No job name

Anal. Chem. 2004, 76, 4756-4764
Trace Determination of Macrolide and Sulfonamide
Antimicrobials, a Human Sulfonamide Metabolite,
and Trimethoprim in Wastewater Using Liquid
Chromatography Coupled to Electrospray Tandem
Mass Spectrometry
1 bel,* Christa S. McArdell, Marc J.-F. Suter, and Walter Giger
Swiss Federal Institute for Environmental Science and Technology (EAWAG), CH-8600 Du¨bendorf, Switzerland An analytical method has been developed and validated
hence, pose a potential threat for the aquatic environment. In the for the simultaneous trace determination of four macrolide
case of antimicrobial agents (including both naturally and syn- antibiotics, six sulfonamides, the human metabolite N4-
thetically derived compounds), the possible maintenance and acetylsulfamethoxazole, and trimethoprim in wastewater.
spread of bacterial resistance is a major point of concern. In 1997, The method was validated for tertiary, secondary, ands
the human consumption of antibacterials in Switzerland exceeded unlike in previously published methodssalso for primary
30 tons per annum, (t/a).8-10 In addition ∼55 t/a of antimicrobials effluents of municipal wastewater treatment plants. This
were used in veterinary medicine. Macrolides (4.3 t/a), sulfon- wide range of application is necessary to thoroughly
amides (5.1 t/a), and fluoroquinolones (3.9 t/a) represent the most investigate the occurrence and fate of chemicals in waste-
important groups in human medicine, next to -lactams (17.5 t/a).
water treatment. Wastewater samples were enriched by
The latter include penicillins and cephalosporins and seem to be solid-phase extraction, followed by reversed-phase liquid
chromatography coupled to tandem mass spectrometry
In industrialized countries, most human use antimicrobials and using positive electrospray ionization. Recoveries from all
other pharmaceuticals reach the aquatic environment, unchanged sample matrixes were generally above 80%, and the
or transformed, mainly via discharge of effluents from municipal combined measurement uncertainty varied between 2 and
wastewater treatment plants (WWTPs). The residual concentra- 18%. Concentrations measured in tertiary effluents ranged
tions of these bioactive compounds in the treated effluents depend between 10 ng/L for roxithromycin and 423 ng/L for
on their removal during wastewater treatment. They can poten- sulfamethoxazole. Corresponding levels in primary ef-
tially pose a hazard for aquatic organisms if the removal is fluents varied from 22 to 1450 ng/L, respectively. Trace
incomplete. In addition, exposure via sewage sludge disposal on amounts of these emerging contaminants reach ambient
land could represent a hazard for soil organisms.
waters, since all analytes were not fully eliminated during
Detailed knowledge of the behavior of antimicrobials in conventional activated sludge treatment followed by sand
wastewater treatment and the aquatic environment will help to filtration. In the case of sulfamethoxazole, the amount
achieve a reliable basis for environmental risk assessment (e.g., present as human metabolite N4-acetylsulfamethoxazole
by providing measured environmental concentrations, MECs).
had to be taken into account in order to correctly assess
MECs can be used in environmental risk assessment studies since the fate of sulfamethoxazole in wastewater treatment.
they provide accurate indications of actual concentrations presentin environmental systems. Investigations on the occurrence and Since 1997, interest in the occurrence and behavior of pharmaceuticals in the aquatic environment has significantly fate of antimicrobial agents in various wastewater treatment steps increased.1-7 One motivation for this attention is the fact that these can be exploited in order to evaluate wastewater treatment chemicals are designed to trigger specific biological effects and, technologies with respect to elimination of specific contaminants.
Reducing the release of residual pharmaceuticals into the aquatic * Corresponding author. E-mail: anke.goebel@eawag.ch. Fax: +41 1 823 environment would presumably decrease any potential environ- mental risks. By monitoring receiving surface waters as well as (1) Stan, H. J.; Heberer, T. Analusis Mag. 1997, 27, 20-23.
(2) Ternes, T. A. Water Res. 1998, 32, 3245-3260.
wastewater treatment plants, locations of particular concern can (3) Halling-Sorensen, B.; Nors Nielson, S.; Lanzky, P. F.; Ingerslev, F.; Holten be identified and mitigated specifically.
¨tzenhoft, H. C.; Jorgensen, S. E. Chemosphere 1998, 36, 357-393.
(4) Daughton, C. D.; Ternes, T. A. Environ. Health Perspect. 1999, 107, 907-
(8) Annual Report; Swiss Importers of Antibiotics (TSA): Berne, Switzerland, ¨mmerer, K. Chemosphere 2001, 45, 957-969.
(6) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; (9) Pharmaceuticals Sold in Switzerland; Swiss Market Statistics, 1999.
Buxton, H. T. Environ. Sci. Technol. 2002, 36, 1202-1211.
(10) Antibiotics Used in Veterinary Medicine; Swiss Federal Office for Agriculture (7) Heberer, T. Toxicol. Lett. 2002, 131, 5-17.
Analytical Chemistry, Vol. 76, No. 16, August 15, 2004 To reach the aims stated above, selective and sensitive Table 1. Investigated Compounds
analytical methods for many different sample matrixes are es-sential. Until now, published methods for antimicrobial agents have focused on wastewater treatment plant effluents and surface waters,11-17 with the exception of fluoroquinolones, which were studied in detail by Golet et al.18-21 Analytical methods for wastewater matrixes other than final effluents including sludge extracts, however, are lacking. Another important aspect that has not yet been sufficiently addressed is the presence of human metabolites of antibacterials in wastewaters. Sulfamethoxazole, for example, is metabolized in the human body and ∼50% of the administered dose is excreted as the inactive human metabolite N4-acetylsulfamethoxazole and only 10% as the unchanged com- active sulfamethazine during the storage of manure has already a Used as internal standards.
been shown by Berger et al., suggesting a similar cleavage of N4-acetylated sulfonamides, for instance in wastewater treatment.23Observed elimination rates may be biased, if the possible retrans- the selected compounds in all compartments of a wastewater formation to the active pharmaceutical is not considered. To the treatment plant as well as for environmental monitoring studies.
best of our knowledge, only one study included N4-acetylsul- Preliminary results on the occurrence of macrolides and sulfon- famethoxazole in the analysis of surface water and WWTP amides in Swiss wastewater treatment plants are presented.
effluentssindicating concentrations of up to 2200 ng/L in WWTPeffluents.24 Unfortunately, the state of treatment has not been EXPERIMENTAL SECTION
reported. This clearly shows the importance of considering the Chemicals and Reagents. HPLC-grade methanol, acetonitrile,
main human metabolite of sulfonamides when assessing the and water are purchased from Scharlau (Barcelona, Spain).
occurrence and fate of sulfamethoxazole in wastewater treatment.
Analytical ethyl acetate, ammonia solution, 25% sulfuric acid, In this article, we present a reliable analytical method for the sodium chloride, sodium hydroxide, ammonium acetate, and trace determination of the most important macrolide and sulfon- formic acid were obtained from Merck (Darmstadt, Germany).
amide antibiotics in the various aqueous compartments of a Sulfamethazine, sulfamethoxazole, sulfadiazine, and roxithro- WWTP, including primary effluent. In addition, the human mycin were purchased from Sigma-Aldrich (Buchs, Switzerland).
metabolite N4-acetylsulfamethoxazole and trimethoprimsfrequently Sulfathiazole, sulfapyridine, trimethoprim, tylosin, josamycin, and used as a synergist to sulfamethoxazoleswere measured. Table erythromycin were obtained from Fluka Chemicals (Buchs, 1 lists the selected macrolides and sulfonamides; their respective Switzerland), and sulfamerazine was from Riedel-de Hae¨n (Seelze, chemical structures are given in Charts 1 and 2. Using solid-phase Germany). Sulfamethazine-phenyl-13C6 was purchased from Cam- extraction combined with liquid chromatography tandem mass bridge Isotope Laboratories (Andover, MA), and sulfamethoxazole- spectrometry (positive electrospray ionization), concentrations d4, sulfadiazine-d4, sulfathiazole-d4, and N4-acetylsulfamethoxazole- down to the low-nanogram per liter range can be determined. The d5 were purchased from Toronto Research Chemicals (North presented method is feasible to study the occurrence and fate of York, ON, Canada). Clarithromycin was kindly supplied by Abbott(Wiesbaden, Germany) and azithromycin by Pfizer (Zurich, (11) Hirsch, R.; Ternes, T. A.; Haberer, K.; Mehlich, A.; Ballwanz, F.; Kratz, K.- Switzerland). Azithromycin is also commercially available from L. J. Chromatogr., A 1998, 815, 213-223.
Sigma-Aldrich (Buchs, Switzerland). Standard solutions for eryth- (12) Sacher, F.; Lange, F. T.; Brauch, H.-J.; Blankenhorn, I. J. Chromatogr., A 2001, 938, 199-210.
romycin-H2O were prepared from erythromycin as described by (13) Lindsey, M. E.; Meyer, M.; Thurman, E. M. Anal. Chem. 2001, 73, 4640-
McArdell et al.15 The acidic solution was readjusted to pH 6 after 4 h using 1 M NaOH to ensure stability during storage. N4- (14) Hartig, C.; Storm, T.; Jekel, M. J. Chromatogr., A 1999, 854, 163-173.
(15) McArdell, C. S.; Molnar, E.; Suter, M. J.-F.; Giger, W. Environ. Sci. Technol.
Acetylsulfamethoxazole was synthesized by acetylation with acetic 2003, 37, 5479-5486.
acid anhydrate according to Neumann with a yield of 70%.25 (16) Giger, W.; Alder, A. C.; Golet, E. M.; Kohler, H.-P. E.; McArdell, C. S.; Identity and purity was confirmed by LC/UV, LC/MS/MS, and Molnar, E.; Siegrist, H. R.; Suter, M. J.-F. Chimia 2003, 57, 485-491.
(17) Vanderford, B. J.; Pearson, R. A.; Rexing, D. J.; Snyder, S. Anal. Chem. 2003,
Internal Standards. Deuterated sulfonamide standards were
(18) Golet, E. M.; Alder, A. C.; Hartmann, A.; Ternes, T. A.; Giger, W. Anal. commercially available in most cases. Erythromycin-13C2 was Chem. 2001, 73, 3632-3638.
(19) Golet, E. M.; Strehler, A.; Alder, A. C.; Giger, W. Anal. Chem. 2002, 74,
tested as an internal standard for the macrolides but proved to be unsuitable due to the significant natural contribution to M + (20) Golet, E. M.; Alder, A. C.; Giger, W. Environ. Sci. Technol. 2002, 36, 3645-
2 from unlabeled erythromycin (11.6%). Similar observations were (21) Golet, E. M.; Xifra, I.; Siegrist, H. R.; Alder, A. C.; Giger, W. Environ. Sci. described by Vanderford and co-workers for erythromycin-13C1.17 Technol. 2003, 37, 3243-3249.
The absence in water samples of all internal standards used was (22) Vree, T. B.; Hekster, Y. A. Pharmacokinetics of sulfonamides revisited; confirmed by enriching a representative samples from each matrix, (23) Berger, K.; Petersen, B.; Buening-Pfaue, H. Arch. Lebensmittelhyg. 1986,
to which no surrogate or instrumental standard was added. No (24) Hilton, M. J.; Thomas, K. V. J. Chromatogr., A 2003, 1015, 129-141.
(25) Neumann, J., Ph.D. Thesis in Pharmacy, Free University of Berlin, 1989.
Analytical Chemistry, Vol. 76, No. 16, August 15, 2004 Chart 1. Chemical Structures of Macrolide Antibiotics
Chart 2. Chemical Structures of Sulfonamide Antibiotics and Trimethoprim
peaks could be detected at the retention times of the used internal josamycin (JOS) for all macrolides. While the surrogate standards standards. Surrogate standards were added prior to enrichment were used for quantification, the instrumental standards were used to assess possible losses during the analytical procedure. Instru- to check the instrument performance during measurement. Its mental standards were added to the final extracts prior to peak area was monitored over the whole measurement series in measurement. The following substances were used as surrogate order to detect problems with, for example, instrument sensitivity standards: sulfamethazine-phenyl-13C6 (13C6SMZ) for SMZ, TRI, or injection volume. If the area of the instrumental standard and SPY, sulfamethoxazole-d4 (d4SMX) for SMX, sulfadiazine-d4 decreases significantly (signal reduction of >20% within same (d4SDZ) for SDZ, sulfathiazole-d4 (d4STZ) for STZ, N4-acetylsul- matrix), the series was stopped and the instrument cleaned.
famethoxazole-d5 (d5N4AcSMX) for N4AcSMX, and tylosin (TYL) Sample Collection and Preparation. Flow-proportional com-
for ERY-H2O, AZI, CLA, and ROX. As instrumental standards posite samples of the primary effluent (1°EFFL) after mechanical sulfamerazine (SMR) was used for all sulfonamides and TRI, and treatment, the secondary effluent (2°EFFL) after biological treat- Analytical Chemistry, Vol. 76, No. 16, August 15, 2004 ment, and the tertiary effluent (3°EFFL) after sand filtration were Table 2. Precursor Ions, Selected Fragment Ions, and
collected. The samples were transferred into amber glass bottles Retention Times of the Measured Compounds
and filtered as soon as possible but no later than 6 h after sampling through 0.45-µm cellulose nitrate filters (Schleicher & Schuell).
The filtered samples were directly extracted or kept at -20 °C in half-filled amber glass bottles in horizontal position until extraction.
The sample volumes were 250 mL for 2°EFFL and 3°EFFL samples and 50 mL for 1°EFFL samples. The latter were diluted with 150 mL of water prior to extraction. After addition of 1 g of sodium chloride, the pH was adjusted to 4 with sulfuric acid, and the surrogate standard (50-100 ng) was added. Solid-phase extraction was performed on 6-mL Oasis HLB sorbent cartridges (200 mg; Waters, Bergen op Zoom, The Netherlands) using a 12- fold vacuum extraction box (J.T. Baker, Phillipsburg, NY). The sorbent material is a copolymer of two monomers, N-vinylpyrroli- done and divinylbenzol. The cartridges were preconditioned with 2 × 1.5 mL of methanol-ethyl acetate (1:1), 2 × 1.5 mL of methanol containing 1% (v/v) ammonia, and 2 × 1.5 mL of water percolated through the cartridges at a flow rate of less than 5 a Used as internal standards.
mL/min. After percolation, the cartridges were washed with 1.5mL of water-methanol (95:5) and the eluent was discarded.
Subsequently, the cartridges were dried completely in a nitrogen gives the retention times of the individual analytes. To prevent flow for 1 h. The analytes were then eluted with 2 × 1.5 mL of sensitivity losses of the mass spectrometer, the eluate of the first methanol-ethyl acetate (1:1) and 2 × 1.5 mL of methanol 8 min and of the last 20 min of the chromatographic run were containing 1% ammonia into 10-mL graduated glass vessels.
Eluates were reduced to ∼50 µL by a gentle flow of nitrogen at Tandem Mass Spectrometry. A triple quadrupole mass
room temperature. After the addition of the instrumental standard spectrometer, TSQ Quantum Discovery (Thermo Finnigan, San (100 ng), the sample volume was adjusted to 0.5 mL with water.
Jose, CA), equipped with electrospray ionization was used for Final extracts were stored in amber glass vials at -15 °C until detection. Analyses were performed in the positive mode, with a spray voltage of 3500 V and an ion-transfer capillary temperature Liquid Chromatography. HPLC analyses were performed
of 350 °C. Nitrogen was used as sheath gas (40 bar) and as using a Rheos 2000 pump equipped with a solvent degasser (Flux auxiliary gas (10 bar), and argon as collision gas (1.5 mTorr).
Instruments AG), a HTS Pal autosampler (CTC Analytics, Zwin- Both mass analyzers were set to unit resolution. Usually, the gen, Switzerland), and a Jones chromatography column oven, protonated molecular ion ([M + H]+) of the compounds was model 7956 (Omnilab AG, Mettmenstetten, Switzerland). Sample selected as precursor ion except for azithromycin, for which the aliquots of 20 µL were injected. Two analytical columns were doubly charged molecular ion ([M + 2H]2+) was chosen as tested for separation. Initially, a 125 × 2 mm Nucleosil 100-5 C18- precursor ion because of its greater abundance under the given HD end-capped column (Macherey-Nagel, Dueren, Germany) conditions. Detection was performed in multiple reaction monitor- equipped with a 8 × 2 mm precolumn containing the same sorbent ing mode using the two most intense and specific fragment ions.
material was used (column 1). Gradient elution was performed Table 2 lists the monitored transitions for the individual analytes.
with water adjusted to pH 4.6 by acetic acid and acetonitrile, both The detection of the compounds was divided in time windows containing 10 mM ammonium acetate. Later, a 150 × 2 mm YMC during the course of the chromatographic run with a dwell time Pro C18, 120 Å, 3 µm (Stagroma, Reinach, Switzerland) column of 100 ms. Figure 1 shows a chromatogram of a 1°EFFL sample.
equipped with a 10 × 2 mm precolumn containing the same In the case of SDZ, STZ, and SMZ, which were not present in the sorbent was applied (column 2) was used. Optimal separation was sample, the peaks obtained from the measurement of a 1°EFFL achieved using column 2 maintained at 30 °C and with a flow rate sample spiked with 25 ng prior to sample preparation are included of 0.15 mL/min. Solvent A was water acidified with 1% (v/v) formic acid, resulting in a pH of 2.1, and solvent B was methanol acidified Method Validation. For the method validation, flow-propor-
with 1% (v/v) formic acid. The run (0.15 mL/min) started at 10% tional composite samples from the respective effluents of a WWTP B for 5 min, followed by a 5-min linear gradient to 15% B, a 5-min (Kloten-Opfikon) were taken. Breakthroughs were determined by linear gradient to 40% B, and another 5-min linear gradient to 45% extracting spiked wastewater samples (duplicate analyses) using B and was terminated by a 10-min linear gradient to 70% B.
two stacked cartridges. A breakthrough on the first cartridge Afterward, the eluent was brought to 100% B in 2.5 min and the triggered an enrichment on the consecutive cartridge, which was column washed at a flow rate of 0.25 mL/min for 10 min. Initial then eluted separately. For the 1°EFFL a 250-mL sample with a conditions were reestablished in 2.5 min, and the column was spiked analyte concentration of 2000 ng/L was used, and for the equilibrated for 10 min at a flow rate of 0.25 mL/min prior to the 3°EFFL a 500-mL sample with a spiked analyte concentration of next analysis. The total time per analysis was 55 min. Table 2 5000 ng/L was extracted. Complete elution of the cartridges was Analytical Chemistry, Vol. 76, No. 16, August 15, 2004 Figure 1. Total ion chromatogram (sum of two transitions) of an extract from a primary wastewater effluent (1°EFFL). a The peaks shown for
SDZ, STZ, and SMZ correspond to a 1°EFFL sample spiked with 25 ng of these compounds since they were not present in the unspiked
sample.
verified by eluting cartridges of spiked samples for a second time amount already present before spiking was then divided by the with 1.5 mL of acetone as a stronger solvent. The acetone extract was then treated as a separate sample. Instrumental limits of Identification and Quantification. For each substance, two
detection (LODs) and limits of quantification (LOQs) were transitions of the precursor ion were monitored. Together with calculated on the basis of standard deviation of the repeated the retention times, they were used to ensure correct peak measurement (n ) 10) of a standard mixture (100 pg on column).
assignment and to evaluate peak purity. For instrumental and The LOD is defined as 3 times and the LOQ as 10 times the surrogate standards, peak purity was tested using the area ratio standard deviation. If the resulting value for the LOQ was below of the two product ions monitored. Their individual ratio was the linear range, the lower limit of the linear range was set as calculated as well as the mean ratio of all samples and its relative LOQ. Sample-based LOD and LOQ were defined as concentrations standard deviation. The ratio in one sample was compared to the in a sample matrix resulting in peak areas with signal-to-noise mean ratio of all the samples measured in one series. The variance ratios (S/N) of 3 and 10, respectively. Since samples typically had to be within the range of twice the standard deviation of the contained analytes in higher amounts, the concentration corre- mean sample ratio. The peak purity of the analytes was tested by sponding to the defined S/N was determined by downscaling, calculating a concentration (as described below) for both product using the measured concentration and the assigned S/N of the ions measured. The respective surrogate product ion was used.
peaksassuming a linear correlation through zero. Instrumental If no surrogate product ion resulting from the same fragmentation precision of the measurement was assessed using an average of reaction can be used, i.e., if no isotope labeled surrogate standard 10 independent injections of 100 pg on column of a standard is available, the sum of both product ions of the compound mixture. The precision of the entire method was determined using assigned as surrogate standard was used for quantification to four replicates of each matrix investigated, spiked with 50 ng of simplify the procedure. The relative average deviation of the analyte prior to extraction. It is indicated by the relative standard calculated concentrations from the two product ions had to be deviation of the measured concentrations of native plus spiked less than 10%. Peaks not fulfilling the requirements for peak purity analyte. For recovery studies over the entire procedure, waste- water samples (duplicate analyses) were spiked prior to extraction Quantification was performed using the ratio of the peak areas with surrogate standard and with 25 and 50 ng of analytes, of the analytes and of the surrogate standard. The sum of the respectively. The calculated amount of antibacterials minus the two monitored product ions was used. An external calibration Analytical Chemistry, Vol. 76, No. 16, August 15, 2004 curve, plotting ratio against concentration, was obtained by diluting Scheme 1. Breakdown Curves of
standards in HPLC water. A standard curve was acquired at the N4-Acetylsulfamethoxazole and Proposed
beginning, at the end, and also in the middle of a measurement Product Ion Structures (Absolute Intensity 3.56
× 105, Collision Pressure 1.5 mTorr)
series. At least five concentration points in the appropriateconcentration range were used for quantification.
Concentrations in the samples were calculated by comparing the peak area ratios of the analytes and their assigned surrogatestandards in the SPE extracts, to the corresponding ratios in thestandard solutions. These results were corrected with the corre-sponding recovery rates obtained in the same matrix and samplebatch to provide accurate amounts. For routine determination,duplicate analyses of all samples were performed. Proceduralblanks, consisting of deionized water, were analyzed with eachset of 12 extractions as a control for laboratory contamination.
Additional instrumental blanks using deionized water were checkedwith each calibration curve in order to uncover potential analyticalinterferences.
RESULTS AND DISCUSSION
Method Development. The crucial parameters for enrich-
ment, separation, and detection of the analytes were identifiedand optimized. The pH of the sample proved to be the mostinfluential variable during sample extraction. A critical impact onthe retention of the analytes on the cartridge material wasobserved, especially for sulfonamides caused by their aminogroups. Our enrichment tests between pH 2 and 6 revealed, asexpected, highest recoveries at pH 4 for the sulfonamides, whilethe recovery of the macrolides and trimethoprim showed nostrong pH dependence. This behavior can be explained by thecharge state of the sulfonamides at the particular pH values (Table1).26-31 With a compound specific pKa of 5-8 for the sulfonaminogroups (pKa 1) and a pKa of 2-2.5 for the arylamin (pKa 2), thesulfonamides are positively charged at pH 2 and negatively at a shows the breakdown curves for N4AcSMX and its four most pH above 5. The interaction with the cartridge material is strongest intense fragments as a function of the collision energy. As for analytes in uncharged forms occurring at a pH of ∼4 in the expected, the collision energy, which gives the most intense case of the sulfonamides. The dilution of the 1°EFFL samples prior signal, increases for the formation of smaller fragments. Tentative to enrichment additionally increased signal intensity provided by product ion structures are given also. These structures have not the mass spectrometer for the sulfonamidessin most cases by a been reported previously but are in agreement with transitions factor of 2. For the macrolides and trimethoprim, no significant known to be typical for sulfonamides.32, 33 During LC/MS/MS measurement, matrix compounds can be While N4-acetylsulfamethoxazole is stable during sample deposited on the instrument’s sample interface, especially on the preparation, erythromycin present in the samples is completely ion-transfer capillary, and can thus significantly reduce instrument sensitivity. The higher the sample volume the more matrix will with the reported instability of erythromycin under acidic condi- be introduced into the mass spectrometer within one run. On the tions resulting in the formation of the inactive erythromycin-H other hand, a high enrichment factor is desirable to achieve the Erythromycin was therefore assessed as the main environmental low limits of detection, which are necessary for the environmental analysis of antimicrobials. The sample volume was optimized by Tandem mass spectrometric conditions were optimized for using a variable splitting device prior to the electrospray interface.
each analyte and internal standard through automated tuning For this experiment, higher sample volumes were chosen.
procedures implemented in the instrument software. Scheme 1 Therefore, 200 mL of 1°EFFL and 1000 mL of 2°EFFL and 3°EFFL, respectively, were enriched and measured in one series.
(26) McFarland, J. W.; Berger, C. M.; Froshauer, S. A.; Hayashi, S. F.; Hecker, The eluent flow and split ratio were varied, so that the instrument S. J.; Jaynes, B. H.; Jefson, M. R.; Kamicker, B. J.; Lipinski, C. A.; Lundy, K.
M.; Reese, C. P.; Vu, C. B. J. Med. Chem. 1997, 40, 1340-1346.
remained sensitive enough for the measurement of up to 30 (27) Bryskier, A. J.; Butzler, J.-P.; Neu, H. C.; Tulkens, P. M. Macrolides; Arnette samples of each matrix. The sample volume used was then adjusted according to the split ratio. Subsequent samples were (28) Vree, T. B.; Hekster, Y. A. Clinical pharmacokinetics of sulfonamides and their metabolites; Karger: Basel, 1987.
measured without the additional splitting device that may pose (29) Lin, C.-E.; Chang, C.-C.; Lin, W.-C. J. Chromatogr., A 1997, 768, 105-112.
(30) Petz, M. Habilitationsschrift in Chemie; Westfa¨lische Wilhelms-Universita¨t,
(32) Volmer, D. Rapid Commun. Mass Spectrom. 1996, 10, 1615-1620.
¨ller, S. R.; McArdell, C. S.; Alder, A. C.; Suter, M. J.-F. J. (31) Neuman, M. Antibiotika-Kompendium; Verlag Hans Huber: Bern, 1981.
Chromatogr., A 2002, 952, 111-120.
Analytical Chemistry, Vol. 76, No. 16, August 15, 2004 Table 3. Linear Ranges and Limits of Quantification
a Concentration estimated from measured samples for a signal-to-noise ratio of 10.
as a source for errors (for example, plugging of the capillary).
analytes are quantitatively enriched by one cartridge and exhaus- The given sample volumes therefore represent a compromise tively eluted by the procedure described above.
between method sensitivity and routine analysis. Two columns For the standard curves, good linearity was observed with were tested for the separation of the selected antibacterial agents.
correlation factors typically above 0.99. The linear range of the In both cases, a reversed-phase end-capped C18 column was measurement varied with the analyte due to differences of the chosensone belonging to the older (column 1) and one belonging ionization efficiencies (Table 3). The instrumental LOQ ranges to the newer (column 2) generation of silica gels. On column 1, between 16 and 100 pg of analyte on column. In the case of the azithromycin produces a peak with substantial tailing. To our sample-based LOQ and LOD, the range and the average of the knowledge, azithromycin has not been included in analytical resulting values for each matrix from different samples are given methods for environmental samples so far, likely also due to in Table 3. Since the LOD and LOQ in an individual sample can analytical difficulties like this. The observed tailing on column 1 be higher or lower than the average LOD and LOQ, all concentra- is probably due to the interaction of the two basic functional tions resulting from peaks with S/N greater than or equal to 3 groups with residual silanol groups and metal impurities of the and 10, respectively, are considered valid results.
column material. In the case of the other macrolides, which The instrumental precision of the method was addressed for contain one amino group less than azithromycin in the lactone various aspects, and the following relative standard deviations were moiety, this interaction was sufficiently suppressed by the addition obtained: the retention time ranged between 0.06 and 0.35% and of ammonium acetate. On column 2, belonging to the new the peak area between 1.3 and 9.2%. The peak area ratios of analyte generation of silica gels, however, good separation was achieved versus surrogate standard varied to a lesser extent in most cases with almost symmetrical peaks for all analytes. In addition, the (between 1.3 and 7%), since the surrogate standard compensates use of ammonium acetate in the eluent was no longer necessary.
for analytical variability. The precision of the entire method This significantly increased the sensitivity of the method for (reproducibility) is indicated by the standard deviation of multiple sulfonamides, which tend to form ammonium adducts during analyses and ranged between 0.5 and 15%. Detailed results are In the case of some 1° EFFL samples, the extracts needed to Accuracies of the method were determined by recovery studies be diluted up to five times in order to obtain good peak shapes over the entire procedure (Table 4). The resulting recoveries for azithromycin, which seems to form complexes with matrix obtained in all matrixes investigated were generally above 80%, compounds. For most analytes, the assumed loss of sensitivity with the exception of TRI where they ranged between 30 and 47%.
due to dilution is partly compensated by the simultaneous For TRI, this was caused by the use of a nonideal surrogate reduction of ion suppression, since signal intensities observed are standard (13C6SMZ), but none better suited could be found.
Recoveries, and thereby LODs and LOQs, of the analytes vary Method Validation. The developed method was validated for
between samples, mainly due to varying matrix effects, if no primary effluents after mechanical treatment, secondary effluents isotopically labeled surrogate standard is available. Correct after biological treatment, and tertiary effluents after sand filtration.
quantification can still be ensured if recovery studies are per- For breakthrough studies, samples representing unnaturally high formed in each matrix and sample batch, as was the case within concentrations and high loads of sample matrix were enriched the work presented here. The combined measurement uncertainty on two stacked cartridges. No quantifiable amounts of the analytes was quantified using data from the method validation as described could be detected on the second cartridge for both sample in example A4 of the EURACHEM/CITAG Guide Quantifying matrixes (1°EFFL and 3°EFFL). When testing for complete Uncertainty in Analytical Measurement.34 Therefore, all uncertainty elution, no quantifiable amounts of analytes could be measured sources were identified and quantified. The main contributions in the acetone eluates of already eluted cartridges. Thus, the result from the repeatability of the measurement, calculated from Analytical Chemistry, Vol. 76, No. 16, August 15, 2004 Table 4. Method Precisions, Accuracies, and Combined Measurement Uncertainties
a SD, standard deviation. b n, number of samples.
Table 5. Sulfonamide and Macrolide Concentrations Measured in Two Municipal Wastewater Treatment Plants in
Switzerland
sample concn ( ADa (ng/L), nb ) 2 a Concentration measured in filtered 72-h flow proportional composite sample. Mean and average deviation (AD) of duplicate determination.
b Number of measurements. c WWTP 1, Kloten-Opfikon (canton Zurich); WWTP 2, Altenrhein (canton St. Gall). d nd, not detected, signal-to-noisebelow 3; nq, not quantifiable, signal-to-noise below 10.
duplicate sample analysis, and its accuracy, represented by tertiary effluents. Samples were taken in February 2003. With recovery studies. The relative values of all uncertainty sources 54 100 and 40 000 inhabitant equivalents, the two investigated are finally combined using statistical methods. The values for the treatment plants are of similar size and have comparable volumes combined measurement uncertainty vary between 2 and 18% with of wastewater inflow. This is also reflected in the similar the analyte and the matrix investigated (Table 4).
concentration ranges found at each plant, with the exception of Wastewater Applications. The developed method was suc-
azithromycin. The latter appears to be more frequently used in cessfully applied to the analyses of wastewater samples from two the catchment area of WWTP2. Sulfamethoxazole and clarithro- urban wastewater treatment plants in Switzerland: WWTP Kloten- mycin were found to be the most commonly used sulfonamides Opfikon, located near the international airport of Zurich (WWTP1), and macrolides, respectively. Sulfadiazineswhich is very rarely and WWTP Altenrhein, located in the canton St. Gall close to the applied in human medicinescould not be quantified in any of the border with Austria (WWTP2). In both cases, mechanically treated samples, nor could sulfathiazoleswhich is almost exclusively used wastewater (primary effluent) passes through conventional acti- vated sludge treatment, followed by secondary settling (secondary effluent). After biological treatment, both treatment plants use sand treatment plants (WWTP2). N4-Acetylsulfamethoxazole is typically filtration as a tertiary treatment step (tertiary effluent). Table 5shows the results obtained from duplicate analyses of 72-h flow- present in high amounts in the primary effluents, but only small proportional composite samples of the primary, secondary, and amounts can be found in the tertiary effluents. If the amount ofsulfamethoxazole present as acetyl metabolite is neglected, the (34) http://www.measurementuncertainty.org/mu/guide/index.html, Quantify- elimination of sulfamethoxazole will be underestimated. Concen- ing uncertainty in analytical measurement/prep. by the EURACHEM trations of the analytes in both tertiary effluent range between 19 Working Group on Uncertainty in Chemical Measurement (ISBN 0 94892615 5).
and 352 ng/L. This clearly shows that the compounds investigated Analytical Chemistry, Vol. 76, No. 16, August 15, 2004 are not eliminated completely and reach receiving surface waters.
including alternative wastewater technologies such as biofilter Compared to results obtained in Germany,14,35 the concentrations technology and membrane filtration.36 Our ongoing studies are found are in the same range but generally lower.
also aimed at achieving complete mass balances of antimicrobialsin wastewater treatment plants, including sewage sludge treatment CONCLUSIONS
steps. Preliminary results show that the method can easily be Solid-phase extraction Oasis HLB cartridges coupled with adapted for the analyses of sewage sludge extracts. Applications reversed-phase liquid chromatography and tandem mass spec- to drinking water, ambient waters, and hospital wastewaters also trometry were successfully applied for the determination of seem to be possible judging from first measurements without selected sulfonamide and macrolide antibiotics, in addition to major changes in the procedure. With this method, we therefore trimethoprim and the human sulfonamide metabolite N4-acetyl- present a powerful tool to fully assess the fate and occurrence of sulfamethoxazole, in municipal wastewater. As a result of this macrolides and sulfonamides throughout their main pathways to method’s applicability to wastewater samples spanning the whole treatment process (including primary effluent samples), it can beused to investigate the fate of these compounds through the ACKNOWLEDGMENT
Abbott GmbH (Wiesbaden, Germany) is acknowledged for various steps of wastewater treatment. The resulting information supplying clarithromycin and Pfizer A.G. (Zu can be used to evaluate the performance of wastewater treatment supplying azithromycin. Partial financial support came from the procedures and to highlight options for the optimization of EU project POSEIDON (EVK1-CT-2000-00047)37 and the EAWAG WWTPs with the aim of minimizing the input of antibiotics into project on human-use antibiotics (HUMABRA) within the frame- ambient receiving waters. Additionally, by including N4-acetyl- work of the National Research Program on antibiotic resistance sulfamethoxazolesthe main human metabolite of sulfamethoxa- funded by the Swiss National Science Foundation.38 We thank Eva zolesthe fate of the most commonly used sulfonamide in human medicine can be investigated more thoroughly.
technical assistance and advice. For helpful comments on the The presented method provides the necessary basis for a manuscript, we acknowledge our colleagues Alfredo Alder, comprehensive study on antibacterials in wastewater treatment (35) Hirsch, R.; Ternes, T. A.; Haberer, K.; Kratz, K.-L. Sci. Total Environ. 1999,
Received for review March 3, 2004. Accepted May 12, ¨bel, A., Ph.D. Thesis, ETH Zurich, in preparation.
(37) http://www.eu-poseidon.com.
(38) http://www.nrp49.ch/pages/.
Analytical Chemistry, Vol. 76, No. 16, August 15, 2004

Source: http://www.nrp49.ch/upload/docs/dokumente/gobel_analytchem_04.pdf