A Varian 450-GC hyphenated with an ion trap 240-MS mass
spectrometer (GC-IT) was employed for MS/MS analyses. For
chromatographic separation a Varian, Inc. FactorFour VF-5 ms
(30 m × 0.25 mm × 0.25 µm) analytical column was used, and
the instrument was operated under the same chromatographic
conditions(GCoven, gasflow, injectorandinterfacetemperatures)
asGC-MSD.One-microlitresampleswereinjectedinsplitlessmode,
kept until 0.5 min, and then split 1 : 50. The IT was operated in EI
ionisation mode with an external ionisation source, where in the
full scan type masses from 70 to 400 m/z were scanned using an
automatic gain control prescan to achieve target TIC of 20 000
counts. For further mass fragmentation in the IT cavity, the MS/MS
and MSn ion preparation modes in the resonant waveform were
utilised. MS Workstation v6.9.3 software was used for control,
automation and processing.
UV irradiation
One-litre tap water solutions of individual test compounds (KP,
EtBP and AcBP) with an initial concentration of 0.6 µM were
exposedtoUVirradiation. Itsdurationdependedonanelimination
efficiency of each test compound, involving three time intervals:
0–20, 0–90 and 0–435 min. Treatment was carried out using a
low pressure (LP) monochromatic UV lamp with a peak emission
at 254 nm and a medium pressure (MP) lamp with pyrex glass
filter. Due to high resistance of EtBP and AcBP to UV irradiation,
the efficiency of photocatalysis was tested by adding 0.01% (v/v)
H2O2 into the test solutions prior to 9-min irradiation by LP lamp.
Sample preparation
LC-MS/MS
Sample preparation was based on solid phase extraction (SPE)
using an Oasis Hydrophilic-Lipophilic Balance (HLB) reversed-
phase sorbent (Waters, Corp., Milford, MA, USA). The extraction
wasperformedonparallelsamplesinacidicandneutralconditions.
Oneseriesof200-mlsampleswasacidifiedtopH2–3with37%HCl,
whileinthesecondseries, thematrixpHwasleftunadjusted(pH6).
The chromatographic separation was performed on a Waters
Acquity ultra-performance liquid chromatograph (Waters Corp.,
Milford, MA, USA), equipped with a binary solvent delivery system
and an autosampler. The injection volume was 5 µl. Separation
was achieved using a 3-cm-long Acquity UPLC BEH C18 (Waters
c
wileyonlinelibrary.com/journal/jms
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2011, 46, 391–401

Application of complementary mass spectrometric techniques
Corp., Milford, MA, USA) column with 1.7-µm particle size and 2.1-
mm internal diameter. Compounds were analysed under positive
[ESI(+)] and negative [ESI(−)] ion conditions. In ESI(+) the mobile
phases (A) MilliQ water and (B) acetonitrile were used, while in
ESI(−) water was replaced by 0.1% formic acid as the mobile
phase A, whereas other LC conditions were kept unchanged. The
elution gradient was linearly increased from 20 to 80% B in 6 min,
decreased back to 20% in 1 min and then finally kept isocratic for
1 min. Thetotalruntimewas8 min. Flowratewas0.2 ml min−1 and
thecolumntemperaturewasmaintainedat40 ◦C.TheUPLCsystem
was interfaced to a hybrid quadrupole orthogonal acceleration
time-of-flight mass spectrometer (QqToF Premier, Waters, Milford,
MA, USA). The instrument was equipped with an electrospray
ionisation interface operating in ESI(−) and ESI(+). The capillary
voltage was set to 2.8 and 3.0 kV in ESI(−) and ESI(+), respectively,
while the sampling cone voltage was varied between 30 and 40 V.
Source and desolvation temperatures were set to 120 and 200 ◦C,
in MS, tandem (MS2) or multiple (MSn) MS experiments and
(b) highly accurate molecular mass measurements.
Quantitative analysis was performed on GC-IT by automatic
data processing using MS Workstation software. The calibration
curves were prepared in the relevant concentration range from 5
to 600 nM by spiking the reference standards of KP, EtBP and AcBP
into tap water. Six calibration points were prepared following the
SPE procedure as described in the Section on Sample preparation.
Both derivatisation agents and both internal standards were taken
into account. The coefficients of determination (R2) ranged from
0.9926 to 0.9999.
RESULTS AND DISCUSSION
The following two sections describe mass-spectrometric identi-
fication of TPs formed by UV breakdown of KP. The results of
GC-MS identification involve the fragment ions obtained by Q and
IT mass analysers and are summarised in Fig. 1 and Table 1. Ta-
ble 2 includes the LC-MS data on KP and TPs, while their chemical
structures are illustrated in Fig. 2.
respectively. The nitrogen desolvation gas flow rate was 530 l h−1
.
For MS experiments, the first quadrupole was operated in rf-only
mode, while detection was performed in the ToF mass analyser.
MS data were acquired over an m/z range 100–1000 at collision
energy of 4 V. For MS/MS operation, the acquisition range was
between m/z 50 and 1000, and argon was used as the collision
gas at a pressure of 4.5 × 10−3 mbar in the T-wave collision cell.
The MS/MS experiments were performed with collision energy,
varied between 10 and 40 V, to generate product ion spectra
providing the most structural information. Data were collected in
centroid mode, with a scan accumulation time set to 0.25 s and
an interscan delay of 0.02 s. The data station operating software
was MassLynx v4.1. Prior to analysis, the instrument was calibrated
over a mass range 50–1000, using a sodium formate calibration
solution. Reproducible and accurate mass measurements, at a
mass resolution of 10 000, were obtained using an electrospray
dual sprayer with leucine enkephalin ([M − H]− = 554.2615, [M
+ H]+ = 556.2271) as the reference compound. The latter was
introducedintothemassspectrometeralternatingwiththesample
via Waters Lock Spray device. Elemental composition of TPs was
calculatedfromaccuratemassesdeterminedbyHRMSatfollowing
conditions: C:0–80, H:0–120, N:0–8, O:0–30; 5.0 ppm tolerance;
Double bond equivalent (DBE): min = −1.5, max = 50.0.
Identification of TPs by GC-MS
The three most notable TPs observed in the GC chromatograms of
UV-treated KP samples eluted at 10.86 min (TP-1), 12.68 min (TP-2)
and 13.18 min (TP-3). As evident in Table 1, the molecular weight
(MW) of KP is 254, whereas MW of TP-1 shows a 44 Da decrease
from the parent compound suggesting its decarboxylation. The
MW of TP-3 is 16 Da higher than that of TP-1, which is indicative
of a hydroxylated intermediate, and TP-2 yields 2 Da lower MW
than TP-3, thus suggesting its oxidised form. Figure 1 shows how
the structures of TP-1, TP-2 and TP-3 and the parent compound
(KP) yield a phenyl fragment ion with m/z 77 and a m-substituted
benzoyl residual at m/z 177, 133, 147 and 149 for KP, TP-1, TP-2
and TP-3, respectively. Furthermore, all four compounds show
an abundant benzoyl fragment ion at m/z 105, whereas m/z 181
implies a benzophenone ring system in case of KP, TP-1 and TP-2.
TP-3 is an exception, where the 2 Da higher fragment ion at m/z
183 suggests a hydrogenated diphenylmethanol ring system. The
remaining fragment ions indicate decarboxylation at m/z 209 in
case of KP and cleavage of a methyl group at m/z 209 and 211 for
TP-2 and TP-3, respectively. Based on the EI mass fragmentation,
the TPs discussed herein were tentatively identified as EtBP (TP-1),
AcBP (TP-2) and m-acetyl-1,1-diphenylmethanol (TP-3).
TP-4 eluted at 9.83 min and it was formed in the long-term
LP irradiation of EtBP samples in TW matrix in highest amounts
but was also present in the MQ using both the LP and MP lamp.
As described in Table 1 and illustrated in Fig. 1, the compound
with a molecular mass 196 revealed fragment ions at m/z 181
indicating the cleavage of a methyl group from a characteristic
benzophenone fragment ion. The EI mass spectrum also shows
the 3-methylbenzoyl fragment ion at m/z 119 and benzoyl at
m/z 105. The fragmentation of 3-methylbenzophenone (TP-4) was
confirmed by EI-MS/MS experiments and a positive match (91%) in
the National Institute of Standards and Technology (NIST) library.
Derivatisation with silylating agents MTBSTFA and MSTFA was
performed on pre-analysed GC-MS samples to enhance the
chromatographic response of less volatile TPs and to facilitate
their structural elucidation, e.g. by indicating the presence of a
carboxyl or hydroxyl moiety in a TP structure. MTBSTFA yields
tert-butyldimethylsilyl (MTBS) esters or ethers, and is the most
Detection approaches, identification and quantification
ToincludebroadestrangeofTPs,screeningwasmadebyoperating
the GC-MS/(MS) and LC-MS/MS analyses on parallel samples. The
detection of TPs in the GC samples was made either by comparing
the total or extracted ion chromatograms of the treated samples
with those of the control. In the LC-MS/MS analysis, data were
acquired in centroid mode and afterwards processed by the
MetaboLynx application manager embedded into MassLynx
v4.1. software (Waters). The algorithm was programmed to detect
products of expected transformation pathways (i.e. hydroxylation,
decarboxylation and hydroxylation + desaturation) and also to
detect unexpected components. The latter were examined in m/z
150–300 scanning range with 10 Da size of a step scan. Metabolite
traces were searched in the time window 0.50–5.00 min at 0.05-
min peak separation. The presence of transformation products
was investigated in treated samples, while untreated samples
were used as control samples.
Depending on the capabilities of the MS instrumentation, we
employed two established strategies to determine the identity of
unknown compounds, based on (a) structural information gained
c
J. Mass. Spectrom. 2011, 46, 391–401
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms

T. Kosjek et al.
Figure 1. KP and TPs: illustration of fragment ions obtained by EI.
commonly applied silylating agent,[16] probably due to the greater
thermal and hydrolytic stability of the derivatives. EI-MS spectra
of the MTBS derivatives show a characteristic MS fragmentation,
which together with a higher molecular mass of a derivative
improvesthereliabilityanddetectabilityoftheanalyses.[17] Amajor
drawback is its lower reactivity compared with the trimethylsilyl
(TMS) derivatives producing reagents,[18] such as MSTFA. The
derivatisation enabled us to detectTP-5, TP-6 and TP-7 eluted from
the GC column at 9.88, 8.49 and 12.02 min, respectively. TP-5 is an
unstable intermediate which is quickly degraded under intense
irradiation.Itwasdetectedandidentifiedinsamplesobtainedfrom
short-termMPirradiationofEtBP,andtracelevelswerealsoformed
in EtBP/LP and KP/MP samples. The compound was extracted only
under acidic sample preparation conditions. TP-5 was detected as
a MTBS ester with a molecular ion at m/z 266 and a cleavage of
tert-butyl group [M − 57]+ at m/z 209, a typical fragmentation
pattern for MTBS derivatives.[16] The remaining fragment ions
are indicative of the cleavage of a tert-butyl-dimethylsilyl group
to form the deprotonated molecular ion of the underivatised
molecule at m/z 151, the cleavage of an oxygen to form m/z
135 and a characteristic benzoyl fragment ion at m/z 105 (Fig. 1).
Furthermore, the EI-MSn experiments yielded m/z 121, which is
indicative for phenyl-carboxyl fragment ion, suggesting that the
structure of TP-5 corresponds to 3-(hydroxymethyl)benzoic acid.
TP-6 was detected as a TMS derivative with a base fragment ion
at m/z 211, resulting from the cleavage of a methyl group from
TMS ending. The deprotonated molecular ion of the underivatised
compound is found at m/z 163, which further fragments to m/z
147 and 119 by cleaving an oxygen or CO2 group, respectively.
The fragment ion m/z 119 is indicative for acetylphenyl fragment
ion as already shown for TP-2 fragmentation (Fig. 1). It is therefore
assumed that TP-6 is an m-acetyl-benzoic acid. Another TMS
derivative detected by GC-MS was TP-7, yielding the following
fragment ions: the benzoyl at m/z 105, benzophenone at m/z
181 and 135 indicating an underivatised TP-7 after a phenyl
ring is cleaved. The remaining three fragment ions in the chemical
structurewerem/z 270formedbycleavageofamethylgroupfrom
a TMS derivative, and m/z 193 and 255 formed with subsequent
cleavages of a phenyl and a methyl group.
During photocatalytic degradation of EtBP with the addition
of 0.01% H2O2 several new TPs emerged in GC chromatograms,
among them TP-12, TP-13 and TP-14 were most notable. Table 1
shows their retention times and describes their mass spectra. All
three compounds have a MW 226, which is 16 Da higher than EtBP
from which the compounds were formed. Their mass spectra show
cleavage of a methyl group at m/z 211, while m/z 197 replaces
181 in case of KP, TP-1, TP-2 and TP-4, indicating a hydroxylated
benzophenone ring system. Furthermore, TP-12, TP-13 and TP-14
involve a phenyl (m/z 77) and benzoyl (m/z 105) fragment ions, as
well as m/z 133, which analogously to TP-1 indicates an m-ethyl-
benzoyl fragment. The crucial fragment ion in the mass spectra of
TP-12 and TP-13 is m/z 149 (Table 1, underlined), which shows a 16
Da increase from m/z 133, suggesting that hydroxylation occurred
on the phenyl ring marked A (Fig. 1). The compounds TP-12 and
TP-13 show identical fragment ions, but with slight differences
between their ratios (Table 1), and are therefore considered
structural isomers with different position of the hydroxyl group on
the ring A. Alternatively, TP-14 involves a fragment ion m/z 121
insteadof149,suggestingthehydroxylationonringB(Fig. 1).After
MTBSTFA derivatisation, the chromatographic responses of TP-12
and TP-14 decreased by >84%, while the decrease in TP-13 was
less significant. Concurrently, a number of new chromatographic
peaks appeared in the derivatised samples. At least four of them
revealed a MW of 340 [M]+ and, typically, a fragment ion at m/z
c
wileyonlinelibrary.com/journal/jms
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2011, 46, 391–401

Application of complementary mass spectrometric techniques
c
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2011, 46, 391–401
wileyonlinelibrary.com/journal/jms

T. Kosjek et al.
c
wileyonlinelibrary.com/journal/jms
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2011, 46, 391–401

Application of complementary mass spectrometric techniques
283 [M − 57]+, corresponding to fragmentation of a tert-butyl
group from a MTBS derivative. Since the MTBS group is a 115-Da
fragment ion, and based on the decrease in the chromatographic
response of TP-12 and TP-14 after derivatisation, we assume that
these derivatives correspond to MTBS ethers of hydroxylated m-
EtBP, i.e. TP-12, TP-13, TP-14 and others (TP-15) which were not
detected in their underivatised form. In support of this assumption
aretheresultsof MS/MSfragmentationobtainedbyEIionisationin
theITmassdetector.Thefragmentionm/z 225isthedeprotonated
mass of the hydroxylated EtBP and by cleavage of ring hydroxyl
group it forms m/z 209 (Table 1). Alternatively, in case of a side
chain hydroxylation, a loss of a water molecule would have been
evident at m/z 207. Since all the MTBS derivatives of TP-12–15
showequivalentfragmentionsbutindifferentratios,itissupposed
that these compounds are structural isomers differing only in the
position of the respective –OH group on the benzophenone
ring system. MTBS derivatives have improved GC characteristics
comparedtotheirunderivatisedanalogues,whichresultsinhigher
sensitivity and in turn a higher number of detected compounds.
While MTBS derivatives are characterised by a great thermal and
hydrolytic stability, the major drawback is the lower reactivity of
MTBSTFA compared to other silylating reagents[16] and a slower
reaction rate due to steric resistance. Therefore, MTBSTFA was
replaced by more a reactive derivatising agent MSTFA in the
further experiments.
During photocatalytic irradiation of AcBP, three new com-
pounds (TP-19, TP-20 and TP-21) emerged in GC chromatograms
(Table 1). All compounds involve m/z 240 in their mass spectra,
which is assumed to be their MW. As illustrated in Fig. 1, all three
compounds, besides characteristic fragment ions at m/z 77 for
phenyl and m/z 105 for benzoyl, involve also the fragment ion m/z
225, which presumably corresponds to a demethylated fragment,
analogous to the fragmentation of TP-2 (in this case, the parent
compound) at m/z 209. TP-19 further shows the fragment ion
at m/z 197, which corresponds to a hydroxylated benzophenone
ring system, similarly as the compounds TP-12, TP-13 and TP-14.
In case of TP-19, the fragment ions distinctive for determining,
which aromatic ring was hydroxylated were m/z 147 and 121
(Table 1, underlined). The first corresponds to 3-acetyl-benzoyl
fragment ion and m/z 121 analogously with TP-14 indicates a
hydroxybenzoyl fragment ion, both of them suggesting that the
hydroxylation occurred on the ring B (Fig. 1). Compared to TP-19,
TP-20 and TP-21 were more demanding for identification since
their area-under-curve only mounted up to 9 and 24% of that of
TP-19, respectively. Thus, the mass spectra of TP-20 yielded less
fragment ions, but likewise for TP-19, the B-ring hydroxylation is
proposed based on the fragment ion at m/z 147. Alternatively,
TP-21 shows a 16 Da higher m/z 163, from which A-ring hydroxy-
lation is assumed. Table 1 shows how after MSTFA derivatisation
at least four peaks presumably corresponding to the TMS ethers
of hydroxyl-3-AcBP (TP-19–22) appear in the GC chromatogram.
All compounds have in common the fragment ion corresponding
to the MW of derivative at m/z 312 and an abundant fragment ion
at m/z 297 formed after the cleavage of a methyl group from the
TMS ether.
Identification by LC-MS
The ESI(+)MS/MS spectrum of protonated KP at [M + H]+ 255
typically gives a decarboxylated fragment ion at m/z 209 and
194 after further cleavage of a methyl group. In addition, the
ESI(+)MS/MS spectrum shows a phenyl fragment ion (m/z 77), a
matching 3-(1-carboxyethyl)benzoyl residual at m/z 177 to yield
the molecular structure as well as a benzoyl fragment ion (m/z 105)
andacorresponding 3-(1-carboxyethyl)phenylresidualatm/z 149.
Table 2 further details the mass spectrum of KP acquired under
ESI(−) conditions and includes the deprotonated molecule at [M
− H]− 253 and m/z 209 formed with the cleavage of CO2 (Table 2).
The mass spectrum of TP-1 shows a protonated molecule at [M
+ H]+ 211andHRMSanalysisgivestheelementalformulaC15H15
O
at −0.9 ppm mass error. The tandem mass fragmentation of the
protonated molecule produced a fragment ion at m/z 133 after
the cleavage of benzene (m/z 77) and, similar to KP, a benzoyl
fragment ion at m/z 105. According to these data, the TP-1 is
proposed to be EtBP which is cross-confirmed by GC-MS. The
retention time and the fragmentation of EtBP are also confirmed
by comparison with an authentic standard (Table 2).
TP-2 with a protonated molecule at [M + H]+ 225 gives an
elemental composition of C15H13O2 and the following fragment
ions are evident in the ESI(+) mass spectrum: the m/z 147
(C9H7O2)for3-acetylbenzoylfragment,m/z119for3-acetylphenyl,
m/z 105 for benzoyl, m/z 91 for benzyl and the m/z 77 for
phenyl fragment ion. Furthermore, m/z 183 is evident, which
corresponds to protonated benzophenone. According to the
MS/MS fragmentation and accurate mass determination, the
suggested structure of TP-2 is AcBP. The same compound was
proposed by GC-MS identification process and was confirmed by
comparison to an authentic standard (Table 2).
The accurate mass measurement of the protonated molecule
of TP-8 yields the elemental composition C15H15O2, which is two
protons higher than AcBP and implies its reduced form. Similar to
KP, TP-8 shows fragment ions at m/z 209 and 194 corresponding
to C15H13O and C14H10O, respectively (Table 2). The first fragment
ion is formed by the cleavage of H2O and the second, analogously
to KP, after subsequent cleavage of a methyl group. Instead of
m/z 147 for AcBP, TP-8 shows a hydrogenated residual at m/z 149
(C9H9O2), i.e. 3-(hydroxyethyl) benzoyl formed by cleavage of a
phenyl group from [M + H]+ 227. Analogously to the fragment
m/z 119 in case of AcBP, TP-8 again produces a hydrogenated
form at m/z 121 (C8H9O). Based on this fragmentation, the TP-8 is
identified as 3-(α-hydroxyethyl)-benzophenone.
FromTable 2, itisevidentthatTP-9hastheprotonatedmolecule
[M + H]+ 243 with an elemental composition C15H15O3, which
suggests the additional oxygenation of the TP-8 molecule. MS/MS
fragmentation yields the cleavage of water at m/z 225 (C15H13O2)
or methanol at m/z 211 (C14H11O2) corresponding to a protonated
3-(phenylcarbonyl)-benzaldehyde fragment ion. Therefore, it is
supposed that the two oxygen atoms are positioned on the ethyl
side chain of the benzophenone ring system (m/z 182, C13H10O).
Further confirmation of the proposed structure is given by the
cleavage of a phenyl group at m/z 133 (C8H5O2). The MS/MS
spectrum shows also a typical benzoyl fragment ion at m/z 105.
The proposed structure of TP-9 is therefore 3-(1,2-dihydroxyethyl)-
benzophenone.
From Table 1, it can be noted that by addition of H2O2 to initiate
the photocatalytic transformation reactions, the majority of TPs
(TP-12–15 formed by photocatalytic transformation of EtBP and
TP-19–22 formed from AcBP) was subjected to the hydroxylation
of the benzophenone ring system.
TP-10 has the identical elemental composition (C15H15O3) as
TP-9 (Table 2), but there are a few decisive fragment ions in its
MS/MS spectrum that determine the different position of one
hydroxyl group: m/z 225 with elemental composition C15H13O2 is
c
J. Mass. Spectrom. 2011, 46, 391–401
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms

T. Kosjek et al.
Figure 2. Proposed breakdown pathway.
the result of cleavage of an H2O molecule, and m/z 210 (C14H10O2)
if formed by the subsequent cleavage of a methyl group. This
implies that one –OH group is positioned on the side chain,
presumably at α-position, where oxidation typically takes place. In
contrasttoTP-9, themassspectrumofthiscompoundinvolvesm/z
199 (C13H11O2) representing a hydroxylated benzophenone, and
m/z 121 (C7H5O2), a hydroxy-benzoyl fragment. In regards to the
latter two fragment ions, it is assumed that the second hydroxyl
group is positioned on the benzophenone ring system and TP-10
is identified as hydroxy-3-(α-hydroxyethyl)-benzophenone.
As evident from Table 2, TP-11 was found after LP-UV treatment
of AcBP samples and was detected only in ESI(−) mode, showing
a deprotonated molecule at [M − H]− 225. The accurate mass
measurement gave an elemental composition of C14H9O3. The
ESI(−)MS/MS fragmentation resulted in the formation of m/z 181
(C13H9O),suggestingadeprotonatedbenzophenone,analogously
totheparentcompoundAcBP. Thelatterisformedbythecleavage
of CO2, which is a typical dissociation of a carboxylic acid in ESI(−)
mode. According to these data, the assigned structure of TP-11 is
3-benzoylbenzoic acid.
During photocatalytic degradation of EtBP with 0.01% H2O2 at
least six new peaks are revealed in the LC-MS chromatogram (TP-8
and TP-12–18; Table 2). All TPs show a protonated molecule at [M
+ H]+ 227, and accurate mass measurements reveal their common
elemental composition C15H15O2. This implies that these TPs are
isomeric oxygenated EtBP derivatives. TP-16 and TP-17 (Table 2)
eluted at 1.6 and 1.9 min, respectively, and were unfortunately not
sufficiently abundant to enable their structural elucidation by CID.
Furthermore, considering the retention time and MS/MS spectra,
the TP eluting at 2.4 min is assigned as TP-8 [3-(α-hydroxyethyl)-
benzophenone], whichhasbeenpreviouslyidentifiedasaproduct
of photolytic breakdown of KP and EtBP (Table 2). As illustrated in
Table 2, TP-18 elutes 0.1 min prior to TP-8 and shows an MS/MS
spectrumstronglyrelatedtothatofTP-8. Accordingtoitsfragment
ion m/z 209 (C15H13O), which is indicative for the cleavage of
water from [M + H]+ 227, TP-18 involves a hydroxyl group on
the side chain and, being a structural isomer of TP-8, the only
feasible position of the hydroxyl group is on β-carbon of the
ethyl group. For these reasons, the chemical structure of TP-18
is assigned as 3-(β-hydroxyethyl)-benzophenone. The remaining
two TPs formed by photocatalytic breakdown of EtBP elute at
3.0 and 3.2 min, and show similar ESI(+)MS/MS patterns (Table 2),
which are further related to those produced by EI of TP-12–15
(Fig. 1). It is therefore assumed that the compounds detected at
3.0 and 3.2 min in the LC chromatograms are representatives of
ring-hydroxylated benzophenones, assigned in Tables 1 and 2 as
TP-12–15.
By exposing AcBP to photocatalysis, four peaks are revealed
in the LC chromatograms, where two, TP-19 and TP-20, and two,
TP-21 and TP-22, elute close together (Table 2). All four TPs show
the same protonated molecule [M + H]+ 241 and the elemental
composition C15H15O3; though, the accurate mass determination
gives an unusually high mass error (+7.5 and +23.6 ppm, Table 2)
in case of TP-21 and TP-22, which may be the consequence of their
low abundance. Table 2 illustrates most notable ESI(+)MS/MS
fragment ions, where the decisive ones are underlined. The
mass spectra of TP-19 and TP-20, analogously to TP-10, involve
m/z 199 (C13H11O2) representing a hydroxylated benzophenone.
c
wileyonlinelibrary.com/journal/jms
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2011, 46, 391–401

Application of complementary mass spectrometric techniques
Furthermore, m/z 147 (C9H7O2) and m/z 119 (C8H7O) both match
the parent compound AcBP and stand for 3-acetylbenzoyl and
acetylphenyl fragment ion, respectively. Furthermore, m/z 121
(C7H5O2) is a hydroxy-benzoyl fragment and again implies a
hydroxylation. From these data, it is assumed that TP-19 and TP-20
represent isomeric B-ring hydroxylated-3-AcBPs, which is cross-
confirmed by EI ionisation as observed in Table 1 and Fig. 1. In
contrast to TP-19 and TP-20, the mass spectra of TP-21 and TP-22
yield m/z 163 (C9H7O3), which is 16 Da or one oxygen higher than
m/z 147 (C9H7O2), suggesting that the hydroxylation in this case
occurs on the A-ring of the benzophenone ring system. Again this
assumption is confirmed by EI fragmentation, as evident in Table 1
and Fig. 1.
Linking the polarity of KP and its TPs with their elution order
we observe that in LC, the parent compound elutes first (1.2 min).
Given the pKa of 4.7 for this weak acid,[14] the negatively charged
species is the dominant form under given LC conditions, thus
yielding lowinteractionswiththeLCcolumn. Onthecontrary, EtBP
is not charged, and gives less polar interactions with the mobile
phase than other TPs, therefore it elutes at 4.1 min (Table 2). The
elution order is generally reverse in case of the GC separation,
where the least polar compounds m-methylbenzophenone and
m-ethylbenzophenone reach the mass detector before the more
polar hydroxylated TPs (Table 2).
fragmentation pattern, and the comparison with the reference
compound was in this case crucial to confirm the identity of
AcBP. As follows, EtBP and AcBP were individually tested on their
breakdown and further fate under UV irradiation, whereas iPrBP
was excluded from further research.
The breakdown pathway
ThepropositionofaUVbreakdownschemeofKPwasbasedonthe
treatment experiments with three individual reference standards,
i.e. KP, EtBP and AcBP. In respect to the formation of both EtBP
and AcBP in treated KP samples, the formation of solely AcBP
in treated EtBP samples, and the absence of both EtBP and KP
in the treated AcBP samples, we assume that the fundamental
breakdown pathway follows as KP → EtBP → AcBP (Fig. 2). In
agreement with the literature,[20–22] which reported KP to be
one of the most rapidly degrading pharmaceuticals, we achieved
very fast transformation of KP. Conversely, KP transformation
products EtBP and AcBP were found more resistant to both MP
and LP irradiation. It is therefore assumed that in the aqueous
environment, KP easily transforms into its major product EtBF and
into AcBP, which would persist longer.
TP-3 and TP-8 are hydrogenated analogues of AcBP with a
tautomeric position of the –OH group, which were absent in
the treated AcBP samples, whereas they occurred in treated EtBP
samples. Therefore, TP-3 and TP-8 are considered intermediates in
the oxidative formation of AcBP from EtBP. As evident from Fig. 2,
TP-9 and TP-10 are formed by subsequent hydroxylation of TP-8
on the α-hydroxyethyl side chain and benzophenone ring system,
respectively.
Figure 2 illustrates an alternative UV-breakdown pathway of
EtBP implying a side chain shortening (TP-4) and side chain
hydroxylation (TP-7), followed by an oxidative cleavage of
the benzophenone ring system (TP-5). A structurally related
m-substituted benzoic acid is TP-6, which is formed by the
oxidative cleavage of benzophenone ring system from AcBP.
Alternatively, an oxidative transformation on the acetyl side
chain of AcBP leads to the formation of 3-benzoylbenzoic acid
(TP-11). TP-5 and TP-6 are important UV-breakdown products,
since they cease further formation of phototoxic benzophenone
TPs, whereas all the remaining compounds still involve the
benzophenone moiety (Fig. 2). Ironically, in spite of their photo-
inducedadverseeffects, certainbenzophenonederivativesarestill
used as sunscreens,[23] due to their recognised ability to absorb
UVA and UVB light.
Identity confirmation with authentic standards
To enable further insight into the photodegradation pathway of
KP, and to confirm the identities of the proposed TPs, the two
compounds EtBP and iPrBP, initially proposed on the basis of their
GC-MS fragmentation, were synthesised. An authentic standard
of the third proposed compound (AcBP) was purchased. To
confirm the identity of the proposed TPs with reference standards,
we referred to the following acceptability criteria[19]: Ion ratios
for a given TP, measured as the peak area of a quantitation
ion divided by the peak area of the qualifier ion should be
within 20% of the average of the reference compounds’ ion
ratios, and the TP was required to have a retention time within
0.20 min of the reference compounds’ average retention time.
From Table 3, it is evident that EtBP and AcBP clearly meet
these requirements, whereas iPrBP failed to be identified as
the TP.
Furthermore, Table 3 shows how both iPrBP and AcBP share
the same principal fragment ions, and also their mass spectra
are similar (data not shown). It was therefore not possible to
distinguish between the two compounds based only on their EI
Whereas OH radicals are hardly involved in degradation of
KP,[24] Fig. 2 distinguishes TP-12–18 and TP-19–22 from other
Table 3. Confirmation of EtBP and AcBP by authentic standards based on their retention time, NIST 98 similarity and quantifier and qualifier ions’
ratio
Compound
Retention time (min)
EtBP
iPrBP
AcBP
Reference
Analyte
10.91
11.37
12.68
10.86
12.68
12.68
NIST 98 probability (%)
Quantifier/qualifier 1
Quantifier/qualifier 2
Reference
Analyte
93
95
No relevant hits
No relevant hits
m/z 209/224 = 2.0
m/z 209/224 = 2.0
m/z 209/181 = 6.1
m/z 209/181 = 6.4
Confirmed
93
No relevant hits
m/z 209/224 = 0.86
m/z 209/224 = 2.0
m/z 209/181 = 2.7
m/z 209/181 = 6.4
Failed
Reference
Analyte
m/z 133/210 = 1.5
m/z 133/210 = 1.6
m/z 133/181 = 2.8
m/z 133/181 = 3.0
Confirmed
Reference
Analyte
Outcome
c
J. Mass. Spectrom. 2011, 46, 391–401
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms

T. Kosjek et al.
compounds by a dashed arrow and classifies them into a group of
compounds formed strictly by a photocatalytic transformation of
EtBP or AcBP, respectively. Except for TP-18, which was subjected
to a side-chain hydroxylation, the remaining six photocatalytic
TPs of EtBP (TP-12 to TP-17) involve a hydroxyl group positioned
on the benzophenone ring system. Similarly, the photocatalytic
TPs of AcBP were hydroxylated either on the A-ring (TP-21 and
TP-22) or on the B-ring (TP-19 and TP-20) of the benzophenone
ring system. This finding is in agreement with Hayashi et al.,[25]
who suggests an enhanced hydroxylation of benzophenone ring
system in the presence of H2O2. Such transformations may not
only occur during water treatment but also in the environment,
where photosensitisers such as humic acids generate reactive
oxygen species causing the ring hydroxylation of benzophenone
derivatives. Benzophenone is a suspected endocrine-disrupting
chemical.[25] Whereas the estrogenic activity was not shown for
the compound itself, it was found that benzophenone acquired
estrogenic activity after ring hydroxylation.[25] As evident in
Fig. 2, the ring hydroxylation occurred in 11 of 22 identified
TPs, and there exists a strong indication that these compounds
are endocrine disruptors, which calls for further estrogenic activity
testing.
Silvia Pirsˇ (Jozˇef Stefan Institute, Slovenia), Dr Oliver Bajt (National
Institute of Biology, Marine Biological Station, Slovenia), Dr David
Heath (Jozˇef Stefan International Postgraduate School, Ljubljana,
Slovenia), Prof. Jurij Svete and Rok Prebil (Faculty of Chemistry and
Chemical Technology, University of Ljubljana, Slovenia).
REFERENCES
[1] R. F. Borne. In Principles of Medicinal Chemistry, 4th ed., T. L. Lemke,
D. A. Williams, W. O. Foye (Eds). Williams & Wilkins: PA, USA, 1995,
535.
[2] European Medicines Agency, Committee for Medicinal Products
for Human Use (CHMP), Press release: European Medicines
Agency confirms positive benefit-risk balance of topical formu-
lations of ketoprofen. European Medicines Agency: 2010,
EMA/CHMP/465633/2010.
[3] S. Ophaswongse, H. Maibach. Topical nonsteroidal antiinflamma-
tory drugs: allergic and photoallergic contact dermatitis and
phototoxicity. Contact Dermatitis. 1993, 29, 57.
[4] M. C. Marguery, N. Chouini-Lalanne, J. C. Ader, N. Paillous. Com-
parison of the DNA damage photoinduced by fenofibrate and
ketoprofen, two phototoxic drugs of parent structure. Photochem.
Photobiol. 1998, 68, 679.
[5] T. Kosjek, E. Heath, A. Krbavcˇicˇ. Determination of non-steroidal anti-
inflammatory drug (NSAIDs) residues in water samples. Environ. Int.
2005, 31, 679.
Nevertheless, it should be stressed that the proposed UV
breakdown scheme is configured based on treatment exper-
iments using only three reference standards. Therefore, the
proposed breakdown scheme cannot be regarded as complete
and warrants confirmation by reference standards of the remain-
ing TPs.
[6] N. M. Vieno, T. Tuhkanen, L. Kronberg. Seasonal variation in the
occurrenceofpharmaceuticalsineffluentsfromasewagetreatment
plantandintherecipientwater. Environ.Sci.Technol. 2005, 39, 8220.
[7] A. Lahoz, D. Hernandez, M. A. Miranda, J. Perez-Prieto, I. M. Morera,
J. V. Castell. Antibodies directed to drug epitopes to investigate the
structure of drug-protein photoadducts. Recognition of a common
photobound substructure in tiaprofenic acid/ketoprofen cross-
photoreactivity. Chem. Res. Toxicol. 2001, 14, 1486.
[8] A. Radschuweit,H.-H. Ruttinger,P. Nuhn,W. Wohlrab,Chr. Huschka.
UV-induces formation of hydrogen peroxide based on the
photochemisty of ketoprofen. Photochem. Photobiol. 2001, 73, 119.
[9] F. Bosca, M. A. Miranda. Photosensitizing drugs containing benzo-
phenone chromophore. J. Photochem. Photobiol. B, 1998, 43, 1.
[10] F. Bosca´, M. A. Miranda, G. Carganico, D. Mauleon. Photochemical
and photobiological properties of ketoprofen associated with the
benzophenone chromophore. Photochem. Photobiol. 1994, 60, 96.
[11] A. Nakajima, M. Tahara, Y. Yoshimura, H. Nakazawa. Determination
of free radicals generated from light exposed ketoprofen.
J. Photochem. Photobiol. A, 2005, 174, 89.
[12] F. Bosca, M. L. Marın, M. A. Miranda. Photoreactivity of the non-
steroidal anti-inflammatory 2-arylpropionic acids with photosensi-
tizing side effects. Photochem. Photobiol. 2001, 74, 637.
[13] F. Bosca, G. Carganicob, J. V. Castell, M. J. Gomez-Lechon,
D. Hernandez, D. Mauleon, L. A. Martinez, M. A. Miranda. Evaluation
of ketoprofen (R, S and R/S) phototoxicity by a battery of in vitro
assays. J. Photochem. Photobiol. B, 1995, 31, 133.
[14] K. A. K. Musa, J. M. Matxain, L. A. Eriksson. Mechanism of photo-
induced decomposition of ketoprofen. J. Med. Chem. 2007, 50,
1735.
[15] C. Prakash, C. L. Shaffer, A. Nedderman. Mass Spectrom. Rev. 2007,
26, 340.
[16] T. Kosjek. Occurrence, fate and removal of pharmaceutical residues
in water treatment. Doctoral dissertation. Ljubljana Jozˇef Stefan
International Postgraduate School, 2009.
[17] K. L. Woo, J. I. Kim. New hydrolysis method for extremely small
amount of lipids and capillary gas chromatographic analysis as
N(O)-tert.-butyldimethylsilyl fatty acid derivatives compared with
methyl ester derivatives. J. Chromatogr. A, 1999, 862, 199.
[18] M. Farre´, M. Petrovic´, D. Barcelo´. Recently developed GC/MS and
LC/MS methods for determining NSAIDs in water samples. Anal.
Bioanal. Chem. 2007, 387, 1203.
CONCLUSIONS
This study highlights the transformation of the widely used
pharmaceutical KP under exposure to UV irradiation. Using com-
plementary chromatographic and mass spectrometric techniques,
i.e. gas chromatography coupled to single quadrupole or ion trap
mass spectrometry and liquid chromatography hyphenated with
QqToF mass spectrometry, we were able to propose chemical
structures of 22 TPs of KP, 18 of which are novel. The identi-
fied compounds were used to construct a breakdown pathway
showing that, in given experimental conditions, neither UV nor
the photocatalytic treatment led to a complete mineralisation of
the parent compound KP. Instead, 20 of the TPs retained the
benzophenone moiety, which causes photo-allergic reactions. Ad-
ditionally, since benzophenone derivatives may also be formed
in the aqueous environment or during water treatment, such
symptoms can be generated after contact with environmental or
tap water containing such compounds. Another risk arises from
hydroxylated benzophenones formed during the photocatalytic
transformation of KP, since they show a potential estrogenic
effect. Hopefully, this study will increase attention into the trans-
formation of pharmaceuticals and possible risks related to their
presence in the environment, which should encourage further
research into more efficient treatment technologies to limit the
entrance of such compounds into the environmental or drinking
waters.
[19] B. W. Brooks, C. K. Chambliss, J. K. Stanley, A. Ramirez. Determina-
tion of select antidepressants in fish from an effluent dominated
stream. Environ. Toxicol. Chem. 2005, 24, 464.
Acknowledgements
[20] I. Kim, N. Yamashita, H. Tanaka. Performance of UV and UV/H2O2
processesfortheremovalofpharmaceuticalsdetectedinsecondary
effluentofasewagetreatmentplantinJapan.J.Hazard.Mater.2009,
166, 1134.
The authors kindly acknowledge the financial support given by
the Slovenian Research Agency (Program group P1-0143). For
their support and helpful advice the authors are grateful to Ms
c
wileyonlinelibrary.com/journal/jms
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2011, 46, 391–401

Application of complementary mass spectrometric techniques
[21] V. J. Pereira, K. G. Lindenb, H. S. Weinberg. Evaluation of UV
irradiation for photolytic and oxidative degradation of phar-
maceutical compounds in water. Water Res. 2007, 41, 4413.
[22] I. Kim, H. Tanaka. Photodegradation characteristics of PPCPs in
water with UV treatment. Environ. Int. 2009, 35, 793.
[23] A. C. de Groot, J. P. Nater, J. W. Weyland. Unwanted Effects of
Cosmetics and Drugs Used in Dermatology, 3rd ed. Elsevier Science
B.V.: Amsterdam, 1994.
[24] I. Kim, N. Yamashita, H. Tanaka. Photodegradation of pharmaceuti-
calsandpersonalcareproductsduringUVandUV/H2O2treatments.
Chemosphere 2009, 77, 518.
[25] T. Hayashi, Y. Okamoto, K. Ueda, N. Kojima. Formation of estrogenic
products from benzophenone after exposure to sunlight. Toxicol.
Lett. 2006, 167, 1.
c
J. Mass. Spectrom. 2011, 46, 391–401
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jms