Fully automated determination of 74 pharmaceuticals in environmental and waste waters by online solid phase extraction-liquid chromatography-electrospray-tandem mass spectrometry.

Abstract

The present work describes the development of a fully automated method, based on on-line solid-phase extraction (SPE)-liquid chromatography-electrospray-tandem mass spectrometry (LC-MS-MS), for the determination of 74 pharmaceuticals in environmental waters (superficial water and groundwater) as well as sewage waters. On-line SPE is performed by passing 2.5 mL of the water sample through a HySphere Resin GP cartridge. For unequivocal identification and confirmation two selected reaction monitoring (SRM) transitions are monitored per compound, thus four identification points are achieved. Quantification is performed by the internal standard approach, indispensable to correct the losses during the solid phase extraction, as well as the matrix effects. The main advantages of the method developed are high sensitivity (limits of detection in the low ng L(-1) range), selectivity due the use of tandem mass spectrometry and reliability due the use of 51 surrogates and minimum sample manipulation. As a part of the validation procedure, the method developed has been applied to the analysis of various environmental and sewage samples from a Spanish river and a sewage treatment plant.

Full text

Page 1

Talanta 83 (2010) 410–424
Contents lists available at ScienceDirect
Talanta
journal homepage: www.elsevier.com/locate/talanta
Fully automated determination of 74 pharmaceuticals in environmental and
waste waters by online solid phase extraction–liquid
chromatography-electrospray–tandem mass spectrometry
Rebeca López-Sernaa, Sandra Péreza, Antoni Ginebredaa, Mira Petrovi´ ca,b,∗, Damià Barcelóa,c
aDepartment of Environmental Chemistry, IDAEA-CSIC, C/Jordi Girona 18-26, 08034 Barcelona, Spain
bCatalan Institution for Research and Advanced Studies (ICREA), C/Lluís Companys, 23, 08010 Barcelona, Spain
cCatalan Institute for Water Research (ICRA), C/Emili Grahit, 101, Cientific and Technologic park of Girona University, 17003 Girona, Spain
a r t i c l e i n f o
Article history:
Received 18 June 2010
Received in revised form
10 September 2010
Accepted 25 September 2010
Available online 1 October 2010
Keywords:
Pharmaceuticals
Water analysis
Online SPE
LC–MS/MS
a b s t r a c t
The present work describes the development of a fully automated method, based on on-line solid-
phase extraction (SPE)–liquid chromatography-electrospray–tandem mass spectrometry (LC–MS–MS),
for the determination of 74 pharmaceuticals in environmental waters (superficial water and groundwa-
ter) as well as sewage waters. On-line SPE is performed by passing 2.5mL of the water sample through
a HySphere Resin GP cartridge. For unequivocal identification and confirmation two selected reaction
monitoring (SRM) transitions are monitored per compound, thus four identification points are achieved.
Quantification is performed by the internal standard approach, indispensable to correct the losses during
thesolidphaseextraction,aswellasthematrixeffects.Themainadvantagesofthemethoddevelopedare
high sensitivity (limits of detection in the low ngL−1range), selectivity due the use of tandem mass spec-
trometry and reliability due the use of 51 surrogates and minimum sample manipulation. As a part of the
validation procedure, the method developed has been applied to the analysis of various environmental
and sewage samples from a Spanish river and a sewage treatment plant.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
It is estimated that approximately 3000 different substances
areusedaspharmaceuticalingredientsworldwidetoday.However,
only a small subset of these compounds (∼150) has been investi-
gated in environmental studies. The worldwide average per capita
consumption of pharmaceuticals per year is estimated to be about
15g, but in industrialized countries the value is much higher and is
estimated to be between 50 and 150g. After administration, most
pharmaceuticals are not completely metabolized. The unmetabo-
lizedparentdrugsandsomemetabolitesaresubsequentlyexcreted
from the body via urine and faeces [1] reaching the Wastewater
Treatment Plants (WWTPs) via wastewater. Reports have shown
that many pharmaceuticals do not degrade during municipal con-
ventional wastewater treatment [2–8] being, therefore, discharged
to the receiving waters. Recent data indicate that, as much as, 80%
of the total load of pharmaceuticals entering a WWTP may be dis-
∗Correspondingauthorat:DepartmentofEnvironmentalChemistry,IDAEA-CSIC,
C/Jordi Girona 18-26, 08034 Barcelona, Spain. Tel.: +34 934006172;
fax: +34 932045904.
E-mail address: mpeqam@cid.csic.es (M. Petrovi´ c).
charged into surface water [9,10]. Disposal of unused or unwanted
medications to the toilet or household waste is another route of
their entry to the environment.
The concentrations of individual compounds in surface waters
are typically in the range of several tens to hundreds of ngL−1,
although concentrations at the ?gL−1level are also reported for
some compounds and specific sites [11]. Generally, these concen-
trations are lower than typical maximum concentrations (in the
tens of ?gL−1) reported for some industrial contaminants (e.g. sur-
factants,plasticizers),butduetotheircontinuousintroductioninto
the environment and bioactive properties, pharmaceuticals may
pose a risk to the aquatic organisms and ultimate to humans. One
of main concerns is contamination of groundwater through surface
water filtration and landfill leakage [1].
Generally, very little is known about the long-term effect and
behaviour of pharmaceutical residues in the aquatic environ-
ment [12], and in groundwater in particular [13]. In addition,
environmental risk assessment is often carried out for individual
pharmaceutical compound (active ingredients), while pharmaceu-
tical compounds are typically detected in mixtures with other
anthropogenic contaminants [11]. Studies have shown that combi-
nations of pharmaceutical compounds exert a much stronger toxic
effect that could be expected from the weak toxic effects related to
0039-9140/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.talanta.2010.09.046

Page 2

R. López-Serna et al. / Talanta 83 (2010) 410–424
411
exposure to each compound individually [14,15], i.e., the combina-
tion of drugs that share a common mechanism of action exhibits
synergistic effects [16].
Therefore,monitoringofwide-rangepharmaceuticalsinsurface
and ground waters is as a prerequisite for proper risk assess-
ment. Nowadays, a large number of analytical methodologies,
mainly using liquid chromatography–tandem mass spectrometry
(LC–MS/MS), are already available for pharmaceutical determi-
nation in both environmental and wastewaters [17–19] being
antibiotics the most widely studied group [20–22]. Majority of
the methods employ rather tedious and time-consuming sample
preparation based on off-line solid-phase extraction (SPE). How-
ever,thegrowingnumberofsamplestobeanalyzedinlaboratories
carrying out monitoring studies requires employment of high-
throughput and fully automated analytical techniques. Because of
these reasons, great effort is going into the development of fast,
cost-effective and “greener” alternative methods for environmen-
tal analysis. Over the past several years, there has been an increase
in the use of automated instruments that integrate extraction,
purification and detection step (i.e. on-line solid phase extraction
systems such as SymbiosisTMand Prospekt-2 systems manufac-
tured by Spark Holland). On line SPE followed by LC–MS/MS that
has been used to analyze trace emerging contaminants in water,
such as drugs of abuse, pesticides, and hormones [23–27]. With
respect to the analysis of pharmaceuticals in aqueous environmen-
tal samples several papers were published recently [24,27–29]. For
example[28],usedon-lineSPEintheanalysisofsixpharmaceutical
indicators in water, while [29] reported on the application on line
SPE for the analysis of macrolide antibiotics.
In this work, a reliable, fully automated method for the
determination of 74 pharmaceuticals in environmental waters
(groundwater (GW) and superficial water (SW)) and wastewa-
ter (WWTP effluent (WWE) and WWTP influent (WWI) has been
developed, validated and applied to real samples. Target com-
pounds, which are listed in Table 1, belong to different medicinal
classesandwereselectedbasedontheirhighhumanconsumption,
ecotoxicological relevance and ubiquity in the aquatic environ-
ment,accordingtotheinformationfoundinthescientificliterature
[30–42].
The objective of this work is to develop an analytical method
for simultaneous analysis of a large number of target compounds
belonging to different therapeutical classes, that will have clear
advantagesandimprovementsoverexistingmethodsintermsof(i)
minimum sample manipulation; (ii) maximum sensitivity; (ii) reli-
ability, and (iv) selectivity and thus to fulfil the stringent criteria
set by the EU regulations (EU Commission Decision 2002/657/EC)
[43].
The developed method was successfully applied to the analysis
of pharmaceutical residues in WWTP as well as river and drinking
water samples.
2. Material and methods
2.1. Chemicals
All pharmaceutical standards were of high purity grade (>90%)
and are listed in the Supplementary data 1.
Bothindividualstockstandardandisotopicallylabelledinternal
standard solutions were prepared on a weight basis in methanol
(MeOH), except fluoroquinolones which were dissolved in a
water–methanol (H2O/MeOH) mixture (1:1) containing 0.2% (v/v)
hydrochloric acid, as they are slightly soluble in pure MeOH [44].
After preparation, standards were stored at −20◦C. Special pre-
cautions have to be taken into account for tetracycline antibiotics,
which have to be stored in the dark in order to avoid their expo-
sure to the light, since it has been demonstrated that they are liable
to photodegradation [45]. Fresh stock solutions of antibiotics were
prepared monthly due to their limited stability while stock solu-
tions for the rest of substances was renewed every three months.
On the other hand, compounds with number (see Table 1) 26, 5, 10,
12 and 8, were obtained as solutions in acetonitrile (ACN), while 67
and 65 were dissolved in MeOH, at a concentration of 1mgmL−1.
A mixture of all pharmaceuticals was prepared by appropriate
dilution of individual stock solutions in MeOH/H2O (25:75, v/v).
Working standard solutions, also prepared in MeOH/H2O (25:75,
v/v) mixture, were renewed before each analytical run. Working
solutions were prepared in amber glass vials while standard mix-
tures were prepared in volumetric flasks wrapped with aluminium
foil, in order to prevent the exposure of tetracycline antibiotics to
light.Aseparatemixtureofisotopicallylabelledinternalstandards,
used for internal standard calibration, was prepared in MeOH and
further dilutions also in MeOH/H2O (25:75, v/v) mixture.
HPLC grade MeOH, ACN, water, hydrochloric acid 37% and
formic acid 98% were supplied by Merck (Darmstadt, Germany).
Ethylenediaminetetraacetic acid disodium salt dehydrate (thereon
Na2EDTA) was 99% from Sigma–Aldrich (Steinham, Germany).
Nitrogenfordrying99.995%ofpuritywasfromAirLiquide(Madrid,
Spain).
2.2. Sample pre-treatment
The method was optimized using groundwater, river water,
WWTP influent and effluent. Amber glass bottles pre-rinsed with
ultra-pure water were used for sample collection. Water samples
were filtered through 1?m fiberglass filters from Whatman (Fair-
field, Connecticut, USA) followed by 0.45?m nylon membrane
filters from Teknokroma (Barcelona, Spain). Na2EDTA 0.1% (m/v)
was added to all samples in order to form complexes with inor-
ganic elements. As it is indicated in [19], this addition improves
in a great extent the extraction efficiency of tetracycline, macrolide
andfluoroquinoloneantibiotics.Thiscouldbeexplainedbythefact
that these compounds can potentially bind residual metals present
in the sample matrix and glassware, resulting in low extraction
recoveries [46–50]. The amount of Na2EDTA added was the same
for all types of water analyzed and was considered to be sufficient
to enabled formation of complexes with inorganic compounds in
all types of matrices, even in waters with high mineral content.
Finally, 200?L of a 0.05ng?L−1standard mixture containing 37
surrogates for the analysis in positive ion (PI) mode, and 14 surro-
gates for the analysis in negative ion (NI) mode (see Table 1), were
added in every 100mL of sample for surrogate control and internal
standard calibration.
2.3. On-line trace enrichment
Preconcentration of the samples and its chromatographic sep-
aration was performed using an automated on-line SPE–LC device
SymbiosisTMPico from Spark Holland (Emmen, The Netherlands).
The base of the SymbiosisTMPico system is a high-end HPLC sys-
temwithahighperformanceinjectorthathandlessamplevolumes
from10?Lupto10mLfullyautomated.Thisequipmentalsocounts
with the AliasTMautosampler that includes positive headspace
pressure, extensive wash routines for minimal carry over and 2
injection modes, offline and online SPE. Offline mode was only
used in the optimization procedure to assess the recovery by com-
paring the peak areas obtained in the on-line analyses of spiked
waters samples with those obtained from the injection of standard
mixtures of the analytes in MeOH/H2O (25:75, v/v) at equivalent
concentrations.
AmeticulousexperimentdesignwascarriedouttooptimizeSPE
(see Table 2A and B). Three different disposable trace enrichment

Page 3

412
R. López-Serna et al. / Talanta 83 (2010) 410–424
Table 1
Target compounds organized in their therapeutical groups and their assigned surrogates.
Therapeutic groups CompoundsNumber CAS Number Corresponding surrogate
Analgesics/anti-inflammatories (12) Ketoprofen
Naproxen
Ibuprofen
Indomethacin
Diclofenac
Mefenamic acid
Acetaminophen
Salicylic acid
Propyphenazone
Phenylbutazone
Phenazone
Codeine
Clofibric acid
Bezafibrate
Fenofibrate
Gemfibrozil
Mevastatin
Pravastatin
Atorvastatin
Paroxetine
Fluoxetine
Diazepam
Lorazepam
Carbamazepine
Loratadine
Famotidine
Ranitidine
Cimetidine
Tetracycline
Doxycycline
Oxytetracycline
Chlortetracycline
Erythromycin
Azithromycin
Tilmicosin
Roxithromycin
Clarithromycin
Josamycin
Tylosin A
Spiramycin
Sulfamethoxazole
Sulfadiazine
Sulfamethazine
Danofloxacin
Enoxacin
Ofloxacin
Ciprofloxacin
Enrofloxacin
Norfloxacin
Flumequine
Trimethoprim
Nifuroxazide
Chloramphenicol
Metronidazole
Atenolol
Betaxolol
Carazolol
Pindolol
Nadolol
Timolol
Sotalol
Metoprolol
Propranolol
Salbutamol
Clenbuterol
Butalbital
Pentobarbital
Phenobarbital
Enalapril
Hydrochlorothiazide
Lisinopril
Furosemide
Glibenclamide
Tamoxifen
1
2
3
4
5
6
7
8
9
22071-15-4
22204-53-1
15687-27-1
53-86-1
15307-86-5
61-68-7
103-90-2
69-72-7
479-92-5
50-33-9
60-80-0
76-57-3
882-09-7
41859-67-0
49562-28-9
25812-30-0
73573-88-3
81093-37-0
134523-00-5
61869-08-7
54910-89-3
439-14-5
846-49-1
298-46-4
79794-75-5
76824-35-6
66357-35-5
51481-61-9
60-54-8
564-25-0
79-57-2
57-62-5
114-07-8
83905-01-5
10850-54-0
80214-83-1
81103-11-9
16846-24-5
1401-69-0
8025-81-8
723-46-6
68-35-9
57-68-1
112398-08-0
74011-58-8
82419-36-1
85721-33-1
93106-60-6
70458-96-7
42835-25-6
738-70-5
965-52-6
56-75-7
443-48-1
29122-68-7
63659-18-7
57775-29-8
13523-86-9
42200-33-9
26839-75-8
3930-20-9
37350-58-6
525-66-6
18559-94-9
37148-27-9
77-26-9
76-74-4
50-06-6
75847-73-3
58-93-5
83915-83-7
54-31-9
10238-21-8
10540-29-1
Ketoprofen-13C-d3
Naproxen-d3
Ibuprofen-d3
Indomethazine-d4
Diclofenac-d4
Mefenamic acid-d3
Acetaminophen-d4
Salicylic acid-˛-13C
Antipyrine-d3
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
Codeine-d3
Clofibric acid-d4
Bezafibrate-d4
Fenofibrate-d6
Gemfibrozil-d6
Carbamazepine-d10
Pravastatin-d3
Atorvastatin-d5
Paroxetine-d4
Fluoxetine-d5
Diazepam-d5
Lipid regulators and cholesterol lowering stain drugs (7)
Psychiatric drugs (5)
Carbamazepine-d10
Loratadine-d4
Famotidine-13C3
Ranitidine-d6
Cimetidine-d3
Demeclocycline
Histamine H2 receptor antagonists (4)
Tetracycline antibiotics (4)
Macrolide antibiotics (4)
Erythromycin-13C-d3
Azithromycin-d3
Clarithromycin-N-methyl-d3
Spiramycin-d3
Sulfamethoxazole-d4
Sulfadiazine-d4
Sulfamethazine-d4
Ofloxacin-d8
Sulfonamide antibiotics (3)
Fluoroquinolones antibiotics (7)
Ciprofloxacin-d8
Enrofloxacin-d5
Norfloxacin-d5
Flumequine-13C3
Carbamazepine-d10
Other antibiotics (4)
Ibuprofen-d3
Metronidazole-hydroxy-d2
Atenolol-d7
?-Blockers (9)
Timolol-d5
Sotalol-d6
Metoprolol-d7
Propranolol-d7
Albuterol-d3
Clenbuterol-d9
Phenobarbital-d5
?-Agonists (2)
Barbiturates (3)
Antihypertensives (3)
Enalapril-d5
Hydrochlorothiazide-d2
Atenolol-d7
Furosemide-d5
Glyburide-d3
Tamoxifen-(N,N-dimethyl-13C2)
Dirutics (1)
Antidiabetic (1)
To trat cancer (1)

Page 4

R. López-Serna et al. / Talanta 83 (2010) 410–424
413
Table 2
Experiments tested during the online SPE optimization procedure.
Type of waterType of cartridgeSample extraction volume (mL)Wash volume after extraction (mL)
(A) Online SPE experiments in
HPLC grade water
HPLC grade waterHySphere Resin GP11
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
2.5
5
PRLP-s1
2.5
5
Oasis HLB1
2.5
5
(B) Online SPE experiments in
real aqueous samples
GW HySpere Resin GP11
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
2.5
5
SW1
2.5
5
WWE1
2.5
5
WWI1
2.5
5
cartridges were evaluated for their efficiency in the on-line SPE of
the target pharmaceuticals from water: Oasis HLB (macroporous
copolymer of divinylbenzene and N-vinylpyrrolidone, 30-?m par-
ticle size) from Waters Corporation (Milford, Massachusetts, USA),
PLRP-s (cross-linked styrenedivinylbenzene polymer, 15–25-?m
particle size) from Spark Holland (Emmen, The Netherlands), and
HySphere Resin GP (polydivinyl-benzene, 5–15-?m particle size)
also from Spark Holland (Emmen, The Netherlands). In order to
evaluate which of these three cartridges yielded higher recov-
eries of target compounds, HPLC grade water was spiked with
100ngL−1of each target compound. The experiment is summa-
rized in Table 2A. After cartridge conditioning with 2mL of MeOH
and 2mL of water (flow rate 5mLmin−1), three different sample
loadingvolumes(1,2.5and5mL)weretested.Theflowthroughthe
cartridge was in all cases 1mLmin−1. Afterwards and prior to the
elution,cartridgeswererinsedwithHPLCgradewaterataflowrate
of 5mLmin−1to complete the transfer of the sample and remove
interferences such as inorganic salts. Two wash volumes (1 and
2mL) were tested in order to optimize it. Upon completion of each
SPE protocol, the trapped analytes are eluted from the cartridge to
theLCcolumn.TwoelutionmodescanbechoseninSymbisisTMPico
device: a “focusing” approach where a pre-selected quantity of sol-
vent or mixture of solvents can be chosen; or a so called “standard”
approach, where the full chromatographic gradient passes through
the SPE cartridge before being led to the LC column. Due to the
elevated number of target compounds and their different chemi-
cal properties, the last option is the more appropriate one because
of the wide range of polarity given by the mixture of the mobile
phases during the gradient. So, the chance of a successful elution
is higher. The full eluate is conducted to the LC column where the
chromatographic separation and the subsequent detection by the
mass spectrometer are carried out. In meanwhile, during the elu-
tion, a new cartridge is put in place and pre-concentration of the
next sample is simultaneously performed. This kind of configura-
tion allows short cycle times, which in our approach are 30 and
37min (the duration of the chromatographic run time) for NI and
PI mode, respectively.
Once selected the cartridge which yielded the best SPE recov-
eries, the same extraction and wash volumes trials were carried
out on real matrices (GW, SW, WWE and WWI) previously spiked
with a standard mixture of target analytes at environmentally
realistic concentrations: 20 and 100ngL−1for GW and SW, respec-
tively, and 50 and 500ngL−1for WWE and WWI, respectively (see
Table 2B). SPE recoveries as well as the method detection limits
(MDLs) achieved in each case where the parameters observed to
choose the more suitable extraction and wash volumes.

Page 5

414
R. López-Serna et al. / Talanta 83 (2010) 410–424
According to the results obtained by preliminary trials,
HySphere Resin GP cartridge, 2.5mL of sample extraction volume
and 1mL of cartridge wash after extraction, were selected for fur-
ther experiments and analysis of water samples.
2.4. LC-ESI-(QqLIT) MS/MS analysis
For chromatographic separation, an analytical column was
used: a reversed-phase Purospher Star RP-18 endcapped column
(125mm×2.0mm, particle size 5?m) from Merck (Dramstadt,
Germany) [19]. For MS/MS analyses, SymbiosisTMPico was con-
nected in series with a 4000QTRAP hybrid triple quadrupole-linear
ion trap mass spectrometer equipped with a Turbo Ion Spray
sourcefromAppliedBiosystems-Sciex(FosterCity,California,USA),
where mass spectrometry detection is carried out. 4000QTrap is
controlled by means of the Analyst 1.4.2 Software from Applied
Biosystems-Sciex (Foster City, California, USA) and a companion
software appendix for controlling the SymbiosisTMPico from Spark
Holland (Emmen, The Netherlands).
The chromatographic conditions were adapted from an analyt-
ical method previously developed and described elsewhere [19].
ForPImode,thisinvolvesaflowrateof0.3mLmin−1,andACN/0.1%
(v/v) formic acid as mobile phases. The proportion of the organic
solvent was programmed to increase from 5 to 95% in the first
20min and then to 100% in the following 2min; afterward the
column was readjusted to the initial conditions. These conditions
were held for 10min to allow re-equilibration of the column before
the next injection. The total time of chromatographic analysis (and
cartridge elution) is 37min. In this mode 57 pharmaceuticals are
analyzed.ForNImode,thisinvolvesaflowrateof0.2mLmin−1,and
ACN:MeOH (1:1, v/v)/H2O as mobile phases. The proportion of the
organic solvent was programmed to increase from 20 to 80% in the
first 15min and then to 90% in the following 2min; afterward the
column was readjusted to the initial conditions by programming
the amount of organic solvent to 20% in 3min. These conditions
were held for 10min to allow re-equilibration of the column before
the next injection. The total time of chromatographic analysis (and
the cartridge elution) is 30min. In this mode 17 compounds are
analyzed. In both modes, the injection volume was 20?L.
For quantitative analysis, the ESI-MS/MS method was modified
and adapted from [19]. For most of compounds two SRM transi-
tions between the precursor ion and two most abundant fragment
ions were monitored (full list of SRMs and instrumental conditions
are given in Supplementary data 2). Only one transition was moni-
tored for the isotopically labelled standards since they are added in
a concentration elevated enough (100ngL−1) to be reliably quan-
tified in its more intense transition. In order to obtain additional
confirmation, especially for compounds showing poor fragmenta-
tion, an Information Dependent Acquisition (IDA) experiment was
performed, with SRM as the survey scan and an Enhanced Product
Ion Scan (EPI), at three different collision energies, as dependent
scan. The obtain spectra were compared with library data based on
EPI spectra at the three collision energies used. This allows broad
accomplishment of the requirements set by the EU regulations
(EU Commission Decision 2002/657/EC) [43] related to identi-
fication and confirmation of pharmaceuticals in LC–tandem MS
analysis.
Improvements of the existing MS/MS method included: (i) a
total of 51 isotopically labelled compounds (37 in PI and 14 in NI
mode) were added before the SPE, (ii) an additional compound, the
antibiotic flumequine, was included; (iii) a second transition has
been tuned for the hydrochlorothiazide, lisinopril, acetaminophen,
pravastatin and norfloxacin. For all these ones, the selection of par-
entionsandoptimumionizationmodewereperformedbyinfusing
100?gL−1individual standard solutions in full-scan mode at dif-
ferent values of declustering potential (DP). In all cases, [M−H]−
for NI and [M+H]+for PI mode were selected. Subsequent identifi-
cation of the two most abundant fragment ions (one for surrogate
standards) and selection of the optimum collision energies (CEs)
and collision cell exit potentials (CXP) for each one was carried out
in the product ion scan mode, also infusing standard solutions of
each individual substance.
In order to obtain enough points per peak to fulfil the European
Directive and, at the same time, to get the highest sensitivity pos-
sible, the dwell time values were adjusted to 12 in PI (providing a
total scan time of 2.15s) and 31ms for NI (with a total scan time of
2.12s), with pauses between ranges of 2 (PI) and 5ms (NI).
3. Results and discussion
3.1. Solid phase extraction
Three parameters were optimized for the performance of the
method in environmental waters (groundwater and superficial
water) and sewage water (influent and effluent to a waste water
treatmentplant(WWTP)):thetypeofcartridge,thesampleextrac-
tion volume and the wash volume after extraction. SPE recoveries
and method detection limits (MDLs) were the criteria used to make
the more appropriate choice for every parameter.
Type of cartridge optimization: Table 2A shown the experimen-
tal set up. Extraction recovery of each compound was compared
among all the experiments realized for every type of cartridge. For
hydrophilic compounds, such as salbutamol, famotidine, sotalol,
ranitidine,cimetidine,HySphereResinGPcartridgesareclearlythe
best performing cartridge. As the hydrophobia increases, the dif-
ferences among the performing of the three cartridges decrease.
For the most hydrophobic compounds (betaxolol, paroxetine,
propyphenazone), Oasis HLB cartridges are the ones with better
performing,nevertheless,differenceswiththeothertwocartridges
compared are not significant, especially with HySphere Resin GP.
In general, in PI as well as NI mode, the best recoveries (near
100%)wereobtainedforHySphereResinGP,forahighernumberof
compounds. In Fig. 3 extraction recoveries of some representative
compounds are shown.
Sample volume optimization: In comparison with conventional
methods, where hundreds or even thousands of mL of sample were
needed [7,19,51–53], in the present method, much smaller sam-
ple size (units of mL) was needed since the whole eluate goes into
the analytical column. Three extraction volumes have been tested
(1mL, 2.5mL and 5mL). In general, volumes that gave best SPE
recovery were 1 and 2.5mL for PI mode, and 2.5mL for NI mode.
In Fig. 4A extraction yield of some representative compounds is
shown.
The next step included experiments with real samples in order
to check the influence of the matrix on the required sample vol-
ume, and consequently on SPE recoveries and MDLs. Less complex
matrices, such as GW and SW showed the same tendency seen
in experiments with HPLC grade water (as the hydrophobicity
of compounds increases the required volume decreased). For the
most hydrophobic compounds, 1mL was the one that gave the
best results. For samples with a complex matrix (WWE and WWI),
preference of smaller volumes (1 and 2.5mL) was even more pro-
nounced.ThatcanbeduetosignalsuppressionintheESIbecauseof
thematrix(seeSection2).Thebiggervolumeofsampleisextracted,
the higher amount of matrix is trapped in the cartridge that sub-
sequently gets to the ESI source. In Fig. 4B and C, SPE recoveries
comparingextractionvolumeswererepresentedforGWandWWE
waters, respectively. In general, 1 and 2.5mL were the volumes
that provided the best recoveries in environmental samples (GW
and SW) as well as in sewage waters (WWE and WWI) with no
big differences between them, so finally, 2.5mL was selected as the

Page 6

R. López-Serna et al. / Talanta 83 (2010) 410–424
415
Fig. 1. Chromatogram in positive ESI mode of a HPLC grade water sample spiked with a mixture of standards at 500ngL−1after being underwent to the online-SPE extraction
through GP, 2.5mL of samples 1mL of wash.
Fig.2. ChromatograminnegativeESImodeofaHPLCgradewatersamplespikedwithamixtureofstandardsat100ngL−1afterbeingunderwenttotheonline-SPEextraction
through GP, 2.5mL of samples 1mL of wash.

Page 7

416
R. López-Serna et al. / Talanta 83 (2010) 410–424
Fig. 3. Cartridge election (HPLC grade water, extraction volume 2.5mL, wash volume 1mL).
sample extraction volume, in PI mode as well as in NI one, for all
type of samples, since it yielded better MDLs.
Wash cartridge step optimization: Two cartridge wash volumes
of water were tested (1mL and 2mL). In spiked HPLC grade water
samples (experiment in Table 2A), polar compounds gave better
SPE recoveries with 1mL (Fig. 5A). This is consistent with the fact
that the solvent used for washing is water, and part of the polar
compoundswillrunwithit.Forthemthelesswashingvolumeused
the best. For the rest of compounds this parameter is not so influ-
ential. In real samples (Table 2B), the same tendency was observed
(see Fig. 5B and C). In general, washing with 1mL of water resulted
inbestrecoveryforahighernumberofcompoundsandwaschosen
for further analyses.
3.2. ESI-(QqLIT) MS/MS detection
OptimizationofMS/MSparameters:Inthepresentmethod,atotal
of 51 isotopically labelled surrogates (37 in PI and 17 in NI mode)
were included which controlled all the steps the samples under-
went, in comparison with only 10 internal standards added to the
sample after the SPE, just before the LC–MS/MS analysis in [19]
whereonlythematrixeffectcanbecorrected.Forasmallnumberof
compounds,thecorrespondingisotopicallylabelledcompoundwas
not commercially available or their price was extraordinarily ele-
vated.Anadditionalcompound,theantibioticforbiddenforselling,
flumequine is now included in the method. A second transition has
been tuned for the hydrochlorothiazide, lisinopril, acetaminophen,
pravastatinandnorfloxacinimprovingthereliabilityofthemethod
compared with [19] where only one transition was registered for
those compounds.
Thus, the resulting method includes 125 substances (74 com-
pounds and 51 surrogates), 94 of them (57 compounds and 37
surrogates) monitored in the PI mode and 31 (17 pharmaceuti-
cals and 14 surrogates) in NI mode (Figs. 1 and 2). Transitions
betweentheprecursorionandthetwomostabundantproductions
for each target analyte were recorded for all compounds with the
only exception of ibuprofen, phenobarbital, flumequine, ofloxacin,
carbamazepine and fenofibrate, for which only one product ion
could be obtained. In total, 146 transitions in positive ionization
mode (corresponding to 57 compounds and 37 surrogates) and
47 transitions in negative ionization mode (17 compounds and 14
surrogates) were recorded in one single retention time window
(Figs. 1 and 2). It should be remarked the fact of that elevated
number of transitions were recorded in one single retention time
window, without losing sensitivity, due to the setting of appro-
priate values for the dwell time and pause between mass ranges.
Adjustingthedwelltimetoanappropriatevalueisakeyparameter
to monitor large number of transitions in the same time segment
and still obtain enough points per chromatographic peak (>15),
which is very important for a precise quantification. Dwell time
in NI (31ms) was higher than in PI (12ms) because the number
of transitions was lower, so the detector can devote more time in
monitoring every transition in each cycle. Nevertheless, the ioniza-
tioninPIisbetterthaninNImode,sothesensitivityforbothmodes
is similar.
3.3. Method performance
Extraction recoveries for target compounds were determined
for all different matrices by spiking samples (n=3) at two levels
of concentrations 20ngL−1and 100ngL−1for HPLC grade water,
GW and SW and 50ngL−1and 500ngL−1for both WWI and WWE.
Thoselevelswerechosenastypicallowandhighconcentrationsfor
most of compounds in those types of waters. For each type of water
samples, recoveries were determined by comparing the concen-
trations obtained after the whole procedure, calculated by internal
standard calibration, with the initial spiking levels. As real sam-
ples (ground, surface and wastewaters) already contained target
compounds, non-spiked samples were analysed in order to deter-
mine their concentrations, which were afterwards subtracted to
the spiked samples. Due to huge quantity of data, and in order to be
easilyobserved,validationparametersarepresentedinfigures(see
Figs. 6 and 7). Complete numerical data is given in Supplementary
data 3. Two types of SPE recoveries are provided. Absolute recov-
eries, determined by comparing the peak areas obtained for spiked
water samples in the on-line SymbiosisTMPico mode of work-

Page 8

R. López-Serna et al. / Talanta 83 (2010) 410–424
417
Fig. 4. Extraction volume election (GP cartridge, wash volume 1mL).
ing, with those achieved from the injection of standards mixtures
of the analytes in MeOH/H2O (25:75, v/v) through off-line mode
at equivalent concentrations. Relative recoveries were calculated
afterwards by comparing absolute recoveries for every compound
and its respective surrogate.
Absolute recoveries achieved were in the range of 50–150% for
the 70%, 73%, 61%, 42% and 36% of target compounds in HPLC grade
water, GW, SW, WWE and WWI, respectively. See Supplementary
data3A.Thus,itwasclearthatasthematrixwasmorecomplex,the
extraction performance and/or the mass spectrometry detection
got worse. For polar compounds, as salbutamol, atenolol, cime-
tidine, famotidine low absolute SPE recovery is obtained (10.1%,
46.0%, 14.4% and 29.2%, absolute recovery in HPLC grade water,
respectively).Thepooraffinityforthecartridgeand/ortheremoval

Page 9

418
R. López-Serna et al. / Talanta 83 (2010) 410–424
Fig. 5. Cartridge wash volume election (GP cartridge, extraction volume 2.5mL).

Page 10

R. López-Serna et al. / Talanta 83 (2010) 410–424
419
A HPLC
0
2
4
6
8
10
12
14
16
18
20
<50
50-60
60-70
70-80
80-90
90-100
100-110
110-120
120-130
130-140
140-150
>150
Range of % relative recovery
Number of compounds
B GW
0
2
4
6
8
10
12
14
16
18
<50
50-60
60-70
70-80
80-90
90-100
100-110
110-120
120-130
130-140
140-150
>150
Range of % relative recoveries
Number of compounds
C SW
0
2
4
6
8
10
12
14
16
<50
50-60
60-70
70-80
80-90
90-100
100-110
110-120
120-130
130-140
140-150
>150
Range of % relative recoveries
Number of compounds
D WWE
0
5
10
15
20
25
<50
50-6060-70
70-80
80-90
90-100
100-110
110-120
120-130
130-140
140-150
>150
Range of % relative recoveries
Number of compounds
E WWI
14
Number of compounds
0
2
4
6
8
10
12
<50
50-60
60-70
70-80
80-90
90-100
100-110
110-120
120-130
130-140
140-150
>150
Range of % recoveries
Fig. 6. Relative SPE recoveries organized in ranges for HPLC grade water, GW, SW, WWE and WWI.
from it during the cartridge wash step was the reason of those low
absolute recoveries, as no matrix was involved. For chlorampheni-
col, the absolute SPE recoveries were 87.1%, 82.5%, 78.9%, 28.9%
and 16.3% for HPLC grade water, GW, SW, WWE and WWI, respec-
tively. In this case, a clear influence of matrix on the extraction
and MS/MS detection occurred. Anyhow, when the SPE recoveries
were corrected by the ones for the corresponding surrogates, the
percentages of compounds with relative SPE recovery around 100%
increased significantly. In this manner, the 92%, 81%, 81%, 68% and
72% of compounds showed a relative SPE recovery between 50 and
150%. Thereby, 111.2%, 117.4%, 97.6% and 122.5% were the rela-
tive SPE recoveries, for the same polar compounds named before,
respectively. And, 98.0%, 99.2%, 76.1%, 74.4 and 89.8% were the rel-
ative SPE recoveries for the chloramphenicol in HPLC grade water,
GW, SW, WWE and WWI, respectively. Consequently, poor per-
centages of absolutely recovery were not considered an obstacle
for their reliable determination in water, as their sensitivity was
fairly good for being corrected by the corresponding surrogate. The
overallmethodprecision,calculatedastherelativestandarddevia-
tion (RSD) was satisfactory, with RSD values ranging from 1 to 30%
for most of the compounds in all matrices.
Regarding sensitivity, Method Detection Limits (MDLs) and
Method Quantification Limits (MQLs) were determined, for envi-
ronmental and wastewater samples, as the minimum detectable
amount of analyte with a signal-to-noise of 3 and 10, respectively.
Spiked GW, SW, WWE and WWI samples (n=3) at the two level
of concentrations indicated before were used for their calculation.
As it can be seen in the Fig. 7 and Supplementary data 3B, MDLs
achieved ranged from 0.01 to 5ngL−1for most of compounds in
HPLC grade water, GW and SW, and from 0.01 to 20ngL−1for the
majority of them in wastewaters.
To ensure correct quantification, precision of the chromato-
graphic method, determined as relative standard deviation (RSD),
was determined from repeated injections (n=5) of a 100ngL−1
spiked HPLC grade water sample during the same day (repeata-
bility) and on different days (reproducibility). RSD achieved were
lower than 20 and 30% for most of compounds for intra- and inter-
day analysis, respectively.

Page 11

420
R. López-Serna et al. / Talanta 83 (2010) 410–424
A GW
0
5
10
15
20
25
30
35
40
>100
50-100
20-50
10-20
5-10
1-5
0.1-1
0.01-0.1
Ranges of MDLs (ng/L)
Number of compounds
B SW
0
5
10
15
20
25
30
35
>100
50-100
20-50
10-20
5-10
1-5
0.1-1
0.01-0.1
Ranges of MDLs (ng/L)
Number of compounds
C WWE
0
2
4
6
8
10
12
14
16
18
20
>100
50-100
20-50
10-20
5-10
1-5
0.1-1
0.01-0.1
Range of MDLs (ng/L)
Number of compounds
D WWI
0
2
4
6
8
10
12
14
16
18
>100
50-100
20-50
10-20
5-10
1-5
0.1-1
0.01-0.1
Range of MDLs (ng/L)
Number of compounds
Fig. 7. MDLs organized in ranges for GW, SW, WWE and WWI.
Regarding quantitative performance in terms of dynamic range,
linear response covered, giving good fits (r2>0.99), four and even
five orders of magnitude for the majority of compounds. Cali-
bration curves were generated in HPLC grade water and linear
regression analysis was used over the concentration range of
0.01–10,000ngL−1. Only, glibenclamide, phenyl-butazone, propy-
phenazoneanddiclofenacshowedanarrowerlinearresponsefrom
their MQLs to 500ngL−1. Thanks to that wide range of linearity, no
sample dilution is needed for highly concentrated samples before
performing the analysis in order to get a concentration inside the
lineal range. For quantification purposes, the internal standard cal-
ibration approach was used, performing thirteen-point calibration
standards daily, and the possible fluctuation in signal intensity was
checkedbyinjectingastandardsolutionattwoconcentrationlevels
after each 8–10 injections.
Influence of matrix effect in the quantitative LC–MS/MS anal-
ysis is a widely observed and studied phenomena [19,25,54]. The
ESI source is highly susceptible to other components present in
the matrix, which may result in a signal suppression or enhance-
ment leading to erroneous results. Natural organic matter, salts,
ion-pairing agents, non-target contaminants have shown to be
responsible for ion suppression. The more complex is the matrix
the stronger matrix effect will be present. Therefore, any ana-
lytical method where MS is used as detection technique should
include a matrix effect study, especially if it deals with complex
samples, as in the present case, wastewaters. If relevant ion sup-
pression (or signal enhancement) occurs, appropriate quantitative
approaches should be applied for its correction and/or minimiza-
tion in order to get an accurate quantification. The most common
approaches consist of the use of suitable calibration, such as exter-
nal calibration using matrix-matched samples, standard addition
orinternalstandardcalibrationwithstructurallysimilarunlabelled
pharmaceuticals or isotopically labelled standards, as well as dilu-
tion of sample extracts [55–58]. In order to evaluate the degree of
ion suppression or enhancement in each target compound, matrix
effects in all types of validated samples (GW, SW, WWE and WWI)
were evaluated by comparing the peak areas from the analysis of
spiked real samples (after subtracting the peak areas correspond-
ing to the native analytes present in the sample), with peak areas
from spiked HPLC grade water. In the absence of matrix effects,
analyte peak areas should be similar in both HPLC grade water
and real samples. Nevertheless, when matrix effects occurs the
signal intensity for the analytes decreases (ion suppression) or
increases (enhancement). Matrix effect was quantified comparing
the areas of compounds in spiked matrix samples with the areas
obtained in spiked solvent. The effect was expressed by percentage
of signal suppression (positive value) or enhancement (negative
values). It is clearly observed an increase in the effect as the matrix
becomes more and more complex. However, the impact of the
matrix is different for every compound. Two extreme examples
werebezafibrate,forwhichratherloweffect(−1.07%,5.72%,39.06%
and 34.51% of matrix effect in GW, SW, WWE and WWI, respec-
tively) is observed, in comparison to phenobarbital for which a
much stronger effect was evidenced with 10.90%, 24.54%, 57.48%
and 84.85% ion suppression for the same samples. It should be
noticed that ion suppression/enhancement is different for every
sample analysed even among the same type samples. Therefore, it
is of high significance to use any of the aforementioned approaches
to correct ion suppression in order to avoid inaccurate quantifica-
tion and underestimate levels of compounds when analyzing real
samples. In our study, the approach used was internal standard
calibration. In general, a corresponding isotopically labelled inter-
nal standard was selected for each compound (51 surrogates for
74 target analytes). Thus, all the therapeutic groups and within
them every family of compounds count with at least an internal
standard. The assignation of an appropriate internal standard for

Page 12

R. López-Serna et al. / Talanta 83 (2010) 410–424
421
Table 3
Average concentrations and relative standard deviation (expressed in brackets) for target pharmaceuticals in drinking water, superficial water (2 points) and effluent
wastewater in the Llobregat River basin (NE Spain).
Therapeutic groupsCompounds Concentration (ngL−1)a
Point #1 – Llobregat
River upstream to the
spill point
Point #2 – Llobregat
River downstream to
the spill point
Point #3 – drinking
water
Point #4 – WWE
tertiary treatment
Analgesics and
antiinflammatories
Ketoprofen
n.d.3.18 (1.56) n.d.57.73 (0.55)
Naproxen
Ibuprofen
Indomethacin
Diclofenac
Mefenamic acid
Acetaminophen
Salicylic acid
Propyphenazone
Phenylbutazone
Phenazone
Codeine
Clofibric acid
Bezafibrate
Fenofibrate
Gemfibrozil
Mevastatin
Pravastatin
Atorvastatin
Paroxetine
Fluoxetine
Diazepam
Lorazepam
Carbamazepine
Loratadine
81.05 (0.27)
186.68 (0.33)
16.27 (0.18)
89.53 (0.25)
n.d.
307.00 (0.59)
208.17 (0.07)
3.25 (1.11)
n.d.
5.90 (1.57)
45.85 (0.8)
8.40 (0.36)
15.89 (0.34)
23.85 (0.36)
1.90 (0.39)
n.d.
n.d.
2.99 (1.19)
n.d.
n.d.
n.d.
22.58 (0.14)
31.28 (0.29)
3.68 (1.68)
67.38 (0.27)
134.75 (0.32)
37.75 (0.29)
176.78 (0.31)
6.76 (0.30)
146.67 (0.91)
333.17 (0.61)
11.10 (0.68)
n.d.
40.27 (0.58)
109.68 (0.41)
24.25 (1.10)
67.32 (0.47)
82.08 (0.49)
2.14 (0.57)
n.d.
n.d.
2.39 (1.10)
n.d.
<LOQ
6.52 (0.64)
41.27 (0.23)
58.43 (0.30)
2.51 (0.80)
n.d.
3.71 (0.15)
n.d.
n.d.
12.82 (2.24)
n.d.
201.20 (0.23)
n.d.
n.d.
n.d.
n.d.
n.d.
0.11 (2.24)
n.d.
n.d.
n.d.
n.d.
27.60 (2.24)
n.d.
2.74 (2.24)
n.d.
n.d.
n.d.
10.48 (1.14)
72.17 (0.33)
43.57 (0.60)
93.88 (0.52)
421.50 (0.26)
17.38 (0.63)
77.83 (1.77)
674.33 (0.26)
22.55 (0.88)
n.d.
56.30 (0.28)
350.12 (0.45)
22.43 (0.16)
217.50 (0.50)
293.67 (0.64)
8.58 (0.54)
n.d.
n.d.
2.71 (1.32)
7.30 (0.91)
15.87 (0.25)
18.92 (0.23)
114.92 (0.26)
156.83 (0.24)
6.99 (1.08)
Lipid regulators
Psiquiatric drugs
Histamine H2 receptor
antagonists
Famotidine
Ranitidine
Cimetidine
Tetracycline
Doxycycline
Oxytetracycline
Chlorotetracycline
Erythromycin
Azithromycin
Tilmicosin
Roxythromycin
Clarithromycin
Josamycin
Tylosin
Spiramycin
Sulfamethoxazol
n.d.
33.87 (0.40)
17.33 (2.45)
n.d.
n.d.
n.d.
n.d.
50.38 (0.55)
14.73 (0.34)
n.d.
n.d.
42.60 (0.27)
1.82 (0.60)
n.d.
39.90 (0.44)
39.70 (0.23)
n.d.
61.23 (0.70)
n.d.
29.00 (0.91)
n.d.
n.d.
n.d.
174.73 (0.42)
71.67 (0.70)
n.d.
n.d.
88.83 (0.35)
0.81 (1.56)
n.d.
68.32 (0.44)
78.38 (0.37)
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
17.00 (0.58)
n.d.
n.d.
3.67 (0.22)
1.41 (0.77)
n.d.
20.54 (0.92)
n.d.
n.d.
197.67 (1.20)
32.05 (1.17)
171.47 (1.06)
n.d.
42.12 (0.86)
n.d.
677.00 (0.28)
1031.67 (0.53)
n.d.
3.90 (0.43)
237.83 (0.23)
3.03 (0.58)
7.17 (0.57)
141.58 (0.33)
140.48 (0.46)
Tetracycline antibiotics
Macrolide antibiotics
Sulfonamide
antibiotics
Sulfadiazine
Sulfamethazine
Danofloxacin
Enoxacin
Ofloxacin
Ciprofloxacin
Enrofloxacin
Norfloxacin
Flumequine
Trimethoprim
Nifuroxazide
Chloroamphenicol
Metronidazole
Atenolol
Betaxolol
Carazolol
Pindolol
Nadolol
Timolol
Sotalol
Metoprolol
Propranolol
Salbutamol
Clenbuterol
Butalbital
Pentobarbital
Phenobarbital
n.d.
1.68 (2.45)
n.d.
4.83 (1.56)
23.28 (0.24)
8.32 (0.79)
5.82 (0.56)
15.83 (0.31)
n.d.
16.43 (0.24)
n.d.
n.d.
n.d.
38.40 (0.43)
n.d.
n.d.
n.d.
n.d.
n.d.
15.28 (0.30)
54.47 (0.15)
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
13.40 (1.17)
112.27 (1.55)
n.d.
4.65 (1.57)
75.017 (0.48)
28.02 (0.90)
40.12 (0.84)
15.17 (0.86)
n.d.
33.53 (0.34)
n.d.
n.d.
44.88 (0.92)
63.17 (0.48)
n.d.
n.d.
n.d.
n.d.
n.d.
44.32 (0.94)
327.40 (1.91)
14.94 (0.39)
4.87 (1.70)
n.d.
n.d.
n.d.
n.d.
n.d.
4.08 (2.24)
n.d.
16.04 (0.96)
15.30 (0.73)
13.28 (0.68)
18.93 (0.79)
32.88 (0.94)
n.d.
0.51 (2.24)
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
38.48 (0.30)
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
20.38 (0.56)
373.84 (1.92)
n.d.
8.27 (1.13)
276.67 (0.46)
151.25 (1.43)
255.67 (0.35)
63.72 (1.02)
n.d.
65.92 (0.37)
n.d.
n.d.
211.83 (0.74)
117.82 (0.56)
n.d.
n.d.
n.d.
n.d.
n.d.
91.98 (0.30)
96.8 (0.25)
51.60 (0.30)
27.05 (0.30)
n.d.
n.d.
n.d.
n.d.
Fluoroquinolones
Other antibiotics
Beta blockers
Beta agonists
Barbiturates

Page 13

422
R. López-Serna et al. / Talanta 83 (2010) 410–424
Table 3 (Continued)
Therapeutic groupsCompoundsConcentration (ngL−1)a
Point #1 – Llobregat
River upstream to the
spill point
Point #2 – Llobregat
River downstream to
the spill point
Point #3 – drinking
water
Point #4 – WWE
tertiary treatment
Antihypertensives
Enalapril
Hydrochlorothiazide
Lisinopril
Furosemide
Glibenclamide
Tamoxifen
n.d.
3.26 (0.16)
n.d.
50.98 (0.50)
0.46 (1.55)
n.d.
n.d.
7.96 (0.32)
n.d.
173.72 (0.63)
1.72 (0.87)
n.d.
n.d.
1.26 (0.05)
n.d.
n.d.
3.51 (1.00)
n.d.
n.d.
33.53 (0.28)
n.d.
1120.33 (0.88)
13.13 (0.14)
n.d.
Diuretic
Antidiabetics
To treat cancer
avalues below the limit of detection and below the limit of quantification were considered 0 to calculate the mean value and the RSD.
substances without a specific one, was based on the similarity of
their chemical structures and/or their retention times. In Table 1
and Supplementary data 2, internal standards used for each sub-
stance, which in this method work as surrogates, are indicated. In
this way, the limitation in the number of internal standards pre-
sented in [19] was clearly overcome.
3.4. Monitoring results
To demonstrate the applicability of the developed method, two
river waters from the Llobregat River (NE Spain), one WWE and the
effluent of a drinking water treatment plant (DWTPE) were ana-
lyzed River samples correspond to Llobregat River (NE Spain) in
twostrategicsitesup-(point#1)anddownstream(point#2)tothe
point of discharge of treated waters form one WWTP respectively.
Point #2 coincides also with the entrance to the DWTP, which was
located a few kilometres downstream to the point #2. Point #3
corresponds to the effluent of DWTP (drinking water) and point
#4 the WWE after the tertiary treatment which was recirculated
towards the discharge point (Fig. 8). Samples from all four points
were collected twice a week during three consecutive weeks (six
samples per point) during November 2009. The object of this sam-
pling was monitoring the feasibility in the reuse of WWE after a
tertiary treatment. Despite the point #3 did not correspond to a
specific type of water validated for this method, it was considered
similar to a groundwater because of their poor matrix and the low
levels of pharmaceuticals expected.
Average concentration for the six samples per point is summa-
rized in Table 3. Levels detected were in the range of hundreds
of pgL−1to low tens of ngL−1for drinking water, and up to
low hundreds of ngL−1for surface water. Levels in wastew-
ater effluent samples were from units to hundreds of ngL−1
depending on the compound or even thousands of ?gL−1in
some cases such as the antibiotic azithromycin and the diuretic
furosemide. Data from the most frequently detected and at
higher concentration compounds is presented in bold. Antibi-
otics,analgesicsandanti-inflammatorieswerethemostubiquitous
compounds. The azithromycin and diclofenac must be remarked
among them, respectively. As expected, higher concentration were
shown at point #4 (WWE after the tertiary treatment). For the
diuretic furosemide, this concentration was especially elevated
(1120ngL−1), but after the spill into the river, the concentration
decreased in a great extent (173ngL−1). Anyhow, that concentra-
tion was still higher regarding to the one in the river upstream in
the point #1 (51.0ngL−1). This tendency was observed for most of
compounds. So it can be said that, after the discharge of effluent
Fig. 8. Sampling location.

Page 14

R. López-Serna et al. / Talanta 83 (2010) 410–424
423
the dilution effect is quite effective, but anyway, the perturbation
can be observed. However, the decrease in the levels of concentra-
tion after the discharge into the river is not the same for all the
compounds, even taking into consideration the quantity already
present in the river upstream. Thus, in addition to the dilution
which is physical phenomenon which should affect all compounds
in the same extent, other process like adsorption to sediments or
suspended solids, biodegradation or even photodegradation must
be taken into account. Anyhow, levels of pharmaceutical at the
entrance of the DWTP (point #2) were low and after the treatment
at DWTP (point #3) drinking water contained undetectable or very
low concentrations for most pharmaceuticals, with the exception
of salicylic acid that was detected at 200ngL−1.
Compoundsoccasionallydetectedordetectedatlowlevelseven
at point #4 were presented in italics. For some of them, quantifica-
tions were only possible at point #4. But after the discharge into
the river the levels decreased under the limits of quantification
or even detection. The macrolides tylosin and roxythromycin, and
the cycline oxytetracycline were some examples of that. The pres-
ence of compounds, whose quantification at point #2 was still got,
could be attributed exclusively to the discharge from the WWTP.
In those cases, purification in the DWTP treatment was responsi-
ble for reducing their levels down the limit of quantification and/or
detection, (e.g. 20–30ngL−1). 23 compounds were not detected in
any sample at any point of sampling.
4. Conclusions
The fully-automated multi-residue analytical method devel-
oped, based on on-line SPE–LC–MS/MS allowed the analysis of
74 multiple-class pharmaceuticals in two environmental types of
water as well as waste water (influent and effluent to a WWTP).
Since the SPE is carried out fully automated, on-line and simulta-
neouslytothechromatographicseparationandmassspectrometry
detection, a minimum sample manipulation is involved, and there-
fore a clear decrease in the error introduction is achieved. In fact,
filtration is the only sample pre-treatment required. In this way,
the method increases in reliability in comparison with conven-
tionaloff-linemethods.Tothisfeaturealsocontributesthefactthat
mostofcompoundscountwithaspecificisotopicallylabelledcom-
pound as surrogate (quasi isotopic dilution approach). The method
yielded detection limits in the low ngL−1range for both environ-
mentalandwastewaters,whatisessentialforpropermonitoringof
the target compounds in those type of samples. Moreover, regard-
ing to selectivity, the method fulfil the stringent criteria set by the
EU regulations (EU Commission Decision 2002/657/EC) [43]. Other
advantages of this method is its high throughput (total analysis
time is 30min in NI mode and 37min in PI mode) and the wide
linear range for most of compounds, which avoids the necessity of
diluting the samples for determining compounds present at higher
concentrations. It must also be remarked the small size of sample
needed, 2.5mL per ionization mode (total of 5mL), what relieves
the storage problems so usual in analytical laboratories. Applica-
tion of the method to the analysis of drinking, surface and effluent
wastewaters showed a widespread occurrence of pharmaceuticals
insuchmatrices,withgenerallevels,whendetected,intherangeof
units and tens of ngL−1for drinking and river water, respectively,
and tens and hundreds of ngL−1in wastewaters.
Acknowledgements
ThisworkhasbeensupportedbytheSpanishMinistryofScience
and Innovation [projects CGL2007-64551/HID, Consolider-Ingenio
2010 CSD2009-00065] and the Unity Through Knowledge Fund
(UKF), which was established by the Croatian Ministry of Science,
Education and Sports through the World Bank Loan No. 7320-
HR. Merck is acknowledged for the gift of LC columns and Spark
Holland for the gift of on-line SPE cartridges. Rebeca López Serna
acknowledges the Spanish Ministry of Education and Science for
the economical support through the FPI pre-doctoral grant.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.talanta.2010.09.046.
References
[1] Y. Zhang, S.U. Gei?en, C. Gal, Chemosphere 73 (2008) 1151.
[2] B. Halling-Sørensen, S. Nors Nielsen, P.F. Lanzky, F. Ingerslev, H.C. Holten
Lützhøft, S.E. Jørgensen, Chemosphere 36 (1998) 357.
[3] H.R. Buser, T. Poiger, M.D. Muller, Environmental Science and Technology 33
(1999) 2529.
[4] K. Kümmerer, Chemosphere 45 (2001) 957.
[5] S. Castiglioni, R. Bagnati, R. Fanelli, F. Pomati, D. Calamari, E. Zuccato, Environ-
mental Science and Technology 40 (2006) 357.
[6] J.B. Ellis, Environmental Pollution 144 (2006) 184.
[7] J.T. Yu, E.J. Bouwer, M. Coelhan, Agricultural Water Management 86 (2006) 72.
[8] J.H. Al-Rifai, C.L. Gabelish, A.I. Schäfer, Chemosphere 69 (2007) 803.
[9] T.A. Ternes, P. Kreckel, J. Mueller, Science of the Total Environment 228 (1999)
89.
[10] T.A. Ternes, M. Stumpf, J. Mueller, K. Haberer, R.D. Wilken, M. Servos, Science
of the Total Environment 228 (1999) 87.
[11] D.W. Kolpin, E.T. Furlong, M.T. Meyer, E.M. Thurman, S.D. Zaugg, L.B. Barber,
H.T. Buxton, Environmental Science and Technology 36 (2002) 1202.
[12] R.P. Schwarzenbach, B.I. Escher, K. Fenner, T.B. Hofstetter, C.A. Johnson, U. Von
Gunten, B. Wehrli, Science 313 (2006) 1072.
[13] G. Massmann, U. Dünnbier, T. Heberer, T. Taute, Chemosphere 71 (2008) 1476.
[14] M. Cleuvers, Toxicology Letters 142 (2003) 185.
[15] C. Carlsson, A.K. Johansson, G. Alvan, K. Bergman, T. Kühler, Science of the Total
Environment 364 (2006) 67.
[16] C.G.Daughton,T.A.Ternes,EnvironmentalHealthPerspectives107(1999)907.
[17] T. Ternes, ACS Division of Environmental Chemistry, Preprints 40 (2000) 98.
[18] M. Farré, M. Petrovic, D. Barceló, Analytical and Bioanalytical Chemistry 387
(2007) 1203.
[19] M. Gros, M. Petrovi´ c, D. Barceló, Analytical Chemistry 81 (2009) 898.
[20] W.M.A. Niessen, Journal of Chromatography A 812 (1998) 53.
[21] A.A.M. Stolker, U.A.T. Brinkman, Journal of Chromatography A 1067 (2005) 15.
[22] M.S. Díaz-Cruz, D. Barceló, Analytical and Bioanalytical Chemistry 386 (2006)
973.
[23] J. Feitosa-Felizzola, B. Temime, S. Chiron, Journal of Chromatography A 1164
(2007) 95.
[24] A. Garcia-Ac, P.A. Segura, C. Gagnon, S. Sauvé, Journal of Environmental Moni-
toring 11 (2009) 830.
[25] C. Postigo, M.J. Lopez De Alda, D. Barceló, Analytical Chemistry 80 (2008) 3123.
[26] M. Tagiri-Endo, S. Suzuki, T. Nakamura, T. Hatakeyama, K. Kawamukai, Analyt-
ical and Bioanalytical Chemistry 393 (2009) 1367.
[27] L. Viglino, K. Aboulfadl, A.D. Mahvelat, M. Prévost, S. Sauvé, Journal of Environ-
mental Monitoring 10 (2008) 482.
[28] R.A. Trenholm, B.J. Vanderford, S.A. Snyder, Talanta 79 (2009) 1425.
[29] J. Ding, N. Ren, L. Chen, L. Ding, Analytica Chimica Acta 634 (2009) 215.
[30] G. Huschek, P.D. Hansen, H.H. Maurer, D. Krengel, A. Kayser, Environmental
Toxicology 19 (2004) 226.
[31] S.F. Owen, E. Giltrow, D.B. Huggett, T.H. Hutchinson, J. Saye, M.J. Winter, J.P.
Sumpter, Aquatic Toxicology 82 (2007) 145.
[32] A.C. Collier, EcoHealth 4 (2007) 164.
[33] B. Chefetz, T. Mualem, J. Ben-Ari, Chemosphere 73 (2008) 1335.
[34] N. Nakada, K. Kiri, H. Shinohara, A. Harada, K. Kuroda, S. Takizawa, H. Takada,
Environmental Science and Technology 42 (2008) 6347.
[35] N. Pounds, S. Maclean, M. Webley, D. Pascoe, T. Hutchinson, Ecotoxicology and
Environmental Safety 70 (2008) 47.
[36] S. Mompelat, B. Le Bot, O. Thomas, Environment International 35 (2009) 803.
[37] A.G. Trovó, R.F.P. Nogueira, A. Agüera, A.R. Fernandez-Alba, C. Sirtori, S. Malato,
Water Research 43 (2009) 3922.
[38] N.J. Morley, Environmental Toxicology and Pharmacology 27 (2009) 161.
[39] V.L. Cunningham, S.P. Binks, M.J. Olson, Regulatory Toxicology and Pharmacol-
ogy 53 (2009) 39.
[40] L.H.M.L.M. Santos, A.N. Araújo, A. Fachini, A. Pena, C. Delerue-Matos, M.C.B.S.M.
Montenegro, Journal of Hazardous Materials 175 (2010) 45.
[41] G.M. Bruce, R.C. Pleus, S.A. Snyder, Environmental Science and Technology 44
(2010) 5619.
[42] J.P. Berninger, B.W. Brooks, Toxicology Letters 193 (2010) 69.
[43] Commission Decision (2002/657/EC) of 12 August 2002 Implementing Council
Directive 96/23/EC concerning the performance of analytical methods and the
interpretation of results, Official Journal of the European Communities L221,
Brussels, Belgium, pp. 8–36.

Page 15

424
R. López-Serna et al. / Talanta 83 (2010) 410–424
[44] E.M. Golet, A.C. Alder, A. Hartmann, T.A. Temes, W. Giger, Analytical Chemistry
73 (2001) 3632.
[45] P. Eichhorn, D.S. Aga, Analytical Chemistry 76 (2004) 6002.
[46] S. Yang, J. Cha, K. Carlson, Journal of Chromatography A 1097 (2005) 40.
[47] S. Yang, J. Cha, K. Carlson, Rapid Communications in Mass Spectrometry 18
(2004) 2131.
[48] F. Hernández, J.V. Sancho, M. Ibᘠnez, C. Guerrero, TrAC – Trends in Analytical
Chemistry 26 (2007) 466.
[49] J.Zhu,D.D.Snow,D.A.Cassada,S.J.Monson,R.F.Spalding,JournalofChromatog-
raphy A 928 (2001) 177.
[50] A.L. Batt, I.B. Bruce, D.S. Aga, Environmental Pollution 142 (2006) 295.
[51] T. Einsle, H. Paschke, K. Bruns, S. Schrader, P. Popp, M. Moeder, Journal of
Chromatography A 1124 (2006) 196.
[52] I. Ali, V.K. Gupta, P. Singh, H.V. Pant, Talanta 68 (2006) 928.
[53] W.C. Lin, H.C. Chen, W.H. Ding, Journal of Chromatography A 1065 (2005)
279.
[54] A. Jeli´ c, M. Petrovi´ c, D. Barceló, Talanta 80 (2009) 363.
[55] L. Alder, S. Lüderitz, K. Lindtner, H.J. Stan, Journal of Chromatography A 1058
(2004) 67.
[56] T. Benijts, R. Dams, W. Lambert, A. De Leenheer, Journal of Chromatography A
1029 (2004) 153.
[57] A. Kloepfer, J.B. Quintana, T. Reemtsma, Journal of Chromatography A 1067
(2005) 153.
[58] J.C. Van De Steene, K.A. Mortier, W.E. Lambert, Journal of Chromatography A
1123 (2006) 71.