Environ. Sci. Technol. 2005, 39, 5209-5218
Environmental Fate of Pharmaceuticals in Water/Sediment Systems DIRK LO ¨ FFLER,† JO ¨ RG RO ¨ MBKE,‡ MICHAEL MELLER,‡ AND T H O M A S A . T E R N E S * ,† German Federal Institute of Hydrology, P.O. Box 20 02 53, D-56002 Koblenz, Germany, and ECT Oekotoxikologie GmbH, Bo¨ttgerstrasse 2-14, D-65439 Flo¨rsheim am Main, Germany
In recent years there has been growing interest on the occurrence and the fate of pharmaceuticals in the aquatic environment. Nevertheless, few data are available covering the fate of the pharmaceuticals in the water/ sediment compartment. In this study, the environmental fate of 10 selected pharmaceuticals and pharmaceutical metabolites was investigated in water/sediment systems including both the analysis of water and sediment. The experiments covered the application of four 14C-labeled pharmaceuticals (diazepam, ibuprofen, iopromide, and paracetamol) for which radio-TLC analysis was used as well as six nonlabeled compounds (carbamazepine, clofibric acid, 10,11-dihydro-10,11-dihydroxycarbamazepine, 2-hydroxyibuprofen, ivermectin, and oxazepam), which were analyzed via LC-tandem MS. Ibuprofen, 2-hydroxyibuprofen, and paracetamol displayed a low persistence with DT50 values in the water/sediment system e 20 d. The sediment played a key role in the elimination of paracetamol due to the rapid and extensive formation of bound residues. A moderate persistence was found for ivermectin and oxazepam with DT50 values of 15 and 54 d, respectively. Iopromide, for which no corresponding DT50 values could be calculated, also exhibited a moderate persistence and was transformed into at least four transformation products. For diazepam, carbamazepine, 10,11-dihydro-10,11-dihydroxycarbamazepine, and clofibric acid, system DT90 values of > 365 d were found, which exhibit their high persistence in the water/sediment system. An elevated level of sorption onto the sediment was observed for ivermectin, diazepam, oxazepam, and carbamazepine. Respective KOC values calculated from the experimental data ranged from 1172 L‚kg-1 for ivermectin down to 83 L‚kg-1 for carbamazepine.
Introduction Pharmaceuticals are applied annually in amounts up to several hundred tons per individual compound in Germany. After their use, they are excreted unchanged and/or as metabolites with feces and urine; therefore, they are introduced directly into wastewater. The extensive investigation of the occurrence of pharmaceuticals in the environment * Corresponding author phone: +49 261 1306 5560; fax: +49 261 1306 5363; e-mail:
[email protected]. † German Federal Institute of Hydrology. ‡ ECT Oekotoxikologie GmbH. 10.1021/es0484146 CCC: $30.25 Published on Web 06/09/2005
2005 American Chemical Society
began in the 1990s, when the first analytical methods were developed allowing for the determination of pharmaceuticals in aqueous matrices such as wastewater, surface water, groundwater, and drinking water. These studies showed that pharmaceutical residues are not totally eliminated during sewage treatment and occur at a similar level as several pesticides, up to the micrograms per liter range, in rivers, streams, lakes, and even in groundwater (1-3). With increasing knowledge of their environmental occurrence, growing interest was focused on their environmental fate. Several groups have investigated the elimination of the pharmaceuticals during sewage treatment (2, 4-6), in surface water, biofilms (7, 8), in soils (9-12) and soil columns (13, 14), and during groundwater recharge and riverbank filtration (15, 16). Various groups also reported the photodegradability of pharmaceuticals (17) and investigated the fate of several pharmaceuticals in surface water/sediment microcosms (18, 19). However, due to the lack of analytical methods almost no data are available about the occurrence of pharmaceuticals in freshwater sediments and their role in the elimination of the pharmaceuticals (20). For instance, it is widely unclear whether there is temporal or spatial loss of pharmaceuticals in surface waters, and whether degradation or sorption onto the sediment and suspended matter are the primary fate processes. Additionally, data on the occurrence of pharmaceutical residues in the environment are mostly limited to the parent compounds, although metabolites of pharmaceuticals are often released into the environment in higher quantities than their parent compounds (21, 22). Therefore, a study was initiated to determine the environmental fate of six human and veterinary pharmaceuticals in the water/sediment system (Table 1). The selection of the pharmaceuticals considered their amounts applied for medical purposes (23-25) and their occurrence in the environment (2, 3). It also covered their clinical potency and their pharmacological and physicochemical properties. Furthermore, four major metabolites were included to allow a comparison of the environmental fate of the parent compounds and the metabolites. Our major objective was to investigate the (bio)degradability of the selected pharmaceuticals in water/sediment systems using sensitive LCtandem MS and radiotracer techniques for analysis. To minimize the influence of sorption in the experiments to a reasonable extent, a sediment was chosen with a relatively low but still common Corg.
Materials and Methods Experimental Setup. Sediment and water were taken from the Wickerbach creek in Flo¨rsheim (close to Frankfurt/Main, Southwest Germany) at a sampling site that is located close to the source of the creek. The site was selected due to the low but still environmentally relevant organic carbon content (Corg) (26) of the sediment (Table 2) and the absence of sewage treatment plant discharges (27). Since, the main purpose of the current study was to investigate the biodegradability of the selected pharmaceuticals, the low Corg minimized the influence of sorption and partitioning processes. Given that the sampling depth was restricted to 5 cm, the sediment taken was mainly under oxic conditions. The sediment was wet-sieved (2 mm mesh) and homogenized. Combined water and sediment were stored at a ratio of (3:1) at 4 °C in the dark for a maximum of 28 d. Characteristics of water and sediment are given in Table 2. The experiments were performed similar to OECD Guideline 308 (28) and consisted of 500 mL amber glass flasks VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5209
TABLE 1. Structures and Labeling Position of the Pharmaceuticals Investigated
the sampling times, the entire water and sediment phase and the soda lime of each test system were taken, separated, and stored at -20 °C until analysis. Materials and Instrumentation for Chemical Analysis. The following reference compounds (CAS Registry Numbers are given in brackets) and radiopharmaceuticals were used in the current study: carbamazepine [298-46-4], clofibric acid [882-09-7], diazepam [439-14-5], ibuprofen [15687-271], ivermectin [71827-03-7], oxazepam [604-75-1] (Sigma, Deisenhofen, Germany); 10,11-dihydro-10,11-dihydroxycarbamazepine (CBZ-diol), 2-hydroxyibuprofen [51146-55-5] (µMol, Luckenwalde, Germany); iopromide [73334-07-3], desmethoxyiopromide [76350-28-2], 5-amino-2,4,6-triiodoisophthalic acid (ATI) [35453-19-1], ATI-(2,3-dihydroxypropyl)-amide (ATH) [111453-32-8], desmethoxyacetyliopromide (DAMI) [154361-51-0] (courtesy of Schering, Berlin, Germany); abamectin [71751-41-2], fenoprop [93-72-1] (Riedel-de-Haen, Seelze Germany); 10,11-dihydrocarbamazepine [3564-73-6] (Alltech, USA); [14C]diazepam [2-14C], radiochemical purity 99.8%, specific activity 7.22 MBq‚mg-1, (Amersham Pharmacia Biotech UK Limited, Little Chalfont, UK); [14C]ibuprofen [carboxyl-14C], radiochemical purity > 99%, specific activity 8.97 MBq‚mg-1 (American Radiolabeled Chemicals Inc., St. Louis, MO, US); [14C]-4-paracetamol [ringUL-14C], radiochemical purity > 99%, specific activity 1.54 MBq‚mg-1 (Sigma-Aldrich, Steinheim, Germany); [14C]iopromide [ring-14C], radiochemical purity > 98%, specific activity 1.86 MBq‚mg-1 (courtesy of Schering AG, Berlin, Germany). Other chemicals and solvents used were acetic acid (100%, p.a.), amyl alcohol, sulfuric acid (suprapur), ammonia (25%, p.a.), ammonium acetate (p.a.), KH2PO4, and K2HPO4 (Merck, Darmstadt, Germany), and acetone, acetonitrile, chloroform, ethyl acetate, methanol, 2-methylpropanol, n-hexane, and toluene (all suprasolv grade, Merck, Darmstadt, Germany).
(Schott, Mainz, Germany) filled with 200 g of sediment and 300 mL of creek water (Figure 1). CO2 traps filled with 30 g of granulated soda lime were sealed tightly to the flasks. Furthermore, the vessels were wrapped with aluminum foil to minimize photochemical reactions. Prior to spiking with pharmaceuticals, the water/sediment systems were equilibrated under test conditions (20 ( 2 °C) for at least 7 d. During the equilibration and the study period, the test systems were slowly shaken avoiding disturbance of the separation between the sediment layer and overlying water as well as re-suspension of sediment fines. The pharmaceuticals were spiked into the water phase at the beginning of the study period of 100 d. The concentrations were chosen to enable for detection limits between 1 and 2.5% of the initial analyte concentration. For iopromide and diazepam, which were the first compounds investigated, a spiking level of 500 ng‚g-1 was chosen. Since the necessary detection limits were attained easily, the spiking level of the remaining compounds was reduced to 100 ng‚g-1. For each sampling time two water/sediment vessels were terminated and sampled in parallel. Samples were taken immediately after spiking and after 0.25, 1, 2, 7, 14, 28, 56, and 100 d. At 5210
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005
Analysis of Pharmaceuticals in Sediment and Water with LC-Tandem MS. The analysis of the nonlabeled compounds was carried out using methods that allowed for an analysis of both the sediment and the water samples (Figure 2). Sediment samples of 50 g were sequentially extracted (29) (see Table 3) for 15 min with 45 mL of organic solvents under ultrasonic treatment (USE), and the organic phase of the combined extracts was removed by evaporation. The extracts were then diluted with groundwater prior to solid phase extraction (SPE), which was used as a cleanup step for the sediment analysis as well as an enrichment procedure for water samples. Sample preparation and the LC-tandem MS measurements were carried out according to methods published recently (Table 4) (29, 30). A short summary of recovery data is shown in Table 5. The limits of quantification (LOQ) in sediment and water samples for all compounds were e1% of the initially applied concentration (C0). Determination of the Total Radioactivity in Solid and Aqueous Sample Material. The analysis of samples containing the radiolabeled compounds was divided into two major steps. At first, the total radioactivity in the various system compartments was determined. Radioactivity in aqueous samples was measured directly by liquid scintillation counting (LSC) using a Tricarb 2500 TR (Canberra Packard, Dreieich, Germany). 14CO2 sorbed to the soda lime trap material was transferred quantitatively into the liquid phase as was the radioactivity present in the sediment, which was combusted in triplicate in an automatic sample oxidizer TriCarb 307 (Canberra Packard, Dreieich, Germany) prior to LSC measurement of these samples. A balance of the radioactivity was then conducted, allowing for a comparison of the quantity of radioactivity initially present in the various compartments of the test system versus the quantity found in the test system during the study period.
TABLE 2. Characteristics of Water and Sediment parameter pH redox potential (mV) organic matter (% dry wt) Corg (% dry wt) microbial biomass (µg of C per g dry sediment) water content (%) clay (%) silt (%) sand (%) soil type
sediment 7.7 269 2.4 1.4 41 20 9.9 12.6 77.5 loamy sand
FIGURE 1. Water/sediment system.
FIGURE 2. Scheme of the analytical method with HPLC-tandem MS detection. Chemical Analysis of 14C-Labeled Pharmaceuticals in Sediment and Water Samples. The second step was the chemical analysis of the radiolabeled compounds in the different system compartments. An USE was conducted as described for the nonlabeled analytes (see Table 3). The sediment extracts were reduced to dryness by evaporation. Water samples were lyophilized. All extracts were dissolved in 1.5 mL of methanol and were then filtered using 0.22 µm PTFE syringe filters. The recoveries of all 14C-labeled analytes attained in lyophilization of spiked surface water and the extraction of spiked sediment samples were always >85% (see Table 5). For radio-TLC analysis, silica gel TLC plates, Merck (Darmstadt, Germany) 20 × 20 cm, 60 µm, with a 254 nm fluorescence coating and a preconditioning zone were used for all chromatographic separations. The compositions of the mobile phases for diazepam were chloroform/acetone (85:15, v/v) (31); for ibuprofen: chloroform/methanol/acetic acid (85:15:0.1, v/v); for iopromide: 2-methyl-1-propanol/
parameter
water
pH redox potential (mV) TOC (mg of C‚L-1) oxygen (mg‚L-1) nitrate (mg‚L-1) phosphate (mg‚L-1) hardness (mg of CaCO3 per L) conductivity (µS‚cm-1)
8.5 382 4.7 7.8 6.2 0.9 222 515
2-propanol/ammonia 25% (50:30:20, v/v) (18); and for paracetamol: toluene/methanol/acetic acid (90:16:8, v/v). The TLC plates were scanned for 10-30 min with an autoradiograph (Berthold, Wildbad, Germany) at an amplifier voltage of 900 V. For the determination of total radioactivity and the chemical analysis of 14C-labeled pharmaceuticals in water and sediment, LOQs were obtained that correspond to e1% of the initially applied quantity (C0). The LOQs for the TLC analysis of water samples were e2.5% of C0, whereas for sediment samples, LOQs were e1% of C0. Estimation of DT50/90 Values. Dissipation times (DT values) of the compounds investigated were estimated for the water compartment and the entire water/sediment system considering first-order kinetics with the elimination rate constants kelim. In the cases where microbial adaptation had to be taken into account, the elimination rate constants were estimated as rates after the onset of the transformation process, taking the end of the lag phase as time zero. All regression coefficients obtained were >0.8. The dissipation times were calculated as DT50 ) ln(2)/kelim and DT90 ) ln(10)/kelim. In those few cases (e.g., ibuprofen, CBZ-OH) where no fit could be achieved, the DT values were estimated graphically. Considering that the duration of the study was 100 d, numerical dissipation values are given only if the calculated number was not higher than 1 yr. All values estimated as being between 1 and 2 yr are given as “>365 d” and those estimated as even higher are given as “.365 d”. Determination of Kd Values. The measured concentrations were used for a rough calculation of KOC and Kd values, determining the distribution pattern between water and sediment. For each of the six compounds with a complete data set for water and sediment, the maximum values were calculated. These maximum values were usually determined after 30 d or later and should hence represent the value closest to distribution equilibrium. Additionally, KOC values were predicted based on the correlation between the octanol/water distribution coefficient KOW and the sorption coefficient KOC (32) thus allowing a comparison of the data. To this end, three semiempirical methods were applied following the approaches recently used by Stuer-Lauridsen et al. (23) KOC ) 0.41 × KOW; Karickoff (33) log KOC ) 0.989 × log KOW - 0.346; and Gerstl (32) log KOC ) 0.679 log KOW + 0.663. Since clofibric acid was deprotonated under the pH conditions in the sediment, the DOW (DOW ) KOW/(1 + 10pH-pKa) (23) was used in the calculation instead of the KOW.
Results and Discussion The chemical and radiochemical analyses were crucial in the fate investigation of the pharmaceuticals, providing data for the subsequent evaluation. The experimental setup of the water/sediment system allowed for a widely parallel development of the independent test vessels under oxic conditions over a period of 100 d. All main physicochemical VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5211
TABLE 3. Composition of the Solvents for the Sequential Use of 14C-Labeled and Nonlabeled Pharmaceuticals composition of solvent in extraction sequences analytes
1st
2nd
3rd
4th
clofibric acid, 2-hydroxy-ibuprofen, ibuprofen, acetone/acetic acid 20:1 (v/v) ethyl acetate ethyl acetate ethyl acetate ivermectin carbamazepine, CBZ-diol, diazepam, oxazepam methanol/ethyl acetate 50:50 (v/v) ethyl acetate ethyl acetate ethyl acetate paracetamol methanol methanol acetone acetone
TABLE 4. Conditions for the SPE and the LC-tandem MS Detection
SPE material solvent for the elution of SPE material HPLC column ionization mode
clofibric acid 2-hydroxyibuprofen
ivermectin
carbamazepine CBZ-diol oxazepam
Oasis MCX 60 mg, Waters 4 × 1 mL acetone 125 × 3 mm Lichrosphere RP18 Merck APCI (-)
Lichrolute EN 250 mg, Merck 4 × 1 mL ethyl acetate 125 × 3 mm Lichrosphere RP18 Merck APCI (-)
RP18ec 500 mg, ICT 4 × 1 mL methanol 125 × 3 mm Lichrosphere RP18ec Merck ESI (+)
TABLE 5. Recovery Rates of Analytical Methods and Mass Balance over the Entire Test Duration for the Compounds Investigated
compound carbamazepine CBZ-diol clofibric acid [14C]diazepam oxazepam [14C]ibuprofen 2-hydroxyibuprofen [14C]iopromide ivermectin [14C]paracetamol
method recovery rates (n ) 3) for surface sediment water (%) (%)
C/C0 recovered in test systems after 100 d (n ) 2)a (%)
111 ( 11 110 ( 12 129 ( 6 95 ( 2 116 ( 11 90 ( 2 111 ( 6 91 ( 9 94 ( 16 87 ( 4
83 ( 3 33 ( 10 55 ( 3 89 ( 6 (92 ( 6)b 33 ( 1 0c (90 ( 1)b 0 0c (89 ( 5) 18 ( 3 0c (75 ( 2)b
102 ( 2 41 ( 3 70 ( 5 98 ( 9 115 ( 8 92 ( 4 40 ( 4 -c 102 ( 20 95 ( 3
a Values with absolute deviation (n ) 2); deviations from 100% were caused by transformation processes or the formation of bound residues. b Total radioactivity in water, sediment, and CO trap. c Sediment not 2 included in analysis.
parameters, such as pH and the redox potential of water and sediment, the oxygen concentration in the water phase and others displayed no significant variation over the study period. Time-concentration curves for the selected pharmaceuticals are shown in Figure 3, and further data on their behavior (e.g., DT values) are provided in Table 6. Since the pharmaceuticals were spiked into the water, their elimination from the water phase was a competitive process of sorption onto the sediment and transformation. It should be noted that only for radiolabeled substances the elimination can be differentiated into transformation, mineralization, and the formation of bound residues, whereas this is not possible for nonlabeled compounds. Carbamazepine and 10,11-Dihydro-10,11-dihydroxycarbamazepine. The antiepileptic carbamazepine was highly recalcitrant toward elimination in the water/sediment, resulting in a DT50 value of 328 d and a DT50 value of 47 d for the water compartment only. The latter value is comparable to a corresponding DT50 of 82 d reported by Lam et al. (19) for carbamazepine in outdoor microcosm studies. Furthermore, this high stability is consistent with the low degradability of carbamazepine during sewage treatment, its ubiquitous presence in surface waters (2, 21, 34), and its low elimination during subsoil passage (16). Carbamazepine also displayed a moderate affinity for the sediment 5212
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005
(Figure 3), which is consistent with its lipophilicity (log KOW ) 2.25) (24) and its presence in a noncharged form under environmental conditions (estimated pKa values of 13.9 for the deprotonation (24) and 95%). [14C]Diazepam was constantly partitioned
TABLE 6. Dissipation Times, Extent of Mineralization and Sediment Sorption, and Persistence Classification of Pharmaceuticals Investigated in Water/Sediment Systems
[14C]paracetamol [14C]ibuprofen 2-hydroxyibuprofen clofibric acid [14C]diazepam oxazepam carbamazepine CBZ-diol [14C]iopromide ivermectin
DT50 (d)a
DT90 (d)a
3.1 ( 0.2 (3.1 ( 0.2) 365)c >365c (>365)c (96 ( 13) 48 ( 5 (10 ( 1)
mineralization (% C0)
sorption to sediment (% C0)
19
64-53b
low
77
17-9
low
nd
