Research Screening of Human Antibiotic Substances and Determination of Weekly Mass Flows in Five Sewage Treatment Plants in Sweden RICHARD H. LINDBERG,* PATRIK WENNBERG, MAGNUS I. JOHANSSON, MATS TYSKLIND, AND BARBRO A. V. ANDERSSON Environmental Chemistry, Umeå University, SE-901 87 Umea° , Sweden
Twelve antibiotic substances for human use, including trimethoprim and representatives of the fluoroquinolone (FQ), sulfonamide (SA), penicillin (PE), cephalosporin (CE), nitroimidazole (NI), tetracycline (TC), and macrolide (MA) groups, were subjected to a screening study at five Swedish sewage treatment plants (STPs) during one week in 2002 and one week in 2003. The analytes were extracted from raw sewage water, final effluent, and sludge by solid-phase extraction (SPE) or liquid-solid extraction (as appropriate) and then identified and quantified by liquid chromatography/tandem mass spectrometry. The most frequently detected antibiotics in the matrices considered in this study were norfloxacin, ofloxacin, ciprofloxacin, trimethoprim, sulfamethoxazole, and doxycycline. The other analytes were only detected in a few samples. Analysis of the weekly mass flows through each STP showed that FQs were partly eliminated from the water during sewage water treatment and the highest amounts of these substances were found in sludge. Sulfamethoxazole and trimethoprim were mainly found in raw sewage water and final effluent, but these substances had balancing mass flows, indicating that they too can withstand sewage water treatment. The mass flow patterns for doxycycline were more complex, with high amounts occurring in sludge in some cases, suggesting that the behavior of this analyte may be more strongly influenced by the treatment process and other variables at individual STPs. The environmental load (the sum of the amounts in the final effluent and sludge) normalized to the number of inhabitants in the catchment area of each investigated STP compared with theoretical predictions based on consumption data (in parentheses) showed good correlations: norfloxacin, 0.8 (0.9); ofloxacin, 0.3 (0.2); ciprofloxacin, 1.3 (3.5); sulfamethoxazole, 0.2 (0.4); trimethoprim, 1.1 (1.0); and doxycycline, 0.7 (0.4) mg per person per week. The results show that reasonably accurate predictions of environmental load of these antibiotics can be time-effectively derived from consumption data without additional measurements. * Corresponding author phone: +46 90 786 66 69; fax: +46 90 12 81 33; e-mail:
[email protected]. 10.1021/es048143z CCC: $30.25 Published on Web 04/02/2005
2005 American Chemical Society
Introduction Antibiotics are used in human medicine to prevent or to treat microbial infections (1), and following medication, substantial amounts of these substances can be excreted unchanged in urine and feces (2). Significant amounts may also originate from the disposal of unused medicine. This intensive use of antibiotics has led to concerns about the impact these substances may have on the environment due to the risk of developing and/or maintaining bacteria resistant to antibiotics. For instance, Ash and Iverson (3) identified sulfonamide- and trimethoprim-resistant bacteria in rivers in the U.S., and the continuous low concentrations of these substances in the rivers might maintain resistance in these organisms. In 2002 the government of Sweden commissioned the Swedish Medical Products Agency (MPA) to investigate the environmental impact of pharmaceuticals and personal care products (4). In 2004 the Swedish MPA concluded in their final report that the knowledge about pharmaceuticals in the environment has to increase since relevant and available chemical/biological data in general are scarce. This includes an increasing knowledge about the actual presence of these substances but also development and validatation of models that predict pharmaceutical concentrations in the environment. The main route for transportation of antibiotics used in human medicine to the environment is via sewage treatment plants (STPs) where they may be eliminated. Thus, it is important to monitor both raw sewage water and the effluent and sludge leaving STPs. Antibiotics have been detected not only in sewage water (5-7) and hospital wastewater (8, 9) but also in ground and river water (10, 11), sludge (12, 13), and soil and manure (14-16). The reported concentrations of antibiotics in water range from subnanogram or low nanogram per liter levels (in ground and river water) to high microgram per liter levels (hospital effluents), with intermediate concentrations in STP effluents. For sludge, reported levels are in the milligrams per kilogram dry weight (dw) range. Commonly used methods for analysis often include solid-phase extraction (SPE) for enrichment and cleanup of aqueous samples, extraction from solid matrices followed by SPE, and finally liquid chromatography/tandem mass spectrometry (LC/MS/MS) for separation and detection (6, 9, 12, 17). In Sweden, fluoroquinolones (FQ), sulfonamides (SA), trimethoprim, penicillins (PE), cephalosporins (CE), nitroimidazoles (NI), tetracyclines (TC), and macrolides (MA) represents groups of antibiotics widely used. The amounts consumed in Sweden 2002 of each individual antibiotic substance included in this study, in tons per year, were as follows: norfloxacin (FQ), 1.4; ofloxacin (FQ), 0.1; ciprofloxacin (FQ), 3.6; sulfamethoxazole (SA), 1.0; trimethoprim, 1.0; phenoxymethylpenicillin or PcV (PE), 27.5; amoxicillin (PE), 3.9; ampicillin (PE), 0.06; cefadroxil (CE), 2.4; metronidazole (NI), 4.3; doxycycline (TC), 0.5; and erythromycin (MA), 1.6 (18). To conduct meaningful environmental risk assessments of human antibiotics, knowledge about the quantities of these substances in effluents and sludge is required. By examining a range of STPs, differing with respect to their catchment area, size, and treatment technique(s), it should be possible to obtain reasonably comprehensive estimates of the environmental load. In this screening study we focused on the substances previously mentioned due to documented stability (19). VOL. 39, NO. 10, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3421
FIGURE 1. Locations of the sewage treatment plants investigated in this study. Furthermore, they represent different groups of antibiotics and relatively large amounts of them are consumed in Sweden. Samples from five STPs scattered across Sweden were collected during one week in the summer of 2002 and one week in the winter of 2003 with the following objectives: to quantify 12 antibiotics in STP raw sewage water, final effluent, and sludge; to determine weekly mass flows of the analytes; to investigate the mass flows when normalized to inhabitants; and finally, to obtain measurement values for comparison with predicted national environmental loads of individual antibiotics. Pharmaceuticals in the environment, including antibiotics, have already been studied on a national scale in a few countries, for example, by Kolpin et al. (11), Calamari et al. (20), and Miao et al. (21), but to our knowledge this is the first study in Sweden of this kind.
Experimental Section Sample Sites and Collection. Sewage water and sludge samples were collected at five different sewage treatment plants in Sweden during August 2002 and February 2003. The following plants were included in the investigation: Stockholm, Henriksdal (STP A); Gothenburg, Ryaverken (STP B); Umeå (STP C); Kalmar (STP D); and Floda (STP E). Their locations can be seen in Figure 1. The water treatment includes chemical removal of phosphorus, primary clarification, active sludge treatment with nitrogen removal (except Umeå and Floda), and secondary clarification. As for sludge, the final products from all STPs are anaerobic digested sludge except at Floda (aerobic stabilized sludge). In Table 1, descriptive data of each STP and its catchment area are shown, and in Figure 2, their general sewage water treatment 3422
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 10, 2005
processes are outlined. Input of antibiotics from industry and hospitals is unknown; however, in Floda this is assumed to be insignificant as no such sources have been identified. In both 2002 and 2003, raw sewage and final effluents were flow-proportionally sampled, continuously during 1 week by use of existing automatic sampling equipment at each STP. Final sample volumes were 2 liters and they were stored in glass or, in a few cases, plastic (high-density polyethylene) bottles. In 2002, sludge samples (0.1 kg/subsample) were collected over 1 week and pooled (n ) 5) to a final weight of approximately 0.5 kg. In addition, nine subsamples were taken at Ryaverken STP for a homogeneity study. Samples, both water and sludge, for field fortification were also taken at Ryaverken STP. In 2003, grab samples (approximately 0.5 kg) from each STP were collected. All sludge samples were the final products from the respective STPs, that is, anaerobic digested sludge except at Floda STP, where aerobic stabilization was used. All samples (sewage water and sludge) were kept in the dark at -18 °C during transport and storage until analysis, approximately 2 weeks. Chemicals. Norfloxacin, ofloxacin, enrofloxacin, diaveridine, cefadroxil, and cephalexin were purchased from Sigma Aldrich (Stockholm, Sweden). Ciprofloxacin, sulfamethazine, and demeclocycline were bought from ICN Biochemicals (Irvine, CA). Sulfamethoxazole, trimethoprim, ampicillin, amoxicillin, metronidazole, and doxycyline were obtained from Duchefa (Haarlem, The Netherlands). 2-Methyl-5nitroimidazole was obtained from Arcos Organics (Geel, Belgium). Penicillin V (PcV) and erythromycin were purchased from Riedel-de Hae¨n (Seelze, Germany). Formic acid and methanol (HPLC-grade) were purchased from J. T. Baker (Deventer, The Netherlands), and acetonitrile (HPLC-grade) was bought from Fischer Chemicals (Zurich, Switzerland). Triethylamine (TEA), dipotassium hydrogen phosphate, acetic acid, and sulfuric acid were obtained from Merck (Darmstadt, Germany). The purified water (resistivity 18.2 MΩ‚cm) used was prepared by passing water through an ELGA Maxima HPLC ultrapure water system (ELGA, High Wycombe Bucks, England), equipped with a UV radiation source. Standard stock solutions of individual antibiotics were prepared in water, except for sulfamethoxazole and sulfamethazine, which were dissolved in methanol. A maximum of 5% (v/v) acetic acid was used to increase the solubility of nitroimidazoles, trimethoprim, diaveridine, and the fluoroquinolones (except for ciprofloxacin). The stock solutions were kept dark in the freezer at -18 °C and were remade every third months. Stability of standard mixtures has been investigated elsewhere (9). Sample Preparation: Sewage Water. The frozen sewage water samples were thawed and filtered through 0.45 µm MF-membrane filters (Millipore, Sundbyberg, Sweden) before acidification to pH 3 with sulfuric acid. The internal standards enrofloxacin, sulfamethazine, diaveridine, cephalexin, 2-methyl-5-nitroimidazole, and demeclocycline (500 ng of each) were added to each sample. An aliquot (1 L) of sample was withdrawn and subjected to extraction. The ENV+ (6 mL, 200 mg) solid sorbent columns (IST, Hengoed, Mid Glamorgan, U.K.) used for the solid-phase extraction (SPE) were conditioned with 5.0 mL of methanol, 5.0 mL of 50% methanol in water, and 5.0 mL of water, pH 3. The samples were applied to the SPE columns at a flow rate of 3 mL/min. Water (pH 3) (5.0 mL) was added in order to wash the sorbent before drying it with air for approximately 1 h. To elute the analytes, 2.0 mL of methanol followed by 5.0 mL of 5% TEA in methanol was used. The eluate was collected in 10 mL glass vials, evaporated to 20 µL in air, and dissolved in 5% acetonitrile in water to a final volume of 1.0 mL.
TABLE 1. Descriptive Data of the Investigated Sewage Treatment Plants city Stockholm Gothenburg Umeå Kalmar Floda
surf. inhab STP ID serveda × 103 hospitals industryb water,c %
influent N in/out, Fe,g poly,h hydi t r, sol.j t r, dry h days sludgee weight,f, % BOD7, mg/L mg/L mg/L g/kg
flowd
5
1440/1350 267
20
200
38/8
18k
6
24
15
C, F, M, P
54
1550/1800 260
30
143
26/11
10k
6
11
22
F, M F none
20 30 50
30 22 29
295 266 111
53/37 45/12 25/25
21k 40l 38l
9 6 3
8 16 19
20 11 14
A
644
2
F, M
B
605
4
C D E
82 50 10
1 1 0
170/190 100/120 13/20
48.5 24.4 4.4
a Inhabitants of catchment area of STP. b Dominating industry in catchment area of STP: C, chemical; F, food; M, mechanical; P, pharmaceutical. Amount of surface water in raw sewage water. d Raw sewage water flow per week 2002/2003, × 103 cubic meters. e Yearly production of sludge (the final product), × 103 kilograms (dry weight), normalized to 1 week. f Percent dry weight of total mass (sludge as final product). g Amount of flocculation agent added during treatment, in milligrams per liter. h Amount of polymer (cationic polyelectrolyte polymer) added during treatment as flocculation agent and/or for dewatering sludge, in grams per kilogram. i Hydraulic retention time, in hours. j Solid retention time (days) in digester (except Floda, STP E; aerobic stabilizator). k Ferrous sulfate. l Ferrous chloride. c
Sample Preparation: Sludge. For the extraction of the fluoroquinolones, trimethoprim, sulfamethoxazole, and metronidazole, the following method was used: the sludge samples were weighed (2.0 g) into a centrifugal glass tube and 10.0 mL of a pH 6 phosphate buffer was added. The samples were extracted by ultrasonication for 10 min and then centrifuged at 4800 rpm for 10 min. The supernatants were withdrawn and placed in separate 30 mL glass vials, 10.0 mL of 5% TEA in methanol/water (25/75) was added to the residues, and they were re-extracted as described above. The two supernatants were combined and filtered through 0.45 µm MF-membrane syringe filters. Prior to analysis with LC/MS, the internal standards were added to each sample vial. The method used for extraction of penicillins, cephalosporins, doxycycline, and erythromycin was identical to the one presented above with the exception that the solvent used in the final extraction consisted of methanol/water (60/ 40) only. To determine the dry weight of sludge samples, a 2.0 g portion were put into a beaker and incubated in an oven at 105 °C for 24 h. The sample was weighed and put back in the oven for another 24 h, to ensure that the sample had reached a constant weight. The values obtained were used for correcting sludge concentrations to dry weight. Liquid Chromatography-Mass Spectrometry. Sample extracts and calibration solutions (10 µL) were injected into a YMC Hydrosphere C18 analytical column, 150 × 4.6 mm i.d., 5 µm particle size (YMC Inc., Wilmington, NC) following a 10 × 4 mm i.d., 5 µm particle size guard column of the same type by use of an AS 3000 autoinjector (Thermo Finnigan, San Jose, CA). The analytes were chromatographically separated by use of a 15-min linear gradient of 95-50% H2O balanced with acetonitrile, both containing 0.1% (v/v) formic acid, at a flow rate of 0.8 mL min-1 generated by a P4000 HPLC pump (Spectra system, Thermo Finnigan) at 25 °C. The mass spectrometer (LCQ Duo ion trap, Thermo Finnigan) was used together with an electrospray ion source in positive ion mode. The source voltage was maintained at a constant 6.0 kV for all analytes except PcV, which was monitored in negative ion mode at -4.5 kV. The heated capillary temperature was set to 250 °C and the MS/MS parameters were optimized semiautomatically for each substance by use of LCQ Duo internal software, while the collision energy used to produce daughter ions was manually optimized. Internal Standards, Identification, and Quantification. In this work, group analogue internal standards were used to improve precision in the quantification of the analytes (9). Table 2 shows the monitored m/z values for parent and daughter ions of the antibiotics and their respective group analogue internal standard, with the exception of erythromycin using demeclocycline as internal standard. Identification of each analyte was based on chromatographic retention
time, and selectivity was ensured by using a single appropriate transition in MS/MS mode. Note that 2-Me-5-nitroimidazole was monitored by selected ion monitoring (SIM) since a higher signal-to-noise ratio was obtained when monitoring the parent ion compared to the daughter ion. The analytes were quantified by selected reaction monitoring (SRM), with the most abundant daughter ion recorded in each case. The internal standard calibration method was used for all aqueous samples, analyte/internal standard (IS) peak area ratios were calculated, and no correction in terms of recovery was performed. However, for sludge extracts the internal standards were added just prior to injection into the LC-MS to allow compensation to be made for variations in injection volumes and ionization efficiency. Therefore, analyte concentrations in sludge were corrected for extraction recoveries. A seven-point calibration curve, generated from standards prepared in H2O/ACN (95/5 v/v) with concentrations ranging from 1 to 1500 ng mL-1 of the antibiotic substances was used for this purpose. It should be noted that erythromycin was determined in the form of its dehydration product, erythromycin[-H2O], since previous findings had shown this to be its predominant form in such extracts (5). Erythromycin was rapidly dehydrated in the stock standard solution and a semiquantitative determination was performed due to insufficient accuracy of the concentration of the dehydrated product in the standard obtained here (with and without lowering the pH in the standard). All sample matrices were monitored without the addition of internal standards in order to avoid underestimating analyte concentrations due to the possible presence of the standards used. Instrumental blank samples consisting of 5% acetonitrile in water were injected regularly to reduce and make corrections (if necessary) for potential memory effects. Quality Assurance. The recoveries of the analytes in the fortified laboratory samples were evaluated with 500 mL of tap water, fortified with 400 ng of each substance. For sludge, 2 g samples were fortified with 20 µg of each analyte and left to equilibrate for 1 h prior to extraction. Six samples were used in both of the extraction efficiency experiments described above. The internal standards were added to the aqueous and sludge extracts before the instrumental determination to minimize variations induced by injection volume and ionization efficiency. The recoveries were determined from the analyte/IS peak area ratios for the spiked samples in a matching standard calibration curve. The recoveries of the internal standards used were evaluated by the same methology as described above. Both water and sludge were fortified in the field at the Ryaverken STP to ensure that the compounds of interest were stable during transport and storage in these biologically active matrices. For this purpose, a final effluent sample (1 L) was fortified with each substance at 10 µg L-1, and to VOL. 39, NO. 10, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3423
FIGURE 2. General principle of the treatment process at the five investigated sewage treatment plants scattered across Sweden: (A) Stockholm (Henriksdal), (B) Gothenburg (Ryaverken), (C) Umeå, (D) Kalmar, and (E) Floda. evaluate analyte stability in sludge, three sets of three samples (2 g each) were used, fortified at 0, 10, and 20 mg kg-1. All samples were immediately frozen at -18 °C after the fortification. The reason for using moderately elevated amounts of analyte substances was to allow possibly high degrees of degradation to be detected and quantified. The same sample preparation procedure for sewage water and sludge, as described above, was used for the samples fortified in-field. Information regarding the homogeneity of the analyte concentrations in the sludge samples was obtained by determining the mean concentration and standard deviation of nine subsamples from Ryaverken STP. 3424
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 10, 2005
To evaluate possible errors during quantification, three blank samples of tap water and three empty vials were extracted by the sample preparation protocols that were applied to the sewage water and sludge, respectively. Calculations. The degree of elimination/removal in percentage was calculated from the analyte concentration in raw sewage water (Crs) and final effluent (Cfe): [(Crs Cfe)/Crs] × 100. Mass flows, in grams per week, of individual antibiotics were calculated from concentrations of each sampling occasion and in each matrix together with the corresponding weekly raw sewage volume (influent and effluent volume assumed to be equal). For sludge the yearly production was
TABLE 2. Monitored Ions of the Target Analytes and Their Respective Internal Standards
g
substance
m/z parent ion
CE,a %
m/z daughter ion
enrofloxacin,b IS norfloxacinb ofloxacinb ciprofloxacinb sulfamethazine,c IS sulfamethoxazolec diaveridine,d IS trimethoprimd cephalexin,e IS PcVf amoxicillinf ampicillinf cefadroxile 2-Me-5-nitroimidazole,g IS metronidazoleg demeclocycline,h IS doxycyclineh erythromycin[-H2O]i
360.1 [M + H]+ 320.0 [M + H]+ 362.1 [M + H]+ 332.1 [M + H]+ 279.0 [M + H]+ 254.0 [M + H]+ 261.2 [M + H]+ 291.0 [M + H]+ 347.8 [M + H]+ 349.3 [M - H]365.8 [M + H]+ 349.9 [M + H]+ 363.8 [M + H]+ 128.0 [M]+ 172.0 [M + H]+ 465.0 [M + H]+ 445.0 [M + H]+ 716.0 [M - H2O + H]+
28 27 26 30 28 29 37 38 25 19 19 22 18
316.2 [M - CO2 + H]+ 276.2 [M - CO2 + H]+ 318.1 [M - CO2 + H]+ 288.2 [M - CO2 + H]+ 203.8 187.9 123.1 230.1 [M - 2CH3O + H]+ 157.8 208.0, 304.9 348.8 159.9 346.6
27 22 22 27
128.0 [M - C2H4O + H]+ 448.0 [M - NH3 + H]+ 428.1 [M - NH3 + H]+ 558.1
a Collision energy (arbitrary units). b Fluoroquinolone group. c Sulfonamide group. Nitroimidazole group. h Tetracycline group. i Macrolide group.
d
No group. e Cephalosporin group. f Penicillin group.
TABLE 3. Quality Assurance Data: Recoveries and Limits of Quantification (LOQ) laboratory fortified samples
substance
tap water recovery % (n ) 6)
sludge recovery % (n ) 6)
enrofloxacin norfloxacin ofloxacin ciprofloxacin sulfamethazine sulfamethoxazole diaveridine trimethoprim cephalexin PcV amoxicillin ampicillin cefadroxil 2-Me-5-nitroimidazole metronidazole demeclocycline doxycycline erythromycin[-H2O]
82 ( 4 90 ( 2 85 ( 2 94 ( 14 84 ( 10 95 ( 11 88 ( 9 85 ( 6 66 ( 8 61 ( 3 55 ( 5 50 ( 3 65 ( 9 57 ( 11 66 ( 10 57 ( 9 51 ( 4 68 ( 6
53 ( 8 62 ( 7 86 ( 17 51 ( 7 62 ( 11 71 ( 15 66 ( 10 64 ( 8 40 ( 8 33 ( 7 14 ( 3 33 ( 7 25 ( 5 41 ( 8 45 ( 10 51 ( 8 55 ( 5 57 ( 10
in-field fortified samples waterb
standard addition quantificationa
recovery % (n ) 1)
sludge recovery % (n ) 9)
sludge mg/kg (n ) 3)
97 95 98
60 ( 2 83 ( 8 50 ( 5
2.0 ( 0.4 (1.7) 0.3 ( 0.2 (0.1) 3.3 ( 0.3 (3.4)
101
limits of quantification water LOQH2O,c ng/L
sludge LOQsludge,d mg/kg
7 6 6
0.1 0.1 0.1
71 ( 13
80
1.1
72
72 ( 10
8
0.1
