Available online at www.sciencedirect.

com

Journal of Chromatography A, 1177 (2008) 150–158

Multi-residue analysis of pharmaceutical

compounds in aqueous samples夽
Anne Togola, Hélène Budzinski ∗
University of Bordeaux 1, CNRS, ISM-LPTC, UMR 5255, 351 crs de la Libération, 33405 Talence, France
Received 31 January 2007; received in revised form 21 October 2007; accepted 24 October 2007
Available online 4 December 2007

Abstract
Pharmaceutical compounds are nowadays an emerging group of organic pollutants in aquatic systems. Several methodologies have already
been published to measure these pollutants in the environment, showing the difficulties to take into account the various compounds belonging to
numerous therapeutical and chemical groups. In order to develop environmental monitoring, there is a need for a less costly and time-consuming
multi-component procedure. The work presented here deals with the development of an extraction procedure which enables the measurement of
a wide spectrum of pharmaceuticals at trace levels (ng l−1 ) with quite simple equipment (i.e. GC–MS with single quadruple as analyzer). The
analyzed compounds comprise anti-inflammatories, antidepressants and hypolipidic drugs. The reliability and sensitivity have been tested on 18
different compounds (7 basic compounds and 11 acidic drugs) extracted simultaneously and analyzed by GC–MS. The optimized procedure has
been successfully applied to the analysis of wastewaters, surface waters and drinking waters from the following areas: first the Cortiou rocky inlet,
in the Mediterranean Sea (South coast of France), highly impacted by the Marseilles wastewater treatment plant effluent and secondly the Hérault
watershed by studying drinking water, surface water and wastewater. In both cases, the level of pharmaceuticals was totally unknown. Results
obtained have demonstrated the suitability of the method for multi-residue analysis of different types of water matrices.
© 2007 Elsevier B.V. All rights reserved.

Keywords: Gas chromatography–mass spectrometry (GC–MS); Pharmaceuticals; Multi-residue analysis; Urban wastewaters; Marine waters; Drinking waters;
Surface waters; Solid-phase extraction (SPE)

1. Introduction Their monitoring is necessary to provide wider knowledge

about their occurrence in the environment, to understand their
The presence of pharmaceuticals in the environment, clas- fate, partition and organism exposure levels [10].
sified as the so-called emerging contaminants, has raised great The quantification of pharmaceuticals in human biological
concern among the scientific community during the last few matrix such as blood, plasma or urine [11] has been devel-
years. Unfortunately, as a result of their growing use, these com- oped for a long time. But similar developments concerning
pounds have been found in aquatic systems, in sewage treatment pharmaceuticals in natural waters present more difficulties:
plant effluents [1,2] as well as in surface waters [3] even detected these compounds are present at low levels and as very com-
in drinking waters [4]. plex mixtures of dozens of different molecules. Simultaneous
Their ubiquity in the environment has prompted researchers extraction of different therapeutic groups is particularly focused
to identify the effects that these compounds could have on on antibiotics and steroids with HPLC–MS–MS or GC–MS
non-target species [5,6] and to develop chronic exposure risk analyses [12].
assessment on aquatic organisms [7] as well as on human beings Some studies have already presented very efficient analytical
[8,9]. procedures designed for specific pharmaceutical classes [13].
Analyzing simultaneously a wide spectrum of pharmaceuticals
夽 Presented at the 1st Thematic Workshop on Chemical Analysis of Emerging
with different physico-chemical properties is rather difficult and
requires a compromise, sometimes not resulting in the obten-
Pollutants, Mao, Menorca, Spain, 27–28 November 2006.
∗ Corresponding author. Tel.: +33 5 40006998; fax: +33 5 40006998. tion of the best conditions for all analytes. Some procedures
E-mail address: [email protected] (H. Budzinski). have allowed to measure pharmaceuticals at trace levels but

0021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.chroma.2007.10.105
 A. Togola, H. Budzinski / J. Chromatogr. A 1177 (2008) 150–158 151

with very tedious extraction procedures [14] and large extrac- Raw water was filtered on GFF filters to separate dissolved
tion volumes, reducing the number [8] of analyzed samples phase and particles. For natural waters, 1 l was filtered whereas
which makes the environmental screening difficult to achieve for Wastewater Treatment Plant Effluent (WWTP effluent),
[15]. Nowadays, multi-residue analytical methods are required 500 ml was used for each extraction. Sample pH was adjusted
in order to provide wider knowledge about the presence of phar- prior to extraction at a value of 2 with HCl (3.5 M). Moreover,
maceuticals in the environment. The work presented here deals internal standards (25–50 ␮l of a methanolic mixture containing
with the development of an extraction procedure that enables 1 ␮g g−1 of each standard depending on the type of waters: the
the measurement of a wide spectrum of pharmaceuticals at trace least for surface waters, the most for wastewaters) were added
level (ng l−1 ), including anti-inflammatory d rugs, antidepres- to the samples.
sants, hypolipidic drugs, etc. The reliability and sensitivity have Before sample loading, SPE cartridges were conditioned with
been tested on 18 different compounds (7 basic compounds and 3 ml of ethyl acetate and 3 ml of Milli-Q-water at adjusted pH
11 acidic drugs) extracted simultaneously by off-line SPE and 2. Water was percolated under vacuum onto the cartridges at a
analyzed with GC–MS. In order to validate the applicability of flow rate of 12–15 ml min−1 and afterwards dried for 1 h under
the method, it has been applied to the analysis of wastewaters, vacuum. After elution with three successive solvents, 3 ml of
surface waters and drinking waters from the Hérault watershed ethyl acetate, 3 ml of ethyl acetate/acetone (50/50; v/v) and 3 ml
and the Cortiou rocky inlet. First, the Hérault watershed has of ethyl acetate/acetone/ammonium hydroxide (48/48/2; v/v/v),
been studied including wastewaters as well as surface waters respectively, the samples were completely evaporated under
and spring waters dedicated to human consumption, with low nitrogen and transferred into GC injection vials in 50–100 ␮l
pharmaceutical levels. On the other hand, seawaters, highly con- of ethyl acetate. For recovery control, pyrene was added to
taminated by the Marseilles wastewater treatment plant effluent the final extracts for basic compounds before GC–MS anal-
[16] and with a high organic matter content, have been monitored ysis and 1-hydroxypyrene was added to the final extracts for
in the Cortiou rocky inlet, in the Mediterranean Sea (Marseilles acidic compounds before the derivatization step, consisting
area, South coast of France). in adding 30 ␮l of MSTFA before incubation at 65 ◦ C for
35 min.
2. Materials and methods All additions of matrix, standards, solvents or reagents were
gravimetrically controlled. Blanks were performed for each
2.1. Chemicals and reagents batch experiment in order to prevent any contamination. No
compounds have been found in blank samples.
Pharmaceutical products (presented in Table 1) as well
as pyrene and 1-hydroxypyrene used as recovery determi- 2.3. GC–MS analysis
nation standards were purchased from Sigma–Aldrich (St.
Quentin Fallavier, France; purity >98%). Deuterated products GC–MS analyses were carried out using an HP 6890
(diazepam d5, amitryptiline d6 and nordiazepam d5) were pur- gas chromatograph from Agilent Technologies (Palo
chased from Euriso-Top (St. Aubin, France, purity >98%). Alto, CA, USA). The capillary column was an HP5/MS
Acetone, ethyl acetate and methanol (HPLC reagent grade, (30 m × 0.25 mm × 0.25 ␮m film thickness; phase: 5%
Scharlau) were purchased from ICS (Belin-Beliet, France). diphenyl, 95% dimethylsiloxane) from Bios Analytique
Hydrochloric acid 37% (reagent grade) and phosphoric acid (L’Union, France). Samples were injected (1 ␮l) into the GC
85% (reagent grade) were obtained from Atlantic Labo (Eysines, in splitless mode at 250 ◦ C using an HP 6890 series injector.
France). Ultrapure water was obtained with a Milli-Q system The carrier gas was ultrapure helium (99.99990%, Linde
(Millipore, Molsheim, France). Sixty-milligram Oasis MCX Gas, Bassens, France) set at constant flow mode (1.3 ml/min).
cartridges were purchased from Waters (St. Quentin en Yvelines, For GC separation, the temperature program started at 70 ◦ C
France). (held for 2 min), set at 10 ◦ C/min to 250 ◦ C and then was held
MSTFA (N-methyl-N-(trimethylsylil)trifluoroacetamide, isothermally at 250 ◦ C for 5 min. The gas chromatograph was
purity >97% from Acros Organics (Noisy-le-Grand, France)) coupled to an HP 5973N mass selective detector (LMSD,
was used as the derivatizing reagent for GC–MS analyses. Agilent Technologies, Palo Alto, CA, USA), operated under
Whatman GFF glass fibre filters (pore size 0.7 ␮m) were electronic impact (EI) mode at 70 eV using scan mode (from 50
purchased from VWR International (Fontenay-sous-Bois, to 600 amu, 2.69 scan s−1 ) and single ion monitoring mode at
France) and Atlantic Labo (Eysines, France). 1.67 scan s−1 (dwell time 70 ms). The transfer line, source and
quadruple temperatures were 280, 230 and 150 ◦ C, respectively.
2.2. Sample pretreatment and solid-phase extraction Each compound has been first characterized individually in
optimization scan mode in order to identify the main ions (m/z ratio) constitut-
ing the mass spectrum and to choose the ions for quantification
Water samples have been collected in amber glass bottles, and for confirmation (Table 1). For acidic compounds, detection
previously detergent washed, acid rinsed and heated at 450 ◦ C has also been investigated after the derivatization step with N-
for 6 h. The samples were filtered on GFF fibre filters imme- methyl-N-(trimethylsylil)trifluoroacetamide (MSTFA). For this
diately after collection and pharmaceutical extraction was done purpose, the solution has been kept for 35 min in an oven at
during the day of sampling. 65 ◦ C after adding 30 ␮l of MSTFA.
152 A. Togola, H. Budzinski / J. Chromatogr. A 1177 (2008) 150–158

Table 1
Studied compounds with various parameters (structure, molecular weight, m/z ratio (quantification and confirmation), internal standards and linearity)
Compound Retention Therapeutic group Chemical structure MW m/z ratio Internal R2
time (min) (g mol−1 ) standards

Aspirin (ASP) 14.0 180 195 [MTMS –COO–CH3 ] Diazepam d5 0.9958

Non-steroidal
anti-inflammatory
drugs (NSAID)
Ibuprofen (IBU) 16.4 206 160 [MTMS –COO–TMS] Diazepam d5 0.9953
(confirmation: 263)

Ketoprofen (KETO) 16.8 254 282 [Mdi-TMS –COO–TMS] Diazepam d5 0.9963

(confirmation: 311)

Naproxen (NAP) 22.9 230 185 [MTMS –COO–TMS] Diazepam d5 0.9934

(confirmation: 302)

Paracetamol (PARA) 23.5 151 206 [MTMS –NH2 ] (confirmation: Diazepam d5 0.9916
295)

Gemfibrozil (GEMF) 24.9 Lipid regulator 250 201 [MTMS –(CH3 )2 –C6 H3 –O] Diazepam d5 0.9996

Salbutamol (SALB) 25.7 Bronchodilator 239 369 Diazepam d5 0.9863

[Mtri-TMS –CH2 –NHC(CH3 )3 ]
(confirmation: 86)

Clenbuterol (CLENB) 26.2 Bronchodilator 276 335 [Mdi-TMS –CH2 (CH3 )3 ] Diazepam d5 0.9908
(confirmation: 86)

Terbutalin (TERB) 27.9 Bronchodilator 225 356 Diazepam d5 0.9854

[Mtri-TMS –CH2 –NHC(CH3 )3 ]
(confirmation: 86)

Diclofenac (DICLO) 29.4 NSAID 295 214 [MTMS –COO–TMS–Cl] Diazepam d5 0.9987
(confirmation: 367)

Diazepam (DZP) 32.4 Antidepressant 284 256 [M–CH2 N] (confirmation: Diazepam d5 0.9989
221)

Caffeine (CAF) 21.8 Stimulant 194 194 [M+ ] (confirmation: 109) Diazepam d5 0.9943

Carbamazepine (CBZ) 28.9 Sedative 236 193 [M–NHCO] (confirmation: Diazepam d5 0.9949
165)

Amitryptiline (AMI) 29.0 277 58 fragment [–CH2 –N(CH3 )2 ] Amitryptiline 0.9503

Antidepressant (confirmation: 202) d6

Imipramine (IMIP) 29.3 280 58 fragment [M–HN(CH3 )2 ], Amitryptiline 0.9558

[–CH2 –N(CH3 )2 ] (confirmation: d6
234)
 A. Togola, H. Budzinski / J. Chromatogr. A 1177 (2008) 150–158 153

Table 1 (Continued )
Compound Retention Therapeutic group Chemical structure MW m/z ratio Internal R2
time (min) (g mol−1 ) standards

Doxepine (DOX) 31.1 279 58 fragment [–CH2 –N(CH3 )2 ] Amitryptiline 0.9677

(confirmation: 280) d6

Nordiazepam (NDZP) 34.2 Benzodiazepines 270 242 [M–HCO] (confirmation: Nordiazepam 0.9858
active metabolite 270) d5

2.4. Method validation acetate/acetone (50/50; v/v)) already developed and validated
[17].
To allow the quantification of the different compounds, the
use of several internal standards has also been investigated. The 2.6. Environmental analysis
accuracy of the quantification method and the recovery of inter-
nal standard extraction have been evaluated with direct injections A sampling campaign was conducted in April 2004 in the
of solutions in ethyl acetate and on spiked samples. Hérault watershed. This area has been chosen in order to com-
pare with other works concerning the potential contamination of
2.5. Analytical development the waters of this area by anthropic effluents characterized for
some of them by the occurrence of gadolinium, a pharmaceu-
After preparing individual mother standard solutions in tical residue of Magnetic Resonance Imaging (MRI) analysis
methanol (10 ␮g g−1 ) stored at 4 ◦ C, diluted mixtures have been [18]. Three kinds of water have been sampled: three sewage
prepared in ethyl acetate (1 ␮g g−1 for each compound). treatment plant wastewater samples, six surface water samples
Elution on MCX cartridges has been tested with three and six tap water samples. All sampling stations are presented
successive elutions, respectively, 3 ml of ethyl acetate, 3 ml in Fig. 1. Spring water samples have been collected just before
of ethyl acetate/acetone (50/50; v/v) and 3 ml of ethyl the chlorination process. The limits of detection of water have
acetate/acetone/ammonium hydroxide (48/48/2; v/v/v). This been calculated on the three kinds of water.
experiment has been compared with optimized C18 and The Cortiou rocky inlet is located near Marseilles (France),
Oasis HLB extraction procedures (elution with 9 ml of ethyl in the Mediterranean Sea. The effluent of the Marseilles city

Fig. 1. Map of the Herault watershed (a) with sampling sites and zoom on the Lergue river for the study of surface water stations (b).
154 A. Togola, H. Budzinski / J. Chromatogr. A 1177 (2008) 150–158

Fig. 2. Map of the rocky Cortiou inlet with sampling stations.

wastewater treatment plant (1,300,000 population equivalents) 3.2. GC–MS analysis

which undergoes no biological treatment pollutes this area. Sev-
eral samples have been collected in the effluent plume (Fig. 2) GC–MS tests have been done in SCAN mode in order
to investigate the influence area of the effluent. to choose the quantification ions. The compounds have been
separated into two groups: basic and acidic compounds. The
3. Results and discussion basic compound group (caffeine, carbamazepine, amitryptiline,
imipramine, doxepine, nordiazepam and diazepam) is directly
3.1. Sample pretreatment and solid-phase extraction injected. Compounds with one or more acidic functional groups
optimization require a derivatization step in order to improve chromato-
graphic performances.
3.1.1. Recoveries for MCX extraction procedures Among the tested derivatizing products and derivatization
Combined extraction using MCX cartridges has given high conditions [19], it has been chosen to use 30 ␮l of MSTFA per
recovery values, presented in Table 2. The obtained values
range between 54 ± 12% (for terbutaline) and 120 ± 6% (car- Table 2
bamazepine), with an average at 88%. Comparison between different solid-phase extraction sorbents in the case of
spiked water (500 ng l−1 for each compound) (n = 6)
3.1.2. Comparison with C18 and HLB extraction Compounds HLB extraction C18 extraction MCX extraction
The previous procedure used previously [17] allowed only
DICLO 91 ± 10 – 105 ± 5
to extract separately the acidic and basic groups, by using two KETO 89 ± 14 – 106 ± 6
kinds of cartridges (C18 for the basic group and HLB for the NAP 79 ± 12 – 90 ± 3
acidic one). The new methodology is also more interesting from ASP 84 ± 18 – 71 ± 12
a practical point of view. Furthermore, basic compound recovery GEMF 78 ± 11 – 81 ± 1
IBU 84 ± 11 – 80 ± 3
rates have been widely enhanced, more particularly as regards
CAF – 70 ± 6 68 ± 11
doxepine and imipramine that were not extracted at all with DOX – 2± 0 98 ± 9
the previous procedure using C18 cartridges. Table 2 shows CBZ – 99 ± 11 120 ± 6
recovery rates depending on solvents and sorbents. Fig. 5 shows IMIP – 1± 1 95 ± 5
compared results between two similar wastewater samples com- AMI – – 95 ± 9
PARA – – 76 ± 4
ing from a WWTP effluent analyzed by the two methods: the
DZP – – 87 ± 3
procedure using C18 and HLB cartridges and the combined pro- SALB – – 62 ± 14
cedure with MCX cartridges. Results show a good similarity CLEN – – 90 ± 17
between both measured concentrations, which allows the use TERB – – 54 ± 12
of the new procedure, easier to manage, more efficient for some DZP – – 104 ± 2
NDZP – – 101 ± 4
compounds and less costly and time consuming than the previous
one. Abbreviations are given in Table 1.
 A. Togola, H. Budzinski / J. Chromatogr. A 1177 (2008) 150–158 155

Fig. 3. Quantification of pharmaceutical compounds for standard solutions, using internal standards (n = 3). Abbreviations are given in Table 1.

sample and to keep the samples for 35 min at 65 ◦ C before injec- Concerning acidic compounds, the rates of quantification for
tion, considering the abundance of ions, the repeatability and standard solutions obtained using diazepam d5 are between 74
robustness of this procedure. and 115% with RSDs between 0 and 20% (Fig. 3).
Therefore, each sample has first been analyzed by GC–MS Caffeine C13 and paracetamol d4 have not been selected as
before the derivatization step, then reinjected after derivatization internal standards: even if the quantification results are correct,
using a second GC–MS method. Both chromatographic meth- the successive steps, especially the evaporation one, lead to
ods are similar; diazepam that presents the same main fragment significant and not reproducible losses. For diazepam d5, amit-
and responses using both methods are used to compare the two ryptiline d6 and nordiazepam d5, recoveries are the same as
injections. those of non-labeled compounds.
Ions (m/z ratios) chosen for quantification, molecular weights Some quantification problems on some compounds, as well
and linearity coefficients (between 2 pg and 5 ng injected) as high RSDs going up to 20% for salbutamol, are mainly due
obtained for GC–MS analyses are presented in Table 1, with to derivatization problems. Indeed for clenbuterol, salbutamol,
the chemical structures of studied pharmaceuticals. For acidic and terbutaline different derivatives are formed [17] but not in
compounds, in the case of multiple possibilities of derivatives, reproducible conditions.
the most abundant and stable derivative has been selected. In the case of basic compounds, reproducible optimum quan-
In the case of carbamazepine, previous studies have shown tification results have been obtained (see Fig. 3). Different
that a thermal degradation phenomenon could happen, occur- deuterated compounds have been selected depending on best
ring in the injection system, which could affect carbamazepine recoveries, that is, nordiazepam d5 for nordiazepam quantifi-
quantification by forming stilbene [20]. These degradation phe- cation (100 ± 5%), diazepam d5 for diazepam, carbamazepine
nomena have been checked in our study and were shown to be and caffeine (respectively, 100 ± 4, 99 ± 4 and 75 ± 3%), and
very small and constant. Degradation rate was constant for all amitryptiline d6 for amitryptiline, imipramine and doxepine
analyses with a variability below 10% and the use of internal (respectively, 88 ± 4, 93 ± 3 and 87 ± 2%).
standards for quantification that are not influenced by thermal Pyrene (for basic compounds) and 1-hydroxypyrene (for
degradation allows to monitor possible degradation (or not) of acidic ones) are used as syringe standards (added just before
carbamazepine by following evolution of response coefficient the injection so not depending on sample preparation) in order
between diazepam d5 (as internal standard) and carbamazepine to quantify internal standards (diazepam d5, amitryptiline d6 and
in standard solutions to control the stability of the system. nordiazepam d5) and calculate recoveries (Fig. 4). The results
Diazepam d5 is proved to be very stable so if the response fac- that have been obtained (between 86 ± 5 and 110 ± 5%) make
tors of carbamazepine in comparison with diazepam d5 do not it possible to use pyrene and 1-hydroxypyrene as standards for
vary, it shows that under our GC conditions the potential thermal controls of pharmaceutical extraction rates.
degradation of carbamazepine is not significant. The different kinds of water studied present various organic
matter contents and matrix complexity. The limits of detection
3.3. Method validation

All available deuterated standards have been used for quan-

tification. Paracetamol d4 and diazepam d5 have been tested
for acidic compound quantification and diazepam d5, amitryp-
tiline d6, nordiazepam d5 and caffeine C13 for the basic group.
Pyrene and 1-hydroxypyrene have been used for quantification
of deuterated compounds (“syringe” standards). Use of “syringe
standards” added after extraction step and just before GC–MS
analysis allows to check internal standard recovery for each Fig. 4. Quantification of internal standards by “syringe” standards (n = 3).
sample. Abbreviations are given in Table 1.
156 A. Togola, H. Budzinski / J. Chromatogr. A 1177 (2008) 150–158

Table 3 Table 4
Limits of detection obtained for natural waters (expressed in ng l−1 , for 1 l Minimal and maximal values measured in the case of the Hérault watershed
extracted for tap and surface and saline waters, 500 ml for wastewater effluent) samples, expressed in ng l−1 (nd: not detected)
Tap water Surface Marine Wastewater WWTP surface water drinking
water water effluent effluent (3) (6) water (6)

Amitryptiline 0.7 2.2 2.6 6.8 Amitryptiline nd–6.0 nd nd–1.4

Aspirin 0.2 2.1 2.1 15.0 Aspirin 23.5–51.5 nd nd
Caffeine 1.5 2.5 2.3 28.6 Caffeine 255.1–2212.7 13.0–107.2 nd–22.9
Carbamazepine 0.8 1.4 2.2 22.3 Carbamazepine 157.3–293.4 nd–56.3 nd–43.2
Clenbuterol 0.6 0.3 1.2 4.0 Clenbuterol nd–5.9 nd nd
Diazepam 0.4 1.4 1.9 13.8 Diazepam nd nd nd
Nordiazepam 0.4 1.4 1.9 13.8 Diclofenac 210.7–486.4 1.36–33.2 nd–2.5
Diclofenac 0.9 0.7 2.6 9.0 Doxepine nd nd nd
Doxepine 0.7 2.1 2.4 16.6 Gemfibrozil 13.3–17.2 nd–2.3 nd
Gemfibrozil 0.1 0.3 1.2 3.2 Ibuprofen 17.7–219.0 nd–4.5 nd–0.6
Ibuprofen 0.1 0.1 1.7 4.8 Imipramine nd nd nd
Imipramine 0.7 1.2 1.6 12.7 Ketoprofen 21.8–1080.6 nd–14.5 nd–3.0
Ketoprofen 0.3 0.7 1.8 11.6 Naproxen 42.1–289.1 nd–9.1 nd–0.2
Naproxen 0.1 1.0 2.1 6.2 Nordiazepam nd–8.3 nd–2.4 nd
Paracetamol 108.1–11308.9 10.6–72.3 nd–210.1
Salbutamol nd nd nd
Terbutaline nd–4.1 nd nd
are very different depending on the origin and kind of water.
They vary between 0.1 and 1.5 ng l−1 for tap water, between 0.1
and 2.5 ng l−1 for surface water and between 3.2 and 28 ng l−1
for wastewater, with variability depending on compounds as pre- GC–MS analyses [20] or even LC–MS–MS [21]. This is partly
sented in Table 3. These low detection levels allow to quantify due to high ratio of reconcentration which is possible for internal
pharmaceuticals in the two studied environments. For the three standards are used at two levels (for quantification and for control
different types of water (tap water, surface water and waster (syringe standard)) and also because gravimetric manipula-
water), spiked samples have been managed (including all com- tions are preferred to volumetric ones. Currently, LC–MS–MS
pounds) and have shown the good performances of the protocol: and GC–MS–MS analyses give lower detection limits (below
recoveries vary between 70 and 110% depending on the com- 1 ng l−1 ) [22] but need more method optimization and are more
pounds and on the matrix. complex to implement. Especially for LC–MS–MS analyses,
Repeatability and reproducibility have been tested in order some difficulties linked to interfering compounds can occur,
to assess the extraction procedure reliability. Repeatability mea- such as matricial interferences and signal suppression [23].
sured on six independent replicates, fluctuates between 3 (for Comparing with other studies which try to develop multi-
aspirin) and 26% (for clenbuterol). Reproducibility fluctuates residue analytical protocols [12], this optimized procedure has
between 2 and 11% on triplicates for basic compounds and been applied to several kinds of waters showing its multi-matrix
between 1 and 17% for the acidic group, with higher value for property. Using an easy to use GC–MS analyzer allows less
clenbuterol (17%). In the case of the spiked samples the RSD costly and easier environmental monitoring than HPLC-MS–MS
were below 20% for all the compounds (below 15% for basic or GC–MS–MS systems.
ones). Concerning the Hérault watershed, results have shown a real
In both cases, the compounds for which the protocol is the contamination of the WWTP effluents but also of the surface
less reliable are paracetamol, clenbuterol, terbutaline and salbu- waters and tap waters (Table 4). Close connection between
tamol. This weakness is partly due to the GC–MS step, due WWTP and the tap water collecting station can explain this
to some difficulties during the derivatization step as explained occurrence [24], already demonstrated in other publications
previously, especially highlighted in natural samples, due to [25,26]. The use of a single extraction procedure for all samples,
the occurrence of numerous interfering compounds that could sensitive and robust whatever the typed sample, irrespective of
disturb derivatization. concentration levels and water sample physico-chemical prop-
This procedure is robust and repeatable, also applicable to erties is an important advance in environmental monitoring.
environmental samples and allows single determination for each This can allow to monitor environmental areas, without previous
studied sample, which is important considering environmental knowledge of contamination levels or other characteristics, as is
monitoring at large scale. needed for first environmental screening.
Concerning the Cortiou rocky inlet (Fig. 5), the measured
3.4. Environmental implementation values are interesting for several reasons. Marseilles wastewater
treatment plant has an important capacity (85,500,000 m3 treated
The minimal and maximal measured values in the case of the and 1,300,000 population equivalents), without secondary treat-
Hérault watershed are presented in Table 4 showing the sensi- ment. In France, it is the most important station of this kind,
tivity of the developed procedure. Compared with other studies allowing evaluation of WWTP effluent in the worst conditions.
these values are below detection limits generally obtained using This station should be renovated in 2007; the results obtained
 A. Togola, H. Budzinski / J. Chromatogr. A 1177 (2008) 150–158 157

Fig. 5. Pharmaceutical concentrations measured in the Cortiou rocky inlet. Abbreviations are given in Table 1.

here allow future comparison studies and monitoring of reme- detection limits (ng l−1 ) obtained for this optimized procedure
diation processes. make it possible to quantify those compounds in little contam-
Some compounds, rarely found in surface water as well as inated environments, such as drinking water, seawater or other
in wastewater effluents have been measured around 10 ng l−1 , environments. Results have highlighted the contamination of
especially for amitryptiline, diazepam and nordiazepam, as two sensitive systems, the Cortiou rocky inlet and the Hérault
antidepressants. The high concentration level of treated wastew- watershed. These results are among the first showing those high
aters is explained by the relationship between pharmaceutical concentration levels in seawater and drinking water. Compared
consumption and occurrence in effluent. Considering 1,300,000 with other studies, this procedure is faster and easier to manage,
inhabitants, the number of antidepressant consumers is high, considering the simultaneous extraction of all pharmaceutical
increasing the content of those drugs in effluent and then compounds studied. This protocol stands comparison with other
allowing their detection. For the same reasons, very high extraction processes using multiple extraction steps [13,21,31]
amounts of non-prescribed drugs have been measured, such as or using complex and costly analysis apparatus [12]. By using
aspirin (8 ␮g l−1 ), caffeine (8 ␮g l−1 ), but above all paraceta- this process, impacted areas have been discovered, presenting
mol (200 ␮g l−1 ). The degradation of these three compounds is potentially human health risks, if we consider concentration
highly related to biological treatment efficiency in wastewater measured in drinking water or environmental risks considering
treatment plants [27–29]. In the absence of biological treat- the Cortiou rocky inlet situation.
ment in the Marseilles WWTP, high concentrations of those Within the limits of current knowledge, risk assessment
compounds are thus measured in the effluent. does not indicate toxic risk, especially concerning human expo-
Concerning the plume of dilution of the effluent and its impact sure [32,33]. As regards aquatic organism exposure, research
on the Cortiou rocky inlet, this preliminary study has focused advances need to be more important before a real risk assess-
on the area close to the emissary, more or less 500 m from the ment, particularly if we consider long time exposure at
emissary for the remotest stations (stations 5 and 6). Currents environmental concentration levels.
influence the plume of dilution and carry effluents towards one
side more than towards the other. This very important effluent,
introduced into a semi-open aquatic system, poorly submitted Acknowledgments
to dilution phenomena can present a high environmental risk
for aquatic organisms living in the area. Considering the first The authors acknowledge financial support from the “Région
results presenting the occurrence of antidepressants in fish living Aquitaine”, the Seine Aval, the ORQUE (Observatoire Régional
in an effluent-dominated stream [30], studies focusing on fish de la Qualité de l’Environnement), the European SWIFT-WFD
exposure and toxicological impact related to this exposure need (contract no. SSPI-CT-2003-502492) and the “GIS ECOBAG”
to be undertaken. programs for research funding, IFREMER (French Research
Institute for Exploitation of the Sea) and Hydrosciences lab-
4. Conclusions oratory (UMR 5469, Montpellier) for technical support and
sampling campaigns. The authors wish to thank the NFS
This procedure, using MCX solid-phase extraction is at the (National Funding for Science) for providing the PhD grant of
same time quick and semi-automatic, allowing numerous sample A. Togola.
processing, reliable, robust and reproducible for drinking water,
surface water as well as wastewaters. A protocol allowing to References
measure 18 different compounds, belonging to 5 therapeutic
groups and having important differences in chemical structure [1] M. Farré, I. Ferrer, A. Ginebreda, M. Figueras, O. Lourdes, T. lluis, V.
has been developed and has been proven to be very interesting Manel, B. Damia, J. Chromatogr. A 938 (2001) 187.
for environmental screening. [2] T. Heberer, J. Hydrol. 266 (2002) 175.
[3] S. Öllers, H. Singer, P. Fässler, S. Müller, J. Chromatogr. A 911 (2001) 225.
The obtained results have shown that the developed tool [4] A. Tauxe-Wuersch, L.F. De Alencastro, D. Grandjean, J. Tarradellas, Water
ensures to follow aquatic environment contamination whatever Res. 39 (2005) 1761.
the water physico-chemical characteristics are. The low levels of [5] M. Cleuvers, Toxicol. Lett. 142 (2003) 185.
158 A. Togola, H. Budzinski / J. Chromatogr. A 1177 (2008) 150–158

[6] N. Laville, S. Ait-Aissa, E. Gomez, C. Casellas, J.M. Porcher, Toxicology [20] L. Damasceno, R. Ventura, J. Ortuno, J. Segura, J. Mass Spectrom. 35
196 (2004) 41. (2000) 1285.
[7] H. Sanderson, D.J. Johnson, T. Reitsma, R.A. Brain, C.J. Wilson, K.R. [21] T. Kosjek, E. Heath, A. Krbavcic, Environ. Int. 31 (2005) 679.
Solomon, Regul. Toxicol. Pharm. 39 (2004) 158. [22] T.A. Ternes, M. Bonerz, T. Schmidt, J. Chromatogr. A 938 (2001) 175.
[8] O.A.H. Jones, N. Voulvoulis, J.N. Lester, Water Res. 36 (2002) 5013. [23] W. Ahrer, E. Scherwenk, W. Buchberger, J. Chromatogr. A 910 (2001) 69.
[9] L.J. Schulman, E.V. Sargent, B.D. Naumann, E.C. Faria, D.G. Dolan, J.P. [24] C. Metcalfe, X.S. Miao, B.G. Koenig, J. Struger, Environ. Toxicol. Chem.
Wargo, Hum. Ecol. Risk Assess. 8 (2002) 657. 22 (2003) 2881.
[10] B. Halling-Sorensen, S.N. Nielsen, P.F. Lanzky, F. Ingerslev, H.C. Holten [25] M. Rabiet, A. Togola, F. Brissaud, J.L. Seidel, H. Budzinski, F. Elbaz-
Lützhoft, S.E. Jorgensen, Chemosphere 36 (1998) 357. Poulichet, Environ. Sci. Technol. 40 (2006) 5282.
[11] Snow, J. Chromatogr. A 885 (2000) 445. [26] B.K. Burnison, A. Hartmann, A. Lister, M.R. Servos, T. Ternes, G. Van
[12] S. Castiglioni, R. Bagnati, D. Calamari, R. Fanelli, E. Zuccato, J. Chro- Der Kraak, Environ. Toxicol. Chem. 22 (2003) 2243.
matogr. A 1092 (2005) 206. [27] J.A. Pedersen, M. Soliman, I.H. Suffet, J. Agric. Food Chem. 53 (2005)
[13] T.A. Ternes, TrAC 20 (2001) 419. 1625.
[14] D.W. Kolpin, E.T. Furlong, M.T. Meyer, E.M. Thurman, S.D. Zaugg, L.B. [28] A. Joss, E. Keller, A.C. Alder, A. Gobel, C.S. McArdell, T. Ternes, H.
Barber, H.T. Buxton, Environ. Sci. Technol. 36 (2002) 1202. Siegrist, Water Res. 39 (2005) 3139.
[15] S. Weigel, K. Bester, H. Hünhnerfuss, J. Chromatogr. A 912 (2001) 151. [29] J.B. Quintana, S. Weiss, T. Reemtsma, Water Res. 39 (2005) 2654.
[16] L. Sarrazin, C. Diana, T. Schembri, P. Rebouillon, Int. J. Environ. Stud. 61 [30] B.W. Brooks, C. Kevin Chambliss, J.K. Stanley, A. Ramirez, K.E. Banks,
(2004) 413. R.D. Johnson, R.J. Lewis, Environ. Toxicol. Chem. 24 (2005) 464.
[17] A. Togola, H. Budzinski, Anal. Bioanal. Chem. 388 (2007) 627. [31] T.A. Ternes, M. Bonerz, N. Herrmann, D. Loffler, E. Keller, B.B. Lacida,
[18] M. Rabiet, F. Brissaud, J.-L. Seidel, S. Pistre, F. Elbaz-Poulichet, Appl. A.C. Alder, J. Chromatogr. A 1067 (2004) 213.
Geochem. 20 (2005) 1587. [32] O.A. Jones, J.N. Lester, N. Voulvoulis, Trends Biotechnol. 23 (2005) 163.
[19] H. Budzinski, M.H. Devier, P. Labadie, A. Togola, Anal. Bioanal. Chem. [33] S. Webb, T.A. Ternes, M. Gibert, K. Olejniczak, Toxicol. Lett. 142 (2003)
386 (2006) 1429. 157.