Environ. Sci. Technol. 2007, 41, 4471-4478
Screening Method for Ecotoxicological Hazard Assessment of 42 Pharmaceuticals Considering Human Metabolism and Excretory Routes JUDIT LIENERT,* KARIN GU ¨ DEL, AND BEATE I. ESCHER Eawag, Swiss Federal Institute of Aquatic Science and Technology, CH-8600 Duebendorf, Switzerland
We assessed the ecotoxicological hazard potential of 42 pharmaceuticals from 22 therapeutic groups, including metabolites formed in humans. We treated each parent drug and its metabolites as a mixture of similarly acting compounds. If physicochemical or effect literature data were missing, we estimated these with quantitative structureactivity relationships (QSAR). Additionally, we estimated micropollutant removal efficiency of urine source separation using pharmaceutical information. On average, 50% of a parent drug was metabolized, and 70% was excreted with urine, albeit with large variations among drugs. Metabolism reduced the toxic potential of all but eight drugs. The subsequently modeled risk quotient was mostly below the threshold of one. However, ibuprofen and its metabolites in a mixture could pose an ecotoxicologal risk; and possibly also acetylsalicylic acid, bezafibrate, carbamazepine, diclofenac, fenofibrate, and paracetamol. Lipophilicity and sale quantities of parent drugs alone were insufficient to estimate their ecotoxicological risk. Urine separation could decrease the ecotoxicological risk of some, but not all drugs. The estimated risk quotients were equal in urine and feces, again with large variations among compounds. Because of scientific limitations of the model and inconsistent literature data the results are somewhat uncertain. However, this new approach allows first tier screening of single drugs, thus supporting decision-making.
Introduction Pharmaceuticals are increasingly detected in sewage effluents and receiving waters (1-4). Obviously, some pass wastewater treatment plants unchanged, while others are transformed or eliminated (5). After 7 years of preparation, European Medicines Agency (EMEA) issued a guidance document on environmental risk assessment of human medicinal products (6). It relies on the risk quotient approach used in the EU also for industrial chemicals and biocides (7), where the predicted environmental concentration (PEC) is compared to the predicted no-effect concentration (PNEC). However, it differs from the EU approach because in the first phase of tier A, only exposure information is assessed. In the second phase, only chronic toxicity studies are accepted to derive the PNEC. So far, studies performing environmental risk * Corresponding author e-mail:
[email protected]; phone: +41-44-823 5574; fax: +41-44-823 5389. 10.1021/es0627693 CCC: $37.00 Published on Web 05/16/2007
2007 American Chemical Society
assessment according to EMEA (6) have been carried out only for a limited number of drugs due to lack of toxicity data and environmental concentrations (8-13). In tier B, the EMEA guideline also calls for an environmental risk assessment of metabolites that constitute more than 10% of the total excretion (6). This is reasonable, because pharmaceuticals are extensively metabolized in humans (14), usually increasing their hydrophilicity. Although human metabolism is well-known from drug development, little is known about how much this reduces their ecotoxicological potential, apart from the general assumption that more hydrophilic compounds are less hazardous to aquatic organisms. For β-blockers, our first study indicated that hazard assessment of parent compounds gives satisfactory results for metoprolol and atenolol, whereas metabolism strongly decreased the ecotoxicity of propranolol (15). Experimental data on metabolites are scarce, and a large European consortium is currently assessing some case-study pharmaceuticals (16). Given the large number of drugs, it seems impossible to conduct realistic ecotoxicological analyses that take account of all these aspects. Here, we propose a screening tool to identify pharmaceuticals, including their human metabolites, with high environmental risk. This tool uses literature data of human metabolism and excretion, pharmaceutical sales data, and physicochemical properties of parent drug and metabolites. For the many cases where ecotoxicological data are missing, baseline toxicity can be modeled with lipophilicity estimates using quantitative structure-activity relationships (QSAR). The inclusion of known specific toxicity effects is possible and has been illustrated for algal toxicity of β-blockers (15). The mixture effects of each parent drug and its metabolites are treated with the model of concentration addition, assuming a similar mode of toxic action of all components. Although to date neither chronic toxicity nor mixture effects of different pharmaceuticals have been assessed with the proposed screening tool, it is suitable to treat these cases, too, provided there are sufficient input data available in the literature. The first aim of this paper was to apply the methodology to a broad range of pharmaceuticals and their human metabolites. It was not the aim to perform a comprehensive risk assessment, but to evaluate if the metabolites’ contribution to ecotoxicity needs to be accounted for. The presented modeling method was motivated by urine source separation, and the second aim of this paper was to estimate the fractions of parent drug and metabolites excreted via urine and feces. Third, we wanted to assess how strongly the ecotoxicological potential of pharmaceuticals could be reduced if urine were kept away from wastewater. Because urine contains such a large fraction of metabolites, it seemed essential to include human metabolism in the modeling procedure. From an engineering point of view (17), the environmental risk from chemicals in wastewater can be reduced by end-of-pipe measures at treatment plants, either by optimizing existing processes (e.g., increasing sludge residence times) or by introducing additional treatment steps (e.g., ozonation). Alternatively, measures at the source might be more effective in the long run. Urine separation has been proposed as a source control measure in wastewater management, mainly because most nutrients in domestic wastewater (generally >80% nitrogen, >50% phosphorus (18)) stem from urine. Separate collection of urine would greatly simplify wastewater treatment (19) and nutrients could be recycled to agriculture (see http://www.novaquatis.eawag.ch). Because pharmaceuticals are usually metabolized to a polar VOL. 41, NO. 12, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4471
water-soluble form (14) it is expected that many are highly concentrated in urine. Hence, separate collection of urine could reduce the amount of drugs in wastewater, an argument receiving increasing popularity in the engineering community.
Materials and Methods Data Collection. We selected 33 pharmaceuticals () parent drugs/active ingredients) for ecotoxicological hazard assessment from nine therapeutic groups: analgesics/antiphlogistics, antibiotics, antilipidemics/lipid regulators, β-blockers, glucocorticoids/corticosteroids, antipsychotics, X-ray contrast agents, sexual hormones, and cytostatics. Additionally, we selected 18 pharmaceuticals from other groups (Table S1, Supporting Information). The total equals 51 pharmaceuticals from 27 therapeutic groups. We found sufficient data for 42 of 51 drugs; 9 were excluded. We collected data on the identity and excretion pathways of human metabolites and, where available, experimental ecotoxicity data (EC/LC50) from pharmaceutical compilations (20-23) and some primary literature sources. We compiled physicochemical data (structure, molecular weight, octanol-water partition coefficient Kow, acidity constant pKa) mainly from the Physical Properties Database (http:// www.syrres.com/esc/physprop.htm). Especially for metabolites, data were often missing. Here, we estimated the physicochemical parameters from the experimental data of the parent compound with increment methods (24, 25), instead of relying on estimation programs, which often have large systematic errors because they construct a molecule from scratch. Since the reported metabolite fractions excreted via urine or feces were often variable or data sources were not exhaustive, we followed a set of rules (Table S2, Supporting Information). We purchased sales quantities from 2004 of 35 pharmaceuticals in Switzerland (26). This corresponds to the 33 pharmaceuticals from nine therapeutic groups, plus carbamazepine and allopurinol. The sales data include drugs dispensed in hospitals, sold over the counter in pharmacies and drug stores, and provided directly in doctor’s practices. We also bought a list of the 100 bestselling pharmaceuticals in Switzerland (26); of all 51 drugs, 12 were listed (Table S1, Supporting Information). Modeling Procedure. We modeled the ecotoxicological potential of the 42 remaining pharmaceuticals following the flow chart in Figure 1. If experimental ecotoxicity data were available for parent compound or metabolites, we computed the toxic ratio TR. Otherwise, we assumed baseline toxicity. If TRparent was larger than 10 (i.e., parent was specifically toxic), we followed the methodology described in Escher et al. (15) and presented in detail in Table S3 (Supporting Information). There was no case with experimental data of metabolites available. The presence of a toxicophore points to a potential specific effect (27) for which an experimental follow-up study is needed. For lack of better experimental information, we estimated baseline toxicity with quantitative structure-activity relationships (QSAR) and treated each parent drug and its metabolites as a mixture of similarly acting compounds. Basic input data to model the toxic potential of parent drug and metabolites (TPmixture, Figure 1) were derived from the literature (estimation of lipophilicity, baseline effect concentrations (EC50), excreted fractions). Moreover, we generated a risk quotient (RQmixture) using simple predictions of drug concentrations in Swiss wastewater. More details are given in Figure 1, Table S3 (Supporting Information), and the references therein. The calculations are straightforward algorithms as is demonstrated with the example of ibuprofen (Table S4, Supporting Information). 4472
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 12, 2007
Results and Discussion Limitations of QSAR Modeling. Many pharmaceuticals are (partially) charged molecules. Since all QSAR equations in the literature are based on neutral molecules, we rescaled the published QSAR lipophilicity indicators. We used the liposome-water distribution ratio at pH 7 (Dlipw) because it is a better estimate of uptake into organisms than the ionization corrected Kow (Table S3A, Supporting Information). Any QSAR is only valid in a given lipophilicity range, typically the range of the training set. For example, the QSAR for Daphnia magna had a rather limited range of lipophilicity: It was derived for a range of log Kow ≈ 0.5-6.0 corresponding to a log Dlipw range of 0.97-5.94 (28). Hence, the results of the very hydrophilic drugs should be treated with caution because they are extrapolations. Log Kow < 0 is a rough indication that the validity of QSAR modeling is questionable, because the general model for baseline toxicity predicts a leveling off due to missing accumulation of the compound in membranes (e.g., methotrexate: log Kow,parent -1.85, Table S6, Supporting Information; log Dlipw,parent -2.13; Figure 3). Modeling of sugars (acarbose: log Kow,parent -3.0, log Dlipw,parent -2.2) is out of range anyway. Additionally, some metabolites were hydrophilic and lay outside the QSAR range. Often, this did not affect the overall outcome (TPmixture), because they were excreted in negligible fractions (e.g., conjugate of salicyluric acid, Table S5, Supporting Information). We modeled all cases, but if surprising results occur, these restrictions must be kept in mind. Limitations Concerning Specific Toxicity. We modeled baseline toxicity only, which per definition results in the same toxic potential (TPmixture) for all bioassays (daphnia, algae, fish). In 13 cases, baseline toxicity (TRparent < 10; Figure 1) was experimentally confirmed. In 29 cases, we assumed baseline toxicity because experimental toxicity data were not available (Table S5, Supporting Information). We rarely found indications for specific toxicity. β-blockers have a specific effect on photosynthesis of algae; we modeled these earlier (15). For oseltamivir, baseline toxicity was recently confirmed in our lab (unpublished results). Sulfamethoxazole and fluoxetine showed specific toxicity in algae; fluoxetine and paracetamol had borderline values in some test systems. As illustration, we modeled specific toxicity for sulfamethoxazole (Supporting Information; Table S10). Limitations Concerning Chronic Toxicity. The proposed method is applicable to chronic effects, provided that QSAR and experimental data of the parent compound are available for chronic endpoints. Since this is hardly ever the case, we limit our presentation to acute effects. This is not a limitation of the model, but is caused by limited input data. Relative Potencies of Metabolites, Excreted Fractions, and Toxic Potential of Mixture. The detailed results are given in the Supporting Information: of metabolites in Table S5, and an overview of the 42 parent drugs (lipophilicity, excretion, toxic potential) in Table S6. Nearly all metabolites formed in the human body were predicted to be less toxic than the parent drug (RPbaseline,i < 100%), ranging from 1% to 100%. We discuss the eight exceptions with RPbaseline,i > 100% below. The average total excretion fraction (fparent + Σfi) of the 42 drugs were as follows (in parentheses the large standard deviations based on the strongly differing minimal and maximal values from the literature): The total excretion via urine was 70% ((35%) and via feces it was 22% ((25%). As parent drug, 22% ((27%) was excreted via urine, and 19% ((24%) was excreted via feces. On average, 49% ((37%) of any compound was excreted as metabolites via urine, and 3% ((7%) was excreted via feces. This confirms the initial assumption that the majority of a pharmaceutical is excreted via urine, and half of it is metabolized before excretion.
FIGURE 1. Modeling procedure to estimate the ecotoxicological hazard potential of drugs after human metabolism. For clarity, details are only given for the case of baseline toxicity of a parent drug and its metabolites (details see Supporting Information, Table S3 and modeled for the example ibuprofen in Table S4). For other cases please refer to Escher et al. (15). Symbols: ) decision (yes/no); 0 next step; O basic input data. aNo case known for pharmaceuticals. bBecause presence of toxicophore does not render any quantitative information, experimental study is required. Excretion of metabolites via feces was often negligible, but feces contain the unabsorbed (non-metabolized) parent drug. Given the huge variability, a larger number of drugs should be analyzed for a more statistically sound overview (212 drugs in ref 29). Due to the variability of literature data, three drugs had an average total excretion >120% (digoxin, ibuprofen, methotrexate; Table S5, Supporting Information), six had 0 (RQmixture > 1) indicates that the drug poses an environmental risk to water organisms. Error bars denote the maximal estimate. In parentheses we show the respective therapeutic groups (see Figure 2). PNECmixture and log Dlipw,parent: Pearson’s R ) -0.976). Metabolism increased log PNECmixture so that higher concentrations are needed until an environmental effect occurs, compared with the effect concentrations of the parent drugs before metabolism (PNECparent). For clofibrate, for instance, metabolism reduced the toxic potential by 98% (Figures 2 and 3). The difference between PNECmixture and PNECparent was not dependent on hydrophobicity. The predicted environmental concentration (PECwastewater), which includes annual sales quantities and molecular weight of parent drugs, was distributed over the whole lipophilicity spectrum (R ) 0.039). Figure 3 also illustrates for which drugs QSARmodeling is questionable (log Dlipw,parent < 0). Estimation of Ecotoxicological Risk of Parent Drug and its Metabolites in Swiss Wastewater (RQmixture). The predictions of the risk quotient are shown in Figure 4 (details in Table S9, Supporting Information). RQmixture > 1 indicates that the drug poses an environmental risk to aquatic organisms, based on Swiss sales data, human metabolism, and for baseline toxicity only. Keeping the restrictions in mind, RQmixture allows for a realistic comparison of the environmental risk of single pharmaceuticals. Only one of our 30 modeled drugs, ibuprofen, had RQmixture > 1 (Figure 4, Table S9, Supporting Information). Diclofenac and fenofibrate had RQmixture > 0.5, and acetylsalicylic acid, bezafibrate, carbamazepine, and paracetamol > 0.1. For the remaining 23 of 30 drugs, we could not find indications that they pose an environmental risk with modeling of acute baseline toxicity. We can safely assume that the modeling procedure based on QSAR is valid for six of the seven top-risk drugs: Apart from fenofibrate none were outliers with respect to lipophilicity estimation (Figure 3). However, the total excretion of the highest-risk drug, ibuprofen, was over-estimated, which also results in an over-estimation of the risk (see above). On VOL. 41, NO. 12, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4475
the other hand, the excretion of carbamazepine was grossly under-estimated (total excretion: 51%); nevertheless it appeared as a top-risk drug. As discussed above, metabolism increased the modeled toxic potential of four drugs (digoxin, citalopram, methotrexate, sotalol; TPmixture; Figure 2), but none of these had a high RQmixture. For six of the seven high-risk drugs (Figure 4), modeling metabolism reduced the toxic potential by over 50% (paracetamol 88%; Figure 2, Table S6, Supporting Information), with the exception of bezafibrate (only 24% reduction of the toxic potential). Hence, not including metabolism would have resulted in an over-estimation of the environmental risk. One of the high-risk drugs, carbamazepine, formed a metabolite with a higher relative potency (RPi) than the parent drug: iminostilbene (RPiminostilbene ) 13; Table S5, Supporting Information); despite this, overall metabolism decreased the toxic potential by 64% compared with excretion of carbamazepine alone. However, this can be partially attributed to the under-estimated excretion mentioned above. There were strong positive correlations between log RQmixture and log sales quantities (Pearson’s R ) 0.749) or log Dlipw,parent (R ) 0.686), respectively. However, it would be difficult to predict the risk quotient for single drugs alone from basic input such as sales data or lipophilicity (illustrated in Figure S2, Supporting Information). All higher-risk drugs, except one, belonged to analgesics/ antiphlogistics (A) or antilipidemics/lipid regulators (C). Additionally, carbamazepine is a highly sold antiepileptic drug (Figure 4). The remaining drugs were distributed erratically: No cluster of therapeutic groups was discernible, apart from cytostatics (I; Figure 4). Here, the following points need consideration: First, mitomycin and especially methotrexate are very hydrophilic and lie outside the range for QSAR-modeling. Second, annual sales quantities are very low. Third, we assumed baseline toxicity because no data on ecotoxicity were available. However, mutagenicity is a probable effect of cytostatics, and mitomycin caused malformations in the offspring of freshwater mollusks (31). As in this study, others have identified highly sold drugs such as acetylsalicylic acid, ibuprofen, carbamazepine, or paracetamol (8-11) to have a high environmental risk quotient. Some drugs that did not rate high on our list such as ciprofloxacin or ethinylestradiol had a risk quotient >1 elsewhere (12). However, these studies did not include toxicity reduction by human metabolism. Generally, the environmental risk based on acute toxicity was judged as low (e.g., 13), but it was also concluded that environmental risk assessment should include chronic toxicity (9-11, 13), which is also one of the shortcomings of our study. Comparison of Ecotoxicological Risk (RQmixture) in Urine and Feces. No distinctive pattern concerning the risk from urine or feces was discernible (Figure 4, Table S9, Supporting Information). Some drugs had equal modeled risks in both fractions, e.g., the top-risk-candidates ibuprofen and diclofenac. Here, removing urine from wastewater would solve half the problem of anthropogenic micropollutants. In other cases, urine source separation would be useless, because the modeled ecotoxicological risk was much higher in feces (e.g., fenofibrate), while for many others urine separation would be efficient (e.g., bezafibrate, a high-risk-candidate in the same therapeutic group as fenofibrate; Figure 4). Over all 30 drugs, the mean RQmixture in urine was 67% (standard deviation ) 32%), and 33% in feces (SD ) 32%) without scaling. To compare the relative ecotoxicological risk of all modeled 30 drugs, we rescaled RQmixture as follows: The sum of RQmixture over the 30 drugs was set to 100%. Now it became obvious that ibuprofen dominated this mixture, its RQmixture being 52% (diclofenac: 13%). With this rescaling, 50.3% of RQmixture was contained in urine and 49.7% was in feces. Thus, 4476
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 12, 2007
general rules are not applicable but depend on the drug; overall, urine and feces contained equal modeled ecotoxicological risks. Above, we showed that for a first rough estimate of the toxic potential (TPmixture) contained in urine (or feces), the excretion pathways of parent drugs plus metabolites are a good approximation (Figure S1, Supporting Information). However, to estimate the overall risk quotient that allows comparing different drugs, this would not suffice: There was no strong relationship between the ecotoxicological risk in urine (RQmixture) and the absolute fraction excreted via urine (fexcreted(parent+metabolites); Pearson’s R ) 0.217) or the excreted proportion in urine (R ) 0.099). The same applied to feces. Outlook. It was not our goal to perform a comprehensive risk assessment of pharmaceuticals. Rather, we wished to explore whether first tier assessment should include metabolite formation or whether this issue is negligible. Our results clearly demonstrate that metabolites matter and should not be excluded a priori: modeled human metabolism reduced the toxicity of nearly all drugs. Hence, studying parent drugs suffices for worst-case assessment, but results in over-estimating the ecotoxicological risk. The presented model was efficient for first tier assessment of the ecotoxicological hazard of pharmaceuticals in wastewater, which was mostly low with the exception of ibuprofen, and possibly six other drugs. Currently, the model does not include special cases such as a specifically acting metabolite, while the parent is baseline toxic, nor chronic effects or specific modes of toxic action of the parent compound. However, it is theoretically possible to model these cases. For chronic effects, experimental data on the parent drug and a valid QSAR are necessary, which are rarely available. Previously, we demonstrated the method for specifically acting compounds with algal toxicity of β-blockers (15). Again, the literature data are not exhaustive enough to classify all parent drugs, and there is a clear need to improve this ecotoxicological data base. The model is also very simplistic by applying concentration addition of baseline toxicity as only mixture model. However, even if one of many metabolites would show a specific effect, while all others and the parent drug were baseline toxicants, applying the model of concentration addition for all mixture constituents is justified. Two-step mixture toxicity simulations including independent specific action of one compound and baseline toxicity of all others are typically indistinguishable from mixture effects modeled with concentration addition of baseline toxicity of all compounds (our own unpublished work). Despite all limitations and uncertainties, the method facilitates the screening of a larger number of compounds, based on known literature data, and helps to detect drugs that need closer scrutiny. We hardly found general patterns concerning therapeutic groups. Although the lipophilicity of a parent drug and the sales quantities give rough indications, such data alone do not suffice to estimate the ecotoxicological risk of a specific drug. We only assessed the risk of 30 single compounds. However, a large number of human drugs are on the market, and the model does not include mixture effects between different pharmaceuticals. Our modeling exercise shows that the few parent pharmaceuticals analytically quantified in surface waters (1-4) constitute only the tip of an iceberg of metabolites, environmental transformation products, and other drugs. Moreover, if the risk quotient (RQ) is based on measured environmental concentrations (MEC), it neglects the contribution of the metabolites to the toxicity and can result in an underestimation of the risk. Here, a RQparent should be estimated from the MEC and the PNECparent and the risk quotients of all metabolites should be added incrementally. Based on such considerations and the fact that one drug alone, ibuprofen, already poses an
ecotoxicological risk, and in view of the many others that were not screened, application of the precautionary principle to keep these micropollutants away from the environment seems justified (32). For single compounds, our model is a worst-case assessment, because it does not include dilution of wastewater during rain and in receiving waters, elimination of pharmaceuticals by wastewater treatment, and regional or temporal variability. As with the estimated environmental risk, degradation of drugs varies considerably, even within one therapeutic group. Of our seven top-risk candidates, the expected removal by conventional biological wastewater treatment was 90% for ibuprofen and paracetamol, and 20-90% for acetylsalicylic acid, bezafibrate, and fenofibric acid (5). Such data can be incorporated into our model to make it more realistic. Also environmental metabolites could be assessed in a second tier with the same model as for human metabolites. Furthermore, we did not include direct disposal of drugs. In a representative German survey 16% of 1306 respondents discarded tablets via toilets, and 43% discarded liquid medicals via sinks or toilets (33). Of 400 English households, 12% discarded pharmaceuticals into sinks or toilets; thus this disposal route requires greater attention (34). For Switzerland, these numbers seem exaggerated: We questioned 501 people about urine separation in a public library, and only 1% stated that they regularly flushed medicals down the toilet, 13% disposed of them with household waste, and 76% returned them to the retailer (unpublished data). Finally, we showed that an increasingly discussed source control measure, urine separation, could successfully decrease the ecotoxicological risk of some, but certainly not all drugs. However, it is possible that the drugs and metabolites from feces can be better removed from wastewater than the hydrophilic drugs from urine. These substances presumably adsorb to fecal matter due to their higher hydrophobicity and end up in sewage sludge, which is incinerated in Switzerland. This hypothesis, however, still lacks experimental support. Protection of receiving waters from pharmaceuticals as the sole rationale for introducing urine source separation is hardly justified, since NoMix technology is still in a pilot phase and is thus associated with drawbacks. Nonetheless, combined with other motivations or in special circumstances, it might be profitable. For instance, measures at point sources such as hospitals or homes for the elderly could be an effective first step. To this end, mass balances for drugs administered in, e.g., hospitals should be carried out. Urine source separation might be an option, but treatment of the entire hospital wastewater also needs consideration. Our screening method could successfully support decision making in such cases. The presented model is an efficient approach to initial hazard assessment. It can easily be extended, for instance, to include other classes of environmental pollutants such as biocides and their metabolites.
Acknowledgments We thank Alfredo Alder, Brenda Bonnici, Timur Bu ¨ rki, Rik Eggen, Rahel Gilg, and Tove Larsen for support, and three anonymous reviewers for helpful suggestions. This study was partially funded by the Swiss Federal Office for the Environment (FOEN) and by the European Union under the 6th framework program in the STREP ERAPharm (SSPI-CT-2003511135).
Supporting Information Available Details of 51 pharmaceuticals, rules for data handling, detailed modeling procedure, and examples ibuprofen and
sulfamethoxazole, excretion values, and modeling results (relative potency of metabolites, toxic potential, lipophilicity estimates, effect concentrations, predicted environmental concentrations, predicted no-effect concentrations, estimated ecotoxicological risk, and specific toxicity). This material is available free of charge via the Internet at http:// pubs.acs.org.
Literature Cited (1) Daughton, C. G.; Ternes, T. A. Pharmaceuticals and personal care products in the environment: Agents of subtle change? Environ. Health Perspect. 1999, 107, 907-938. (2) Heberer, T. Occurrence, fate, and removal of pharmaceutical residues in the aquatic environment: a review of recent research data. Toxicol. Lett. 2002, 131 (1-2), 5-17. (3) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; Barber, L. D.; Buxton, H. T. Pharmaceuticals, hormones, and other organic wastewater contaminants in US streams, 1999-2000: A national reconnaissance. Environ. Sci. Technol. 2002, 36 (6), 1202-1211. (4) Roberts, P. H.; Thomas, K. V. The occurrence of selected pharmaceuticals in wastewater effluent and surface waters of the lower Tyne catchment. Sci. Total Environ. 2006, 356 (1-3), 143-153. (5) Joss, A.; Zabczynski, S.; Gobel, A.; Hoffmann, B.; Loffler, D.; McArdell, C. S.; Ternes, T. A.; Thomsen, A.; Siegrist, H. Biological degradation of pharmaceuticals in municipal wastewater treatment: Proposing a classification scheme. Water Res. 2006, 40 (8), 1686-1696. (6) EMEA, European Medicines Agency. Guideline on the Environmental Risk Assessment of Medicinal Products for Human Use; Doc. Ref. EMEA/CHMP/SWP/4447/00; 2006; accessible at http://www.emea.eu.int/pdfs/human/swp/444700en.pdf, info@ emea.eu.int, last visit to website: 10.08.2006. (7) European Commission. Technical Guidance Document in Support of Commission Directive 93/67/EEC on Risk Assessment for New Notified Substances, Commission Regulation (EC) No 1488/ 94 on Risk Assessment for Existing Substances, and Directive 98/8/EC of the European Parliament and of the Council Concerning the Placing of Biocidal Products on the Market; EUR 20418; Office for Official Publications of the European Communities: Luxembourg, 2003. (8) Stuer-Lauridsen, F.; Birkved, M.; Hansen, L. P.; Lutzhoft, H. C. H.; Halling-Sorensen, B. Environmental risk assessment of human pharmaceuticals in Denmark after normal therapeutic use. Chemosphere 2000, 40 (7), 783-793. (9) Jones, O. A. H.; Voulvoulis, N.; Lester, J. N. Aquatic environmental assessment of the top 25 English prescription pharmaceuticals. Water Res. 2002, 36, 5013-5022. (10) Sanderson, H.; Johnson, D. J.; Wilson, C. J.; Brain, R. A.; Solomon, K. R. Probabilistic hazard assessment of environmentally occurring pharmaceuticals toxicity to fish, daphnids and algae by ECOSAR screening. Toxicol. Lett. 2003, 144 (3), 383-395. (11) Ferrari, B.; Mons, R.; Vollat, B.; Fraysse, B.; Paxeus, N.; Lo Giudice, R.; Pollio, A.; Garric, J. Environmental risk assessment of six human pharmaceuticals: Are the current environmental risk assessment procedures sufficient for the protection of the aquatic environment? Environ. Toxicol. Chem. 2004, 23 (5), 1344-1354. (12) Huschek, G.; Hansen, P. D.; Maurer, H. H.; Krengel, D.; Kayser, A. Environmental risk assessment of medicinal products for human use according to the European commission recommendations. Environ. Toxicol. 2004, 19 (3), 226-240. (13) Carlsson, C.; Johansson, A. K.; Alvan, G.; Bergman, K.; Kuhler, T. Are pharmaceuticals potent environmental pollutants? Part I: Environmental risk assessments of selected active pharmaceutical ingredients. Sci. Total Environ. 2006, 364 (1-3), 6787. (14) Sheldon, L.; Umana, M.; Bursey, J.; Gutknecht, W.; Handy, R.; Hyldburg, P.; Michael, L.; Moseley, A.; Raymer, J.; Smith, D.; Sparacino, C.; Warner, M. Biological Monitoring Techniques for Human Exposure to Industrial Chemicals; Noyes Publications: Norwich, NY, 1986. (15) Escher, B. I.; Bramaz, N.; Richter, M.; Lienert, J. Comparative ecotoxicological hazard assessment of beta-blockers and their human metabolites using a mode-of-action-based test battery and a QSAR approach. Environ. Sci. Technol. 2006, 40 (23), 74027408. (16) Knacker, T.; Duis, K.; Ternes, T.; Fenner, K.; Escher, B. I.; Schmitt, H.; Ro¨mbke, J.; Garric, J.; Hutchinson, T.; Boxall, A. B. A. The VOL. 41, NO. 12, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4477
(17) (18) (19) (20) (21) (22) (23)
(24) (25) (26) (27)
EU-project ERAPharm - Incentives for the further development of guidance documents? Environ. Sci. Pollut. Res. 2005, 12, 6265. Larsen, T.A.; Lienert, J.; Joss, A; Siegrist, H. How to avoid pharmaceuticals in the aquatic environment. J. Biotechnol. 2004, 113 (1-3), 295-304. Larsen, T. A.; Peters, I.; Alder, A.; Eggen, R.; Maurer, M.; Muncke, J. Re-engineering the toilet for sustainable wastewater management. Environ. Sci. Technol. 2001, 35 (9) 192A-197A. Larsen, T. A.; Gujer, W. Separate management of anthropogenic nutrient solutions (human urine). Water Sci. Technol. 1996, 34 (3-4), 87-94. Beyer, K. H. Biotransformation der Arzneimittel, 2nd ed.; Springer-Verlag: Berlin, Germany, 1990. Dollery, C. Therapeutic Drugs; Churchill Livingstone: Edinburgh, UK, 1991. Baselt, R. C.; Cravey, R. H. Disposition of Toxic Drugs and Chemicals in Man, 4th ed.; Chemical Toxicology Institute: Foster City, CA, 2000. Arzneimittel-Kompendium der Schweiz (Swiss Pharmaceutical Compendium); Documed AG: Basel, Switzerland, 2006; Available at http://www.kompendium.ch, last visit to website: 27.06.2006,
[email protected]. Hansch, C.; Leo, A. Exploring QSAR. Fundamentals and Applications in Chemistry and Biology; American Chemical Society: Washington, DC, 1995. Hansch, C.; Leo, A.; Hoekman, D. Exploring QSAR. Hydrophobic, Electronic and Steric Constants; American Chemical Society: Washington, DC, 1995. IMS Health GmbH. Hergiswil, Switzerland, 2005; http:// www.ihaims.ch, last visit to website: 14.08.2006, info@ ch.imshealth.com. Sinclair, C. J.; Boxall, A. B. A. Assessing the ecotoxicity of pesticide transformation products. Environ. Sci. Technol. 2003, 37 (20), 4617-4625.
4478
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 12, 2007
(28) Zhao, Y. T.; Cronin, M. T. D.; Dearden, J. C. Quantitative structure-activity relationships of chemicals acting by non-polar narcosis - Theoretical considerations. Quant. Struct-Act. Relat. 1998, 17 (2), 131-138. (29) Lienert, J.; Bu ¨ rki, T.; Escher, B. I. Reducing micropollutants with source control: Substance flow analysis of 212 pharmaceuticals in feces and urine. In Proceedings of IWA Advanced Sanitation Conference, Aachen, Germany, 12.-13.3.2007, pp. 15/1-15/9; submitted to Water Sci. Technol. (30) Boxall, A. B. A.; Sinclair, C. J.; Fenner, K.; Kolpin, D.; Maud, S. J. When synthetic chemicals degrade in the environment. Environ. Sci. Technol. 2004, 38 (19), 368A-375A. (31) Nakano, E.; Watanabe, L. C.; Ohlweiler, F. P.; Pereira, C. A. D.; Kawano, T. Establishment of the dominant lethal test in the freshwater mollusk Biomphalaria glabrata (Say, 1818). Mutat. Res. Genet. Toxicol. Environ. Mutagen. 2003, 536 (1-2), 145154. (32) Rogers, M. D. Risk analysis under uncertainty, the Precautionary Principle, and the new EU chemicals strategy. Regul. Toxicol. Pharm. 2003, 37, 370-381. (33) START. Strategien zum Umgang mit Arzneimittelwirkstoffen im Trinkwasser. Medikamentenentsorgung in privaten Haushalten. (in German); ISOE, Institut fu ¨ r sozial-o¨kologische Forschung GmbH: Frankfurt a.M., Germany, 2006; http:// www.start-project.de/downloads/start-Entsorgungsstudie2006.pdf; e-mail:
[email protected]. (34) Bound, J. P.; Voulvoulis, N. Household disposal of pharmaceuticals as a pathway for aquatic contamination in the United Kingdom. Environ. Health Perspect. 2005, 113, 1705-1711.
Received for review November 21, 2006. Revised manuscript received April 5, 2007. Accepted April 9, 2007. ES0627693