Residue Analysis of the Pharmaceutical Diclofenac in Different Water

Jul 3, 2003 - Do Concentrations of Ethinylestradiol, Estradiol, and Diclofenac in European Rivers Exceed Proposed EU Environmental Quality Standards? ...
13 downloads 17 Views 127KB Size
Environ. Sci. Technol. 2003, 37, 3422-3429

Residue Analysis of the Pharmaceutical Diclofenac in Different Water Types Using ELISA and GC-MS ANPING DENG,# MARKUS HIMMELSBACH,§ QING-ZHI ZHU,† SIEGFRIED FREY,| MANFRED SENGL,| WOLFGANG BUCHBERGER,§ REINHARD NIESSNER,‡ AND D I E T M A R K N O P P * ,‡ Institute of Hydrochemistry and Chemical Balneology, Technical University Munich, Marchioninistrasse 17, D-81377 Munich, Germany, Institute of Chemistry, Department of Analytical Chemistry, Johannes-Kepler-University Linz, Altenbergerstrasse 69, A-4040 Linz, Austria, and Bavarian Water Management Agency, Kaulbachstrasse 37, D-80539 Munich, Germany

A highly sensitive and specific indirect competitive enzyme-linked immunosorbent assay (ELISA) for the determination of diclofenac in water samples was developed. With pure water, the limit of detection (LOD, S/N ) 3) and IC50 were found to be 6 ng/L and 60 ng/L, respectively. The analytical working range was about 20-400 ng/L. Highest cross-reactivity (CR) of 26 tested pharmaceuticals, metabolites, and pesticides was found for 5-hydroxydiclofenac (100%). Other estimated values were well below 4% and, therefore, are negligible. The assay was applied for the determination of diclofenac in tap and surface water samples as well as wastewater collected at 20 sewage treatment plants (STPs) in Austria and Germany. Humic substances were identified as main interference in surface water. Wastewater samples which were only submitted to filtration and dilution yielded about 25% higher diclofenac concentrations using the ELISA compared to GC-MS. However, the ELISA turned out to be a simple, inexpensive, and accurate method for the determination of diclofenac both in influent and effluent wastewater after rather simple sample preparation, i.e., filtration, acidification, and readjustment to neutral pH-value, and at least 10-fold dilution with pure water.

Introduction Within the last years a number of investigations have been reported which clearly demonstrate the widespread occurrence of therapeutic agents in the environment, notably in the aquatic compartment at low ppt or ppb (ng - µg/L) concentrations, thus establishing these compounds as a new * Corresponding author phone: + 49 89 2180 78252; fax: + 49 89 2180 78255; e-mail: [email protected]. # On leave from Sichuan University, Chengdu, China. § Johannes-Kepler-University Linz. † An Alexander von Humboldt Fellow on leave from the Department of Chemistry, Xiamen University, Xiamen, China. | Bavarian Water Management Agency. ‡ Technical University Munich. 3422

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 15, 2003

class of pollutants. At least 60 pharmaceuticals have been identified in aquatic matrices (1-5), among them analgesics, antibiotics, antiepileptics, β-blockers, β2-sympathomimetics, and blood lipid regulators. The pharmaceuticals and their metabolites may originate from both human and veterinary usage and can be released directly after passing through sewage treatment plants, after sludge dispersion on farm land, and by other sources such as landfill leachates and nonpoint discharges from agricultural runoff. Years after the first detection of pharmaceuticals in wastewater and surface water, new powerful analytical techniques such as GC/MS and LC-ESI/MS have proven the presence of these compounds even in groundwater and drinking water (6-11). Current knowledge about the long-term effects of low-level exposure of such compounds to both aquatic ecosystems and humans is limited, and chronic effects may not become apparent for many years. However, because of the potential of a wide range of xenobiotic chemicals to interact and disrupt the endocrine systems of animals and humans, the presence of microcontaminants in the aquatic environment is of increasing public concern (12-15). Diclofenac belongs to the most frequently detected pharmaceutically active compounds in the water-cycle. It is considerably stable under normal environmental conditions. The most probable degradation pathway for in-situ elimination is photodegradation which will be influenced by additional key parameters such as eutrophic conditions, degree of suspended particulate material, or the depth of the watercourse (16). Approximately 75 tons of this nonsteroidal antiinflammatory prescription drug are annually sold in Germany (2). Average concentrations in the low ppb range were detected in influents and effluents of municipal sewage treatment plants (STPs) and surface waters in Austria, Brazil, Germany, Greece, Spain, Switzerland, and the United States (16, 17). Under recharge conditions, diclofenac has also been detected in groundwater and sporadically at trace level concentration in raw and treated drinking water (16). Both gas and liquid chromatography combined with MS(-MS) detection have been applied for the analysis of diclofenac after preconcentration by solid-phase extraction from water samples (10, 18-20). Highest sensitivity was reported for GC/MS/MS with MDLs of 0.3-4.5 ng/L and LOQs of 1, 5, and 50 ng/L using drinking water, surface water, and STP effluent, respectively. As a useful alternative capillary electrophoresis (CE) - MS was reported (21). Immunochemical techniques provide another approach for analyzing pharmaceuticals by taking advantage of the highly selective binding by antibodies. Whereas a great number of immunoassays have been developed and used for analysis of pesticides which are found at similar levels in the aquatic ecosystem, only very few tests have been applied for pharmaceutical compounds (22, 23). Available test kits for pharmaceuticals are optimized for biological specimens such as blood and urine. Applicability to environmental samples was neither considered nor examined for the vast majority of these kits. To the best of our knowledge, only a few groups have shown the feasibility of adapting clinical assays to the analysis of water (24-29). In various river and drinking water samples and in wastewater detection limits in the very low ng/L range have been achieved for hormones such as progesterone, norethisterone, ethinyl estradiol, 17βestradiol, and testosterone and the anticancer drugs methotrexate and bleomycin. As far as we know, the preparation of antibodies against diclofenac has not been reported so far. In the present study, we focused on the generation of 10.1021/es0341945 CCC: $25.00

 2003 American Chemical Society Published on Web 07/03/2003

anti-diclofenac antibodies and the development of an ELISA for this pharmaceutical and its use for the analysis of tap water, surface water, and wastewater samples.

Experimental Section Reagents. All reagents were of analytical grade unless specified otherwise. The sodium salt of diclofenac (2-(2,6dichlorophenyl]amino)benzeneacetic acid), chicken egg albumin (OVA), porcine thyroglobulin (TG), gelatin, milk powder, casein, tributylamine, isobutyl chloroformate, 1,4dioxane, methanol, ethanol, dimethyl sulfoxide (DMSO), and acetonitrile were purchased from Sigma-Aldrich (Munich, Germany). Tetramethylbenzidine (TMB), 30% hydrogen peroxide (H2O2), and Tween 20 were obtained from Merck (Darmstadt, Germany). Goat anti-rabbit IgG-horseradish peroxidase conjugate (GaRIgG-POD) and Sea Block blocking buffer were purchased from Pierce (Rockford, U.S.A.). Bovine serum albumin (BSA) was from Serva Feinbiochemica (Heidelberg, Germany), and humic acid was from Carl Roth (Karlsruhe, Germany). Freund’s complete and incomplete adjuvants were obtained from Difco Labs (Detroit, U.S.A.). β-Glucuronidase/arylsulfatase solution was obtained from Roche Applied Science (Mannheim, Germany). Preparation of Immunogen and Coating Antigen. BSA and TG were used as carrier proteins for the preparation of hapten-protein conjugates. After converting diclofenac sodium salt into the free acid it was coupled to the proteins by the mixed anhydride method. Briefly, diclofenac (0.1 mmol) was dissolved in 3 mL of dry 1,4-dioxane. Then equimolar amounts of tributylamine and isobutyl chloroformate were added. After mixing, the solution was placed in a 4 °C refrigerator for about 30 min, after which it was added dropwise to a 4 mL aqueous solution of 40 mg of protein. The pH was set to about 8.5 with 0.1 mol/L NaOH and reexamined twice within the next hour. After stirring for 4 h at room temperature, the precipitate was separated by centrifugation, and the remaining solution was exhaustively dialyzed in PBS using Visking dialysis tubing from Serva Feinbiochemica (Heidelberg, Germany). Finally, the conjugates were lyophilized and stored at -20 °C until use. The quantity of hapten coupled to BSA was estimated by MALDI-TOF MS analysis (AnagnosTec GmbH Luckenwalde, Germany). Coupling density was high with approximately 31 molecules of diclofenac per molecule of protein. The diclofenac-TG conjugate was used as coating antigen, and hapten density was not determined. Antisera Production. Two adult random-bred rabbits were immunized with the diclofenac-BSA conjugate. Primary immunization was performed intradermally at 10 multiple sites by injecting 0.1 mL of the emulsified immunogen, which was prepared by dissolving 2 mg of conjugate in 1 mL of sterilized physiological saline solution emulsified with 1 mL of Freund’s complete adjuvant. Booster injections were administered at 7, 16, 19, 22, and 28 weeks after priming using Freund’s incomplete adjuvant. After the first booster, an immunogen switch was performed, and further immunizations were done with the diclofenac-BSA conjugate. Rabbits were bled through the ear vein. Serum samples obtained from final bleeding 3 weeks after the last booster injection were dispensed in 1-mL aliquots and stored frozen in liquid nitrogen. The serum from the rabbit which showed highest titer and sensitivity was used in this study. Optimized ELISA Procedure. Indirect competitive ELISA format was adapted for the analysis of diclofenac. ELISA was run as follows. Microtiter plates (96 flat-bottom wells with high binding capacity; Greiner, Frickenhausen, Germany) were coated with the coating antigen (diclofenac-TG conjugate, 20 ng/mL; 200 µL/well) in coating buffer (0.05 M sodium carbonate buffer, pH 9.6). The plates were covered with adhesive plate sealing film (ThermalSeal, Sigma) to

prevent evaporation. After overnight incubation at 4 °C, the plates were washed with PBS-Tween (0.01 M PBS, pH 7.4, containing 0.15 M NaCl and 0.1% Tween 20) using an automatic plate washer (1296 026 Delfia Platewash, Wallac ADL, Freiburg, Germany). Binding sites not occupied by the coating antigen were blocked with blocking buffer (PBS containing 1% casein; 300 µL/well) for 1 h at room temperature. Plates were then washed as before. For construction of the calibration curve a diclofenac stock solution (0.5 g/L) was prepared with methanol and then further diluted with pure water to obtain standard solutions which cover the concentration range between 0.01 and 10 µg/L. Samples or standard solutions (100 µL/well) and diluted rabbit antiserum (1:20 000 in PBS; 100 µL/well) were added and incubated at room temperature for 1 h. After washing with PBS-Tween, GaRIgG-POD was added (1:8000 in PBS; 200 µL/well), incubated at room temperature for 1 h, and plate washed as before. The substrate solution (TMB/H2O2) was prepared by mixing TMB stock solution in DMSO (1%, w/v, 100 µL), H2O2 (5%, 10 µL), substrate buffer (0.1 M sodium acetate buffer, adjusted to pH 5.8 by adding 1 M citric acid solution, 500 µL), and pure water (10 mL). Substrate solution (200 µL/well) was added, and the plates were shaken for about 15 min for color development. Finally, the enzyme reaction was stopped with sulfuric acid (5%; 100 µL/well), and the absorbance was read at 450 nm with a plate reader (Easy Reader 340 ATC, SLT Labinstruments). All determinations were made at least in triplicate. The sigmoidal standard curves were set up using Rodbard’s four-parameter function and were plotted in the form of B/B0 × 100 (%) against log C (where B and B0 were the values of absorbance measured at the standard concentrations and at zero concentration, respectively). Cross-Reactivity Determination. The relative sensitivity of the immunoassay toward the compounds listed in Table 1 was determined by assaying a dilution series of each compound in water containing 10% of methanol (for fenoprofen, mefenamic acid, and tolfenamic acid, due to their lower solubility in 10% methanol, they were assayed in 10% DMSO). All chemicals were tested in the concentration range of 0.001-1000 µg/L. The IC50 values (concentration of inhibitor that produces a 50% decrease of the maximum normalized response) were compared and expressed as a percent IC50 based on 100% response of diclofenac. Water Samples. Two tap water samples were obtained from different local water works using groundwater as the source. Three random surface water samples were collected near the banks of the lakes Ammersee and Wo¨rthsee as well as the river Windach. Thirteen wastewater samples (composite samples of raw influents and final effluents) were made available from five different sewage treatment plants (STPs) in south Bavaria and effluent samples from fifteen STPs in Austria during the period September-November 2002. The STPs investigated are located in rural areas connected to a small number of households as well as in a large city. The STPs are connected to sewage systems which service regions ranging between 1000 and 2 000 000 residents. All STPs utilize three commonly used treatment steps: preliminary and final clarification and an aerator tank. Additionally, seven STPs are equipped with phosphate elimination, nitrification, and denitrification treatment steps. All the samples were analyzed on the sampling day or stored overnight in the fridge and analyzed on the following day, to keep microbial degradation to a minimum. Drinking and surface water samples were analyzed after dilution or without any preparation. In contrast, wastewater was analyzed after filtration (glass microfiber filters GF/C; Whatman Cat. No. 1822 047) to remove particles >1.2 µm. The filtrates were diluted with pure water and then applied to the ELISA procedure. GC-MS Procedure. Analysis was done at two different laboratories using slightly modified procedures (methods A VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3423

TABLE 1. Cross-Reactivity of Selected Pharmaceuticals, Metabolites, and Pesticides with Diclofenac Antiserum

a Inhibitor concentration for 50% inhibition in the competitive ELISA. b Percentage of cross-reactivity defined as (IC of diclofenac/IC analogue) 50 50 × 100. Compound was supplied by *Sigma, ** Serva, *** Mikromol (Luckenwalde, Germany), or # Novartis Pharma AG (Basel, Switzerland).

and B). Consistently, sample volume was 1 L, and the pH of filtered water samples was adjusted to about pH 2 with 1 N HCl to enhance trapping of diclofenac on the SPE sorbent. Method A: For solid-phase extraction 1 g of Bondesil C18 (Varian, Palo Alto, CA) was manually filled into glass cartridges and consecutively conditioned with 2 × 5 mL acetone, 2 × 3424

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 15, 2003

5 mL methanol, and 2 × 5 mL water (pH 2). During the enrichment of the sample the flow rate was approximately 15 mL/min. Afterward the cartridges were dried by air and then eluted with 3 mL of methanol. Before derivatization the extract was brought to dryness in a gentle nitrogen stream. The derivatization was carried out by adding 300 µL of a 2%

solution of pentafluorobenzylbromide in toluene and 3 µL of triethylamine to the dry sample and heating it at 90 °C for 1 h. After cooling, 700 µL of toluene and 20 µL of desmetryn in toluene (10 µg/mL) as internal standard were added. This solution was used for analysis by GC-MS. The instrumentation for GC-MS analysis consisted of a HP 5890 Series II gas chromatograph equipped with a splitsplitless autoinjector and coupled to a HP 5989 A quadrupole mass spectrometer (Agilent, Palo Alto, CA). Separations were carried out by means of a HP 1701 column, 30 m × 0.25 mm i.d., film thickness 0.25 µm. Helium was used as mobile phase at 70 kPa. The injection volume of 2 µL was injected in the splitless mode at 250 °C. After 1 min at 90 °C the temperature of the oven was raised to 150 °C at a rate of 30 °C/min, to 210 °C at a rate of 3 °C/min, and to 280 °C at a rate of 15 °C/min. The interface temperature was 300 °C, the temperature of the ion source (electron impact) 100 °C, and the quadrupole temperature 174 °C. The mass spectrometer was operated in the SIM-mode at the following m/z ratios (the bold numbers indicate the m/z ratios used for quantitative analysis, whereas the other m/z ratios were used for qualitative confirmation of peak identity): 213, 198, 171, 475, 242, 214. Method B: To the water sample were added 2,3dichlorophenoxyacetic acid and 2-(2,4-dichlorophenoxy)propionic acid (each 100 ng/µL solution in acetone), which were used as surrogate standards for the overall procedure. Automated SPE was done with 6-mL plastic cartridges filled with 1 g of RP-C18 material (Strata C18-E, Phenomenex, Torrance, CA). The solid phase was consecutively conditioned with 10 mL of methanol and 30 mL of water. Enrichment of the sample was done at a flow rate of about 5 mL/min followed by drying with nitrogen and elution with 4 mL of acetone. The extract was brought to a volume of about 100 µL in a gentle stream of nitrogen and finally to dryness in an oven at 50 °C. For the control of the derivatization step, 100 ng of phenoxyacetic acid was added. The derivatization was carried out by adding 200 µL of a 2% solution of pentafluorobenzylbromide in cyclohexane and 2 µL of triethylamine to the dried sample and heating it at 100 °C for 2 h. After cooling, this solution was analyzed by GC-MS. The instrumentation for GC-MS analysis consisted of a 6890N gas chromatograph equipped with a split-splitless autoinjector 7683 and coupled to a 5973N quadrupole mass spectrometer (Agilent, Palo Alto, CA). Separations were carried on a HP-5MS column, 30 m × 0.25 mm i.d., film thickness 0.25 µm. Helium was used as mobile phase at 94 kPa. The injection volume of 2 µL was injected in splitless mode at 275 °C. After 2 min at 65 °C the temperature of the oven was raised to 180 °C at a rate of 30 °C/min and then to 300 °C at a rate of 5 °C/min. This final temperature was held for 12 min. The interface temperature was 300 °C, the temperature of the ion source (electron impact) was 230 °C, and the quadrupole temperature was 150 °C. The mass spectrometer was operated in the SIM-mode at the following m/z ratios (the bold numbers indicate the m/z ratios used for quantitative analysis, whereas the other m/z ratios were used for qualitative confirmation of peak identity): 179, 181, 214, 216, 242, 244.

Results and Discussion Optimization of Assay Conditions. To develop a highly sensitive ELISA, the assay conditions such as the appropriate concentration of the immunoreagents (coating antigen, antiserum and GaRIgG-POD) used in the assay, the suitable blocking reagent, the effect of the temperature, etc. should be carefully optimized. Criteria used to evaluate the optimization were maximum absorbance (B0), dynamic range, IC50 and detection limit (LOD).

FIGURE 1. Dose-response curve for diclofenac calibrators obtained with spiked ultrapure water samples. Error bars represent (1 standard deviation about the mean. It was found that the best combination of the immunoreagents used in this study was a concentration of 20 ng/ mL of diclofenac-TG conjugate for coating and dilutions of 1:20 000 for antiserum and 1:8000 for GaRIgG-POD conjugate, respectively. It is well-known that the unspecific adsorption of the peroxidase labeled secondary antibody onto the surface of the plate is one of the important factors which could lead to low sensitivity of the ELISA. This could be prevented with a suitable blocking reagent. Five blocking reagents (casein, OVA, milk powder, gelatin and Sea Block) were examined for their blocking capacity. The blocking reagents were prepared with PBS at concentrations of 0.5%, 1.0%, 2.5%, and 5% (the commercial blocking solution, Sea Block, was diluted with PBS at 1:2.5, 1:5, 1:10, and 1:20). It was found that casein at a concentration of 1% was optimal i.e., background signal was well below 10%. The effect of the temperature during coating and blocking steps was also examined. When the plates were coated overnight at room temperature, a larger edge effect, i.e., higher standard deviation of signals in the outer wells was observed compared to plates which were coated at 4 °C. Blocking with casein was more efficient at room temperature than at 37 °C. Therefore, plates were routinely coated at 4 °C overnight followed by a blocking step at room temperature for 1 h. Characteristics of the ELISA Calibration Curve. Based on the optimized conditions, the diclofenac calibration curve was constructed in the concentration range of 0.01-10 µg/L. A typical curve is shown in Figure 1. The relative standard deviation of the measured absorbance for three replicates at each standard concentration was lower than 5%. The LOD at a signal-to-noise ratio (S/N) of 3 and IC50 were found to be 6 ng/L and 60 ng/L, respectively. The analytical working range (the linear part of the curve between 20% and 80% of inhibition) was about 20-400 ng/L. For 10 standard curves consecutively performed during 2 weeks the IC50 varied in the range of 40-80 ng/L (interassay CV 18.7%). Effects of Organic Solvents and Humic Acid on the ELISA. Usually ELISAs are performed in pure aqueous phase; however, in the last years an increasing number of reports have proved that ELISAs can be carried out at different concentrations of organic solvents. The tolerance to organic solvents might be an advantage for environmental analysis because many contaminants are hydrophobic and need to be extracted from the aqueous samples. The sodium salt of diclofenac is highly soluble in water; however, due to its very low concentration in environmental samples it may be necessary to perform preliminary enrichment of diclofenac by solid-phase extraction. If suitable antibodies would be VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3425

FIGURE 2. Effect of organic solvents on the dose-response curve for diclofenac calibrators: (2) methanol, (b) ethanol, (O) DMSO, (0) acetonitrile. Values are the mean (1 SD of three replicates. available also immunoaffinity chromatography could be applied. Further, natural matrix constituents such as humic material are known to interfere with some immunoassays. Therefore, the investigation of the effect of humic acids on the ELISA performance should be an integral part of assay development, if environmental samples should be analyzed. In the present paper, four organic solvents, methanol, ethanol, DMSO, and acetonitrile, were tested for their compatibility with the ELISA. For that, a series of diclofenac standard solutions were prepared with organic solvents at concentrations of 2-50% or with humic acid at 0.2-10 mg/ L. A common effect was observed both for organic solvents and humic acid, i.e., with increasing concentration in the sample the standard curve shifted gradually to higher diclofenac concentrations and, therefore, a loss in sensitivity was observed. For the organic solvents, this is presented in Figure 2. With increasing concentration of organic solvent the IC50 values increased at an extent depending on the kind of solvent. A significant increase of the IC50 was observed for methanol, ethanol, or DMSO at amounts higher than 20%, whereas even 10% acetonitrile leads already to a significant loss in sensitivity. If 10 times loss in sensitivity could be considered as acceptable for the diclofenac-ELISA when performed in organic solvent, the maximum concentrations of the solvents were estimated with 50% of methanol, 40% of ethanol and DMSO, and 20% of acetonitrile. This estimation might also be useful for the selection of an appropriate solvent which can be applied as an eluent if an immunoextraction method should be developed in future. According to Figure 3, the assay tolerated humic acid concentrations lower than 1 mg/L, i.e., only minor effect was discovered at diclofenac concentrations below 0.1 µg/L. With increasing concentration of humic acid in the sample the sigmoidal shape of the calibration curve got steadily lost, i.e., definite quantification becomes impossible. The intrinsic mechanisms of interference are unknown; however, we speculate that it may be a combination of different effects. Humic acids are a complex mixture of aromatic and aliphatic hydrocarbon structures that have attached functional groups (30). For example, unspecific binding of the analyte to humic material could lead to flattened curves as it was found with higher humic acid concentrations. On the contrary, recognition of partial humic acid structures by the antibody or even its unspecific binding through hydrogen bonding, nonpolar interactions, and ionic interactions could result in some lower optical densities i.e., overestimated analyte concentrations as was observed for diclofenac concentrations less than 0.1 µg/L and low humic acid concentrations. To summarize, real samples which contain humic acids should be analyzed 3426

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 15, 2003

FIGURE 3. Effect of a commercial humic acid (HA) on the doseresponse curve for diclofenac calibrators obtained with spiked water samples: (0) ultrapure water, (O) 0.2 mg/L HA, (b) 0.5 mg/L HA, (sideways solid triangle) 1 mg/L HA, (2) 2 mg/L HA, (1) 5 mg/L HA; (9) 10 mg/L HA. by the ELISA only after appropriate dilution. However, this may interfere with the probably very low concentration of diclofenac in surface water and the sensitivity of the assay, as will be discussed below. Specificity of the Antiserum. Antiserum specificity was tested by cross-reactivity (CR) measurements. CR values of the antiserum to 14 pharmaceuticals, 6 diclofenac metabolites, and 6 pesticides of similar chemical structure are presented in Table 1. The highest CR was found for 5-hydroxydiclofenac (100%), which is structurally very similar to the analyte. CR values for the other compounds mentioned above were well below 4% and, therefore, are negligible. Any substituents at the dichlorophenyl ring lead to a significant loss in antibody binding. This finding clearly demonstrates that, under the optimized conditions, the developed ELISA is highly specific for diclofenac. Analysis of Water Samples. Three types of water samples, i.e., tap water, surface water, and wastewater were collected and analyzed by the ELISA. Diclofenac was below the detection limit in the two tap water samples, whereas it was clearly detected in all three surface water samples at concentrations between 15 and 19 ng/L. Using GC-MS analysis, in one surface water sample diclofenac was not detected; it was found at 2 and 6 ng/L in the other two samples which was significantly lower compared to ELISA. To test the accuracy of the diclofenac-ELISA, the collected tap and surface water samples were fortified with diclofenac at concentration levels of 0.02, 0.05, and 0.2 µg/L. Both original and diluted (1:5, v/v) samples were used for fortification. For surface water, recovery at the individual spiking level was calculated after subtraction of the corresponding background concentration. All samples were run at least in triplicate, and relative standard deviations were lower than 7%. Data are summarized in Table 2. The average recovery of diclofenac in tap water was almost 100% regardless of whether original or diluted samples were considered. Results were different with surface water. While original samples gave overestimated diclofenac concentrations between 114 and 143%, recovery after dilution was between 103 and 108%. This is a clear indication of a matrix interference. All surface waters were odorless but slightly cloudy and with a tawny color. The latter could be caused by dissolved organic matter (DOM), synonymous with dissolved organic carbon (DOC). DOC is the organic carbon not retained on a 0.45-µm porosity membrane after filtration of a water sample. In the present investigation, DOC analysis revealed 3.76, 4.68, and 9.74 mg/L for water samples from lakes Ammersee, Wo¨rthsee, and river

TABLE 2. Diclofenac Concentrations (a) and Recoveries (b) Found in Fortified Tap and Surface Water by ELISA surface water Wo1 rthsee

Ammersee added diclofenac

tap water

1:5 dilution

original

Windach 1:5 dilution

original

original

1:5 dilution

(a) Concentration (µg/L) 0

0

0.02 0.05 0.2

0.018 0.052 0.185

0.018 (0.003)a 0.041 0.080 0.239

0.005 0.023 0.064 0.207

0.014 (nd)a,b 0.038 0.072 0.225

0.004 0.026 0.058 0.212

0.016 (0.002)a 0.042 0.086 0.320

0.003 0.023 0.057 0.235

(b) Recovery (%) 0.02 0.05 0.2 mean a

90 104 93 96

115 124 111 117

90 118 101 103

120 116 106 114

110 108 104 107

100 108 116 108

Number in parentheses is corresponding concentration determined by GC-MS. b Not determined.

TABLE 3. Diclofenac Analysis of Selected Samples from Five Different Sewage Treatment Plants (STPs) by ELISA

TABLE 4. Determination of Diclofenac in Wastewater Samples from 20 STPs by ELISA and GC-MS

diclofenac (µg/L) WTP dilution 1:200 1:100 1:50 1:25 1:10 mean SD CV (%) a

130 140 152 143

Munich I

Munich II

Weil

a

b

a

b

a

b

3.11 2.90 2.70 2.97 2.92 0.17 5.8

2.06 2.16 2.06 2.04 2.08 0.05 2.4

1.84 2.24 1.69 1.85 1.91 0.24 12.6

1.12 1.03 1.18 1.13 1.12 0.06 5.4

2.16 2.84 2.46 2.16 2.41 0.32 13.3

1.97 1.55 2.23 2.29 2.01 0.34 16.9

Influent.

b

Eching

a 8.14 8.33 8.48 8.45 oorc 8.35 0.15 1.8

Walleshausen

b

a

b

2.07 1.61 2.35 2.32 2.09 0.34 16.3

3.45 3.81 2.84 2.77 3.22 0.50 15.5

2.46 2.80 3.07 2.54 2.72 0.28 10.3

Effluent. c Out of quantification range.

Windach, respectively. Humic substances typically compose about 50-90% of the DOC of an average surface water (31). Concluding from this, humic substance concentration was at least between 2 and 5 mg/L in undiluted surface water samples analyzed in this study, and, therefore, observed matrix interference could easily be explained considering Figure 3. Highest overestimation of diclofenac was detected in the river water sample in accordance with the highest DOC concentration of the three surface water samples. In the 1:5 diluted samples the humic substances are still present at a concentration which could affect the diclofenac-ELISA. The observed slight overestimation could be an indication of that influence. As a conclusion, the overestimation of diclofenac concentration by the ELISA compared to GC-MS in surface water samples was most likely caused by the presence of humic substances. These results illustrate that for surface waters with concentrations of diclofenac typically in the very low ppt-range, i.e., close to the LOD of the ELISA, its applicability is mainly limited by the occurrence of humic substances in the samples. In the case of wastewater, all samples were filtered and diluted with pure water before analyzed by ELISA. To establish the lowest dilution level which does not show any matrix interference in the ELISA, both influent and effluent samples from five STPs were submitted to different dilutions of 1:10 to 1:100, except for water from STP Eching which was diluted 1:25 to 1:200. Diclofenac concentrations of the samples were calculated from the standard curve generated with pure water and run on the same plate and multiplied by the dilution factor. As shown in Table 3, diclofenac concentrations measured at different dilutions were very similar with coefficients of variation in the range of 1.8-16.9% (n ) 4), which not only demonstrated acceptable accuracy of the assay but also indicated that for wastewater samples 10 times dilution with pure water was completely sufficient to

STP

ELISA (µg/L)

GC-MS (µg/L)

Munich Munich Ib Munich IIa Munich IIb Weilb Echingb Weilb,c Echingb,c Walleshausenb,c Atterseeb Mattig Hainbachb Reichramigb Wolfgangsee-Ischlb Mittlere Gusenb Steyrb Trattnachtalb Bad Hallb Mondseeb Traunseeb Welser Heideb Bad Leonfeldenb Eberstallzellb Freistadtb Riedb X h ( SD (n ) 24) p-level from paired t-test: 0.0089

2.92 2.08 1.91 2.01 1.12 2.09 1.41 1.48 2.12 0.87 1.08 1.34 1.34 0.16 0.93 0.94 0.33 0.85 0.81 0.76 0.74 0.76 0.75 0.42 1.22 ( 0.66

2.0 0.79 2.4 0.81 1.15 1.26 1.18 0.86 1.13 1.41 0.88 1.17 1.35 0.19 0.30 0.90 0.14 0.87 0.70 0.72 0.58 0.77 0.43 0.48 0.93 ( 0.51

Ia

a

Influent.

b

Effluent. c Second sample collection.

eliminate possible matrix effect. As a conclusion, a diclofenac concentration as low as 60 ng/L (LOD) could be detected in wastewater by the developed ELISA. Average concentration was higher in influent wastewater (3.76 ( 2.34 µg/L) compared to effluents (2.00 ( 0.51 µg/L). If the significantly higher diclofenac concentration in the influent sample from STP Eching (8.35 µg/L) is not considered in the statistical calculation, the average concentration in the influents is somewhat lower (2.62 ( 0.50 µg/L; n ) 4). From these data an average degradation rate of about 25% can be calculated. To confirm the applicability of the ELISA for the determination of diclofenac in wastewater, 24 samples were analyzed both by ELISA and GC-MS. The results are summarized in Table 4. Regression analysis gave a good linear relationship of both methods (ELISA vs GC-MS: r ) 0.70, slope ) 0.90, intercept ) 0.37). However, the average concentration of 0.93 ( 0.51 µg/L as determined by GC-MS was about 25% lower than the corresponding concentration from ELISA (1.22 ( 0.66 µg/L). Several explanations for this finding would be possible. For the determination of diclofenac VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3427

TABLE 5. Analyses of Selected Wastewater Samples from Five STPs by ELISA and GC-MS after Different Treatment diclofenac (µg/L) ELISA

a

STPa

untreated sample

enzymatically treated sample

Munich Ia Munich Ib Munich IIa Munich IIb Weilc Echingc Walleshausenc X h ( SD

2.92 2.08 1.91 2.01 1.41 1.48 2.12 1.99 ( 0.46

1.98 1.39 1.24 0.84 1.12 1.23 1.6 1.34 ( 0.34

Influent.

b

Effluent. c Second sample collection.

d

9

2.28 1.45 1.27 0.93 1.03 0.95 1.6 1.34 ( 0.46

enzymatically treated sample npd npd npd npd 1.52 1.19 1.16 1.29( 0.16

acidified sample 2.0 0.79 2.4 0.81 1.18 0.86 1.13 1.35 ( 0.57

Not processed, np.

by ELISA, besides filtration water samples were only submitted to dilution. In contrast, GC-MS analysis required adjustment of sample pH to about pH 2.0 to yield undissociated analyte for adjacent trapping on C18 SPE cartridge. Higher ELISA values will be found if, beside the parent drug, also metabolites and/or drug conjugates or unknown chemicals are recognized by the antibody. According to Degen et al. (32), in man mainly hydroxylated metabolites in phase I are produced, which undergo conjugation reactions with endogenous metabolites in phase II reactions and are then excreted in the urine in the form of glucuronides accounting for roughly 60% of the dose administered. Further, 5-10% of diclofenac is eliminated from the body as parent molecule also conjugated to glucuronic acid. A detailed study on the CR of all metabolites and conjugates was impossible, because corresponding compounds were not commercially available. In this study, some metabolites were made available as a gift from NOVARTIS Co. Besides 5-hydroxydiclofenac, which showed identical binding to the antibody as the parent molecule, the other metabolites exhibited only a very low CR below 2%. However, ELISA analysis of wastewater samples which were submitted to enzymatic treatment with β-glucuronidase/arylsulfatase mixture yielded about 33% lower diclofenac concentrations (1.34 ( 0.34 µg/L) compared to the untreated samples (1.99 ( 0.46 µg/L) (Table 5). This might be an indirect indication for the presence of diclofenac glucuronide(s) and their binding by the raised antibodies. Because of structural similarity to the immunogen a higher binding affinity of the glucuronide which was formed as an ester from the parent drug compared to the nonconjugated chemical is most probable. In this case, the ELISA would yield higher diclofenac concentrations than GC-MS which detects only the nonconjugated drug. Glucuronides are known to be rather unstable, depending on the type of conjugate, i.e., esters are less stable than ethers. The question came up whether this diclofenac glucuronide, presumably contained in the samples, was covered by the GC-MS method. This would have been the case if the conjugate was cleaved after routine acidification of the samples. No corresponding data were found in the literature. Therefore, three effluent samples were analyzed by ELISA after different sample preparation, i.e., (a) untreated, (b) after enzymatic digestion, and (c) after acidification to yield pH 2.0 and readjustment to original pH value after 12 h. As a comparison, diclofenac was determined by GC-MS (Method B) according to (a) routine analysis and (b) after submitting samples to enzymatic digestion followed by adjustment to pH 2.0. Results were summarized in Table 5. Both enzymatically treated and acidified samples yielded almost identical mean diclofenac concentrations regardless of whether they were analyzed by ELISA or GC-MS. From this finding it can be concluded that the overestimation of diclofenac in wastewater by the ELISA was caused most likely by a higher affinity of the antibodies 3428

GC/MS acidified sample

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 15, 2003

to diclofenac glucuronide compared to the unchanged drug and not because of a significant contribution of other diclofenac metabolites or matrix constituents to the measuring signal. Both enzymatic cleavage and acidification yielded free diclofenac and, as a consequence, almost identical analytical results using immunochemical and gas chromatographic methods. However, the amount of diclofenac glucuronide present in wastewater should be rather low, i.e., much less than the free chemical. As a conclusion, the ELISA can be considered to be a simple, inexpensive, and accurate method for the determination of diclofenac in wastewater after rather simple sample preparation, i.e., filtration, acidification, and readjustment to neutral pH-value and dilution. In further studies, the diclofenac antibodies will also be used for the preparation of immunoaffinity supports to allow selective extraction of the pharmaceutical from environmental samples.

Acknowledgments We acknowledge the support by the Alexander von Humboldt-Foundation (Q.-Z.Z.). We thank R. Eisenmann (GCM.S.) and B. Apel (DOC analysis) for skillfull technical assistence. Further, the generous gift of diclofenac metabolites by Novartis Pharma AG (Basel, Switzerland) is gratefully acknowledged.

Literature Cited (1) Halling-Sørensen, B.; Nors Nielsen, S.; Lanzky, P. F.; Ingersiev, F.; Holten Lu ¨ tzhøft, H. C.; Jørgensen, S. E. Chemosphere 1998, 36, 357-393. (2) Ternes, T. A. Water Res. 1998, 32, 3245-3260. (3) Jones, O. A. H.; Voulvoulis, N.; Lester, J. N. Environ. Technol. 2001, 22, 1383-1394. (4) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; Barber, L. B.; Buxton, H. T. Environ. Sci. Technol. 2002, 36, 1202-1211. (5) Dietrich, D. R.; Webb, S. F.; Petry T. Toxicol. Lett. 2002, 131, 1-3. (6) Heberer, T.; Schmidt-Ba¨umler, K.; Stan, H.-J. Acta Hydrochim. Hydrobiol. 1988, 26, 272-278. (7) Seiler, R. L.; Zaugg, S. D.; Thomas, J. M.; Howcroft, D. L. Ground Water 1999, 37, 405-410. (8) Sacher, F.; Lange, F. T.; Brauch, H.-J.; Blankenhorn, I. J. Chromatogr. A 2001, 938, 199-210. (9) Moeder, M.; Schrader, S.; Winkler, M.; Popp, P. J. Chromatogr. A 2000, 873, 95-106. (10) Ternes, T. A. Trac-Trend Anal. Chem. 2001, 20, 419-434. (11) Reddersen, K.; Heberer, T.; Du ¨ nnbier, U. Chemosphere 2002, 49, 539-544. (12) Schulman, L. J.; Sargent, E. V.; Naumann, B. D.; Faria E. C.; Dolan, D. G.; Wargo, J. P. Hum. Ecol. Risk. Assess. 2002, 8, 657680. (13) Seiler, J. P. Toxicol. Lett. 2002, 131, 105-115. (14) Straub, J. O. Toxicol. Lett. 2002, 135, 231-237. (15) Koschorreck, J.; Koch, C.; Ro¨nnefahrt, I. Toxicol. Lett. 2002, 131, 117-124.

(16) Heberer, T. Toxicol. Lett. 2002, 131, 5-17. (17) Koutsouba, V.; Heberer, T.; Fuhrmann, B.; Schmidt-Baumler, K.; Tsipi, D.; Hiskia, A. Chemosphere 2003, 51, 69-75. (18) O ¨ llers, S.; Singer, H. P.; Fa¨ssler, P.; Mu ¨ ller, S. R. J. Chromatogr. A 2001, 911, 225-234. (19) Ravina, M.; Campanella, L.; Kiwi, J. Water Res. 2002, 36, 35533560. (20) Jux, U.; Baginski, R. M.; Arnold, H.-G.; Kro¨nke, M.; Seng, P. N. Int. J. Hyg. Environ. Health 2002, 205, 393-398. (21) Ahrer, W.; Scherwenk, E.; Buchberger, W. J. Chromatogr. A 2001, 919, 69-78. (22) Sherry, J. P. Crit. Rev. Anal. Chem. 1993, 23, 217-300. (23) Dankwardt, A. In Encyclopedia of Analytical Chemistry; Meyers, R. A., Ed.; John Wiley & Sons Ltd.: Chichester, 2000; pp 63186344. (24) Aherne, G. W.; English, J.; Marks, V. Ecotox. Environ. Safe. 1985, 9, 79-83. (25) Aherne, G. W.; Hardcastle, A.; Nield, A. H. J. Pharm. Pharmacol. 1990, 42, 741-742.

(26) Shore, L. S.; Gurevitz, M.; Shemesh, M. Bull. Environ. Contam. Toxicol. 1993, 51, 361-366. (27) Snyder, S. A.; Keith, T. L.; Verbrugge, D. A.; Snyder, E. M.; Gross, T. S.; Kannan, K.; Giesy, J. P. Environ. Sci. Technol. 1999, 33, 2814-2820. (28) Huang, C. H.; Sedlak, D. L. Environ. Toxicol. Chem. 2001, 20, 133-139. (29) Valentini, F.; Compagnone, D.; Gentili, A.; Palleschi, G. Analyst 2002, 127, 1333-1337. (30) Leenheer, J. A.; Croue´, J.-P. Environ. Sci. Technol. 2003, 37, 18A26A. (31) Janos, P. J. Chromatogr. A 2003, 983, 1-18. (32) Degen, P.; Faigle, J. W.; Ge´rardin, A.; Moppert, J.; Sallmann A.; Schmid, K.; Schweizer, A.; Sulc, M.; Theobald, W.; Wagner, J. Scand. J. Rheumatol. 1978 Suppl. 22, 17-29.

Received for review March 5, 2003. Revised manuscript received May 20, 2003. Accepted May 20, 2003. ES0341945

VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3429