Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin

Transcription

1 KEI 2007 RE-06 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Jeongim Park Myung-Hyun Kim Kyungho Choi Young-Hee Kim Min-Young Kim

2

3 Research Staff Jeongim Park (Project Leader, Korea Environment Institute) Kyungho Choi (Seoul National University) Young Hee Kim (Seoul National University) Myung Hyun Kim (Korea Environment Institute) Min Young Kim (Research Institute of Public Health & Environment, Seoul) Project Advisory Committee Chan Gu Park, Do Hyun Baek, Dong Chun Shin, Kyoung Hee Shin, Kyoung Soon Kim, Soo Ho Sung, Soon Cho, Yong Seung Shin Copyright c 2007 by Korea Environment Institute All rights reserved. No part of this publication may be reproduced or transmitted in any form or any means without permission in writing from the publisher Publisher Jeong Hoi Seong Published by Korea Environment Institute Bulgwang Dong (290 Jinheungno), Eunpyeong Gu, Seoul, Republic of Korea Tel.(822) Fax.(822) Published and printed in December 2007 ISBN Price 8,000

4 FOREWORD Human and veterinary pharmaceuticals are widely used to treat disease and improve the quality of life. However, during their use, human and veterinary pharmaceuticals may be released into the environment. With the detection of pharmaceuticals in the environment and concerns regarding antibiotic resistance and endocrine disruption, the environmental science community, the public, and policy makers now consider pharmaceuticals as emerging environmental pollutants. The Korea Environment Institute has been investigating this issue since Pharmaceuticals in the Environment and Management Approaches in Korea (KEI, 2005) raised concerns regarding a previously hidden environmental issue in Korea. The report introduced the sources and pathways of pharmaceuticals in the environment, fate, occurrence, and detection from the environment, and risk assessment methods. An Approach for Developing Aquatic Environmental Risk Assessment Framework for Pharmaceuticals in Korea (KEI, 2006) prioritized pharmaceutical substances for environmental risk assessment (ERA) in Korea and provided a basis for developing a proper strategy of pharmaceutical ERA. At the end of the study, ERA for selected antibiotics was conducted in order to assess the proposed ERA strategy. The study suggested that exposure assessment, a key element of ERA, had been hampered by the continuing difficulties and expenses involved in measuring the low ppt concentrations of pharmaceuticals in the environment. As a succeeding study, this report aims to investigate the utilization of computerized exposure models to assess the environmental exposure of pharmaceuticals. At the same time, potential risk management measures are proposed in order to minimize the deposition of pharmaceuticals into the environment. I believe this report will be useful to researchers and policy makers concerned with the environmental risk of chemicals, particularly pharmaceuticals.

5 We would like to thank PhRMA (The Pharmaceutical Research and Manufacturers of America) for their support, particularly for the use of the PhATE TM model. Special thanks are extended to the advisory committee: Dr. Chan Gu Park, Professor Do Hyun Baek, Professor Dong Chun Shin, Dr. Kyoung Soon Kim, Dr. Yong Seung Shin, Dr. Kyoung Hee Shin, Mr. Soo Ho Sung, and Dr. Soon Cho. Ms. Marcia Plese read drafts of the report and made numerous helpful suggestions that greatly improved it. We are grateful to Ms. Hyun Jung Hong for her assistance of GIS works and Mr. David Matte for his editorial support. Finally, this report is also indebted to anonymous reviewers for their insightful comments. December 2007 Korea Environment Institute President Jeong Hoi Seong, Ph.D.

6 Abstract Pharmaceuticals are indispensable as they cure those who suffer from disease and their availability improves the quality of life. Veterinary medicines are also widely used to treat disease and improve the productivity of livestock farming. However, during their use, human and veterinary pharmaceuticals have the potential of being released into the environment. In recent years, the possible environmental (ecological) risk of pharmaceuticals in the aquatic environment has become a matter of increasing public concern. Potential ecological effects from the presence of pharmaceuticals in the environment have generally focused on the following two concerns: 1) the release of antibiotics into the environment increases the chance of antibioticresistant microorganisms and promotes the spread of resistant genes, and 2) when drugs affecting hormonal systems reach organisms in nature, it may result in a reproductive disturbance in the ecosystem. Environmental risk assessment (ERA) is considered the best scientifically based approach for evaluating the potential effects of contaminants on communities and ecosystems. The process includes problem formulation, exposure assessment, effects assessment, and risk characterization. Accurate exposure assessment is a key element of ERA. However, exposure assessment has been hampered by the continuing difficulties and expense involved in measuring the low ppt concentrations of pharmaceuticals in the environment. In addition, real time monitoring data provides only snapshots of contaminant concentrations. Thus, when faced with the task of assessing the environmental exposure of pharmaceuticals, the utilization of exposure models becomes essential. The first objective of this report is to apply computerized exposure models to assess the environmental concentration of human and veterinary pharmaceuticals in the Han River and the Kyungahn stream, a major branch of the Han River. PhATE TM and SWMM are identified as appropriate exposure models for this study based on data availability, researchers previous

7 experience with models, and accessibility to models. The models investigated in this study intend to provide rapid predictions regarding the potential environmental fate of a compound. In this study, model predicted PECs are compared to field data that either have been published previously or were empirically measured during this study (Chapters 2, 3, and 4). A second objective of this study is to estimate the total environmental concentration of pharmaceuticals, from both human and animal use, by integrating the simulation results from PhATE TM and SWMM. Although the introduction routes into the environment for veterinary pharmaceuticals are different from those for human use, both human and animal pharmaceuticals eventually end up reaching surface water. Therefore, estimating the environmental concentration of such pharmaceuticals based solely on either human consumption or animal consumption results in an underestimation of environmental exposure. In this study, a workable framework to estimate PECs for these dual usage pharmaceuticals is suggested (Chapter 5). Finally, a third objective is to perform an environmental risk assessment (ERA) for selected pharmaceuticals. Hazard quotients (the ratio of EC to PNEC) based on PECs are compared to those based on MECs. This exercise will demonstrate the applicability of modeling approaches in the risk assessment of pharmaceuticals in the environment. The potential benefit of using molecular level biomarkers to assess pharmaceutical toxicity is also discussed and methods are presented (Chapter 6). In order to minimize the deposition of pharmaceuticals into the environment, potential risk management actions are suggested; disposal labeling on pharmaceutical products, discharge guidelines for pharmaceutical manufacturing facilities, pretreatment of hospital wastewater, modification of wastewater treatment plant infrastructure or operating parameters, standardizing guidelines for the handling and disposal of unused medicine, and efficient dispensing practices and packaging (Chapter 7).

8 Acronym List ACRs ADIs APVMA APIs BASINS BCF BOD5 CEPA CHMP COD CRS CSOs CVM CVMP DR3M QUAL DSL DWS EA EC50 ECETOC EIA EIC EMEA EPA EPI suite ERA Acute to Chronic effect Ratio Acceptable Daily Intakes Australian Pesticides and Veterinary Medicines Authority Active Pharmaceutical Ingredients Better Assessment Science Integrating Point and Nonpoint Source Bioconcentration Factor Biological Oxygen Demand Canadian Environmental Protection Act Committee for Medicinal Products for Human Use Chemical Oxygen Demand Chemical Ranking and Scoring Combined Sewer Overflows Center for Veterinary Medicine (USA) Committee for Medicinal Products for Veterinary Use Distributed Routing Rainfall Runoff Model that includes quality simulation Domestic Substances List Drinking Water Supply Environmental Assessment Median Effect Concentration European Centre for Ecotoxicology and Toxicology of Chemicals Environmental Impact Assessment Expected Introduced Concentration European Medicines Agency Environmental Protection Agency (US) Estimation Program Interface Suite (US EPA) Environmental Risk Assessment

9 FDA F&DA GC MS GIS GREAT ER model HPLC HQ HSPF model Kd KFDA Kow KPMA KRC K water LAS LOD LOEL LOQ MDL MEC MHLW MOCT MOE MOHW MORAG NA NEPA ND NIER Food and Drug Administration (US) Food and Drug Act (Canada) Gas Chromatography Mass Spectrometry Geographic Information System Geo referenced Regional Exposure Assessment Tool for European Rivers High Performance Liquid Chromatography Hazard Quotient Hydrological Simulation Program Fortran Coefficient of sludge water partition Korea Food and Drug Agency Octanol/water partition Coefficient Korea Pharmaceutical Manufacturer Association Korea Rural community & Agriculture corporation Korea Water Resources Corporation Linear Alkylbenzene Sulfonate Limit of Detection Lowest Observed Effect Level Limit of Quantification Method Detection Limit Measured Environmental Concentration Ministry of Health, Labor and Welfare (Japan) Ministry of Construction and Transportation (Korea) Ministry of Environment (Korea) Ministry of Health and Welfare (Korea) Veterinary Manual of Data Requirements and Guidelines Not Applicable National Environmental Policy Act Not Detected National Institute of Environmental Research (Korea)

10 NOEC NOEL NSNR OECD PEC PhACs PhATE TM model PhRMA PIE pka PNEC POTWs ppb ppt RQ SS STORM model STPs SWMM TKN TSS UCL USGS VDD VICH VMD VMPs WRC WWTPs No Observed Effect Concentration No Observed Effect Level New Substances Notification Regulations Organization for Economic Cooperation and Development Predicted Environmental Concentration Pharmaceutically Active Compounds Pharmaceutical Assessment and Transport Evaluation model Pharmaceutical Research and Manufacturers of America Pharmaceuticals in the Environment Acid Dissociation Constant Predicted No Effect Concentration Publicly Owned Treatment Works Parts Per Billion Parts Per Trillion Risk Quotient Suspended Solid Storage, Treatment, Overflow, Runoff Model Sewage Treatment Plants Storm Water Management Model Total Kjeldahl Nitrogen Total Suspended Solids Upper Confidence Limit US Geological Survey Veterinary Drugs Directorate International Cooperation on Harmonization of Technical Requirements for Registration of Veterinary Products Veterinary Medicines Directorate Veterinary Medicinal Products Water Research Center Wastewater Treatment Plants

11

12 Contents FOREWORD Abstract Acronym List Chapter 1. Introduction 1 Chapter 2. Review and Background 8 1. Occurrence Studies of Pharmaceuticals in the Korean Aquatic Environment 8 2. Environmental Safety Regulations for Pharmaceuticals The European Union The United States Canada Australia Japan Models for Predicting PECs of Pharmaceuticals Models for Predicting PECs of Human Pharmaceuticals Models for Predicting PECs of Veterinary Pharmaceuticals Selecting Models for This Study 57 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM Constructing PhATE TM for the Han River Watershed Data for the Han River Target Compounds and Collection of Compound Specific Data 66

13 2. Modeling Human Pharmaceuticals in the Han River Watershed with PhATE TM Point by Point Comparisons of Model Data to Field Data Field Measurement Data Point by Point Comparisons 78 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream Introduction Materials and Methods Study Area Selection of Target Pharmaceuticals Application of SWMM Results and Discussion Simulation of Runoff Flowrate Simulation of Runoff Quality Conclusions 114 Chapter 5. Environmental Exposure Modeling of Pharmaceuticals with the PhATE TM SWMM Combined Model in the Kyungahn Stream Materials and Methods Target Pharmaceuticals Model Application Results PhATE TM model in the Kyungahn stream Integrated Concentration of Sulfamethoxazole and Trimethoprim 123

14 Chapter 6. Environmental Risk Assessment for Selected Pharmaceuticals Introduction Materials and Methods Daphnia Chronic Test Fish Test Risk Assessment Results and Discussion Daphnia Test Fish Test Risk Assessment Conclusions 144 Chapter 7. Summary and Conclusions Summary of This Study Suggested Potential Risk Management Actions 147 References 153 Appendix 1: PhATE TM Input Data in the Han River 161 Appendix 2: SWMM Model Verification Data 174 Abstract in Korean 177

15 Tables Table 2 1. Limits of detection for the pharmaceuticals analyzed Table 2 2a. Summary of occurrence for pharmaceuticals analyzed in Korea (WWTPs and surfacewater) Table 2 2b. Summary of occurrence for pharmaceuticals analyzed in Korea (livestock WWTPs and neighboring surfacewater) Table 2 3. Pharmaceutical ERA regulations in other countries Table 2 4. Summary of key inputs for PhATE TM Model Table 2 5. Summary of 12 scenarios in VetCalc Table 2 6. Information on the reviewed models Table 3 1. Key inputs for PhATE TM model to run for the Han River Table 3 2. Summary of key information on seven STPs included in this study Table 3 3. Physicochemical properties of selected pharmaceuticals Table 3 4. Field measurement data generated during this study Table 3 5. Field measured and PhATE TM predicted concentrations of four human pharmaceuticals for each segment Table 4 1. Total number of major livestock animals in the study area.. 84 Table 4 2. General characteristics of selected antibiotics Table 4 3. Information requirements for SWMM Table 4 4. Data sources for SWMM Table 4 5. Characteristics of subcatchments in the study area Table 4 6. Rainfall events included in the study... 93

16 Table 4 7. Table 4 8. Table 4 9. Pharmaceutical concentrations during rainfall on May 12, Pharmaceutical concentrations during rainfall on May 16, Pharmaceutical concentrations during rainfall on June 21, Table Baseline values for variable sensitivity analysis in SWMM.. 98 Table Values of each variable as determined during runoff model calibration Table Infiltration coefficient of the Horton equation Table Unit discharge of target chemicals Table 5 1. Major sewage treatment plants in the Kyungahn stream Table 6 1. Acute to Chronic Ratios of the selected antibiotics Table 6 2. Environmental concentrations of the selected pharmaceuticals Table 6 3. PNECs of the selected pharmaceuticals Table 6 4a. Hazard Quotients (HQs) of selected pharmaceuticals in the Han River Table 6 4b. Hazard Quotients (HQs) of selected pharmaceuticals in the Kyungahn stream

17 Figures Figure 1 1. Flow of human and veterinary pharmaceuticals into the environment... 2 Figure 2 1. Progress of European Union (EU) legislation and technical guideline on ERA of human pharmaceuticals Figure 2 2. Guideline on the Environmental Assessment of Medicinal Products for Human Use Figure 2 3. The locations of the 11 watersheds included in PhATE TM Figure 2 4. PhATE TM Model output (print results) Figure 2 5. PhATE TM Model output (graphic results) Figure 2 6. PhATE TM Model output (map results) Figure 2 7. Field measured and PhATE TM predicted concentrations of surrogate compounds Figure 2 8. The cumulative probability distribution of all caffeine concentrations in US surface waters reported by the USGS and PECs generated by PhATE TM Figure 2 9. Predicted surface water concentrations of Paroxetine using the PhATE TM model Figure GREAT ER model outputs user interface Figure Output screen of SWMM Figure VetCalc Figure 3 1. Main screen of PhATE TM including the Han River Figure 3 2. PhATE TM Study area and locations of STPs Figure 3 3. STPs data screen in PhATE TM for the Han River Figure 3 4. Segmentation scheme of the Han River for PhATE TM... 66

18 Figure 3 5. Chemical properties screen in chemical data for PhATE TM Figure 3 6. Human Use/Loss screen and Treatment screen in chemical data for PhATE TM Figure 3 7. PhATE TM modeled segmental concentrations of Acetaminophen in the Han River Figure 3 8. PhATE TM modeled segmental concentrations of Cimetidine in the Han River Figure 3 9. PhATE TM modeled segmental concentrations of Roxithromycin in the Han River Figure PhATE TM modeled segmental concentrations of Chloramphenicol in the Han River Figure Sampling sites of human pharmaceuticals in this study Figure Field measured and PhATE TM predicted concentrations for each segment Figure 4 1. Study area of the Kyungahn stream for SWMM Figure 4 2. Chicken manure composting facility as non point sources of veterinary pharmaceuticals in the study area Figure 4 3. Schematic configuration of SWMM blocks Figure 4 4. Subcatchments of the study area Figure 4 5. Sensitivity analysis at Kyungahn 1 st bridge (peak flow) Figure 4 6. Sensitivity analysis at Kyungahn 1 st bridge (total flow) Figure 4 7. Calibration on SWMM parameters at Kyungahn 1 st bridge using the 2007/05/16 event (runoff flow) Figure 4 8. Calibration on SWMM parameters at Kyungahn 1 st bridge using the 2007/05/24 event (runoff flow) Figure 4 9. Regression curve between simulated and measured data (2007/05/24)

19 Figure Cow, pig, and chicken distribution in the study area Figure Storm events in Figure Comparison of measured and simulated values with SWMM at Kyungahn 1 st bridge for the 2007/05/16 event Figure Comparison of measured and simulated values with SWMM at Kyungahn 1 st bridge for the 2007/05/12 event Figure Comparison of measured and simulated values with SWMM at Kyungahn 1 st bridge for the 2007/06/21 event Figure 5 1. Diagram for integrated environmental concentrations of pharmaceuticals used for both human and animals Figure 5 2. Site map of STPs in the Kyungahn stream watershed Figure 5 3. Segmentation scheme for PhATE TM in the Kyungahn stream Figure 5 4. Input data setup in PhATE TM for the Kyungahn stream Figure 5 5. PhATE TM modeled segmental concentrations of pharmaceuticals Figure 5 6. Comparison of measured and simulated values with PhATE TM in the Kyungahn main stream Figure 5 7. Surface water concentrations of Sulfamethoxazole in the Kyungahn stream Figure 5 8. Adjusted environmental concentrations of pharmaceuticals in the Kyungahn stream Figure 6 1. Reproduction effects of 21 d chronic exposure to Daphnia magna Figure 6 2. Induction of vitellogenin mrna in male medaka Figure 6 3. Cumulative total eggs by exposure to 17 beta estradiol and enrofloxacin Figure 6 4. Induction of vitellogenin mrna in male medaka by exposure to 17 beta estradiol and enrofloxacin

20

21 Chapter 1. Introduction 1 Chapter 1. Introduction Jeongim Park Pharmaceuticals are indispensable for curing those who suffer from disease and their availability improves the quality of life. Veterinary medicines are also widely used to treat disease and improve the productivity of livestock farming. However, during their use, human and veterinary pharmaceuticals have the potential of being released into the environment. In recent years, the possible environmental (ecological) risk of pharmaceuticals in the aquatic environment has become a matter of increasing public concern. Pathways into the environment Pharmaceuticals are introduced into the environment through various routes. Generally they enter the environment during manufacture or after use by both humans and animals (Figure 1 1). Potential pathways for human pharmaceuticals to enter the environment include: 1) release from pharmaceutical manufacturing facilities, 2) disposal of unused pharmaceuticals by patients, hospitals, or distributors either to wastewater or to solid waste, or 3) patient excretion of pharmaceuticals and their metabolites to wastewater. Human wastes are typically treated in sewage treatment plants (STP). During wastewater treatment, a drug may be degraded via hydrolysis, oxidation, or biodegradation, or the drug may adsorb to solids that are isolated in sludge. Pharmaceutical concentrations in STP effluents depend on the removal efficiency of the STP treatment processes. STP effluents are generally considered the primary source of human pharmaceuticals into the aquatic environment. In addition, a STP release of untreated sewage during a storm flow or a potential transportation accident could be non routine episodes that admit additional pharmaceutical contamination into the environment (William, 2005).

22 2 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Veterinary pharmaceuticals may pose more threat to ecosystems than human pharmaceuticals because of their characteristics of environmental release. When they are used in fish farms or excreted directly on land, veterinary medicines are released directly into the environment. They also indirectly enter into the environment when manure containing excreted pharmaceuticals is applied onto land. Since veterinary pharmaceuticals are typically non point source pollutants, it is more difficult to effectively manage their contamination compared to human pharmaceutical contamination. Once released into the environment, veterinary pharmaceuticals and their metabolites have the potential to run off directly into surface waters or leach into groundwater. Household Human Use Excretion Sewage Liquid waste Effluent Disposal Pharmaceutical facilities Hospitals Solid Waste Wastewater Treatment Sludge Effluent Sludge Landfill/ Incineration Manure Soil Runoff Fish farm Surface water & Sediment Animal Use Groundwater Figure 1-1. Flow of human and veterinary pharmaceuticals into the environment. Occurrences and potential ecological effects Pharmaceuticals have been detected in the environment since the late 1990s. Studies in Austria, Australia, Brazil, Canada, France, Germany, Greece, Italy,

23 Chapter 1. Introduction 3 Japan, Korea, Spain, Switzerland, Sweden, the Netherlands, UK, and the United States have reported the occurrence of pharmaceuticals in the aquatic environment. Through 2004, nearly 43,000 samples were analyzed and pharmaceuticals were detected in nearly 33% of the samples (William, 2005). More than 100 pharmaceuticals from various therapeutic classes have been detected in sewage influent/effluent, surface water, groundwater, and even drinking water (Heberer and Stan, 1997; Stackelberg et al., 2004). In addition, it is worthwhile to note that there are many pharmaceutical substances that have never even been surveyed. Potential ecological effects from the presence of pharmaceuticals in the environment have generally focused on the following two concerns: 1) the release of antibiotics into the environment increases the risk of the emergence of antibiotic resistant microorganisms and promotes the spread of resistant genes, and 2) when drugs affecting hormonal systems reach organisms in nature, it may result in a reproductive disturbance in the ecosystem. Key elements of ERA Exposure Assessment Environmental risk assessment (ERA) is considered the best scientifically based approach for evaluating the potential effects of contaminants on communities and ecosystems. The process includes: problem formulation, exposure assessment, effects assessment, and risk characterization. Accurate exposure assessment is a key element of ERA. Environmental exposure is defined by the contaminant concentrations that occur in the environment. Environmental exposures can be estimated (the Predicted Environmental Concentration; PEC), using a model, or measured (the Measured Environmental Concentration; MEC). With the development of new analytical methodologies and the ability to quantify concentrations of analytes in the part per trillion (ppt) range, large scale monitoring programs have been conducted and are currently ongoing in many countries, including Korea. Exposure assessment, however, has been hampered by the continuing difficulties and expense involved in measuring low ppt concentrations of pharmaceuticals in the environment. In addition, real time monitoring data, although helpful, is generally site and time

24 4 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin dependent and provides only a snapshot view of contaminant concentration. However, exposure models have the capability of extrapolating the measured data over time and space. Thus, when faced with the task of assessing the environmental exposure of pharmaceuticals, the use of predictive tools, such as models, becomes essential. Modeling the Environmental Concentration of Pharmaceuticals In the United States, the Food and Drug Administration (FDA) requires an estimate of the concentration of human pharmaceuticals at the point of entry into the aquatic environment (Expected Introduction Concentration, EIC) (US FDA, 1998). Similarly, the European Union (EU) provides guidance on the initial calculation of the predicted environmental concentration (PECsurfacewater) in surface water (EMEA, 2006). The EIC of 1 ppb or the PECsurfacewater of 0.01 ppb is known as the action limit because higher values for EIC or PECsurfacewater will demand further environmental assessment and testing. Both EIC and PECsurfacewater are calculated using similar transfer functions that are derived based on a series of assumptions, including: - The pharmaceutical is evenly used over the year and throughout the geographic area - The pharmaceutical enters the environment only via the sewage water system - There is no biodegradation or retention of the pharmaceutical in sewage treatment plants - Metabolism in the patient is not taken into account - The effluent is completely mixed in the surface water (PECsurfacewater only) As seen above, the estimation models used in initial assessment tiers employ conservative assumptions for input values and these assumptions may be highly variable or uncertain. When additional input data is available to refine the assessment, or the initial assessment indicates that further investigation is warranted, more realistic modeling estimates need to be obtained.

25 Chapter 1. Introduction 5 For calculating PECs that attempt to reflect the complexity of the environmental, spatial and temporal variability, and the various fate and transport processes, spatially explicit models based on watersheds or catchments have been developed. Use of the catchment approach is particularly useful in estimating contaminants that enter the environment solely from human use. A GIS is often used to manage watershed and hydrologic data. Currently there are two GIS based models generally used for predicting PECs for human pharmaceuticals in the environment: PhATE TM (Anderson et al., 2004) and GREAT ER (Schowanek and Webb, 2002). Compared to human exposure modeling, exposure modeling for veterinary pharmaceuticals is even more complicated, due to the various environmental fates of veterinary pharmaceuticals. Exposure assessment models for soil, surface water and groundwater must be respectively developed and validated (Montforts, 2006) to allow for accurate veterinary environmental exposure modeling. Objectives of This Study The first objective of this report is to apply computerized exposure models to assess the environmental concentration of human and veterinary pharmaceuticals in the Han River and the Kyungahn stream, a major branch of the Han River. PhATE TM and SWMM are identified as appropriate exposure models for this study based on data availability, researchers previous experiences with models, and accessibility to models. All models are, by their nature, imperfect representations of the system they are intended to model. Models can be categorized into three groups based on their applicability: screening, primary and secondary (FOCUS, 1995). The models investigated in this study are screening level models. Screening level models do not intend to represent reality accurately, but rather, their purpose is merely to provide rapid predictions of the potential environmental fate of a compound. In this study, model predicted PECs are compared to field data that either have been published previously or are empirically measured during this study. A second objective of this study is to estimate the total environmental concentration of pharmaceuticals, from both human and animal use, by integrating the simulation results from PhATE TM and SWMM. As recognized

26 6 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin in the previous study (Park, 2006), there are many substances used for both human and veterinary purposes, such as trimethoprim. Although the introduction routes into the environment for veterinary pharmaceuticals are different from those for human use, both human and animal pharmaceuticals eventually end up reaching the surface water. Therefore, estimating the environmental concentration of such pharmaceuticals based solely on either human consumption or animal consumption results in an underestimation of environmental exposure. In this study, a workable framework that could estimate PECs for these dual usage pharmaceuticals is suggested. Finally, a third study objective is to perform an environmental risk assessment (ERA) for selected pharmaceuticals. Environmental hazards of the pharmaceuticals are obtained from either published literature or are evaluated using standard aquatic toxicity test methods. The toxicity information is used to determine predicted no effect concentrations (PNECs) of each pharmaceutical. Hazard quotients (the ratio of EC to PNEC) based on PECs are compared to those based on MECs. This exercise will demonstrate the applicability of modeling approaches in the risk assessment of pharmaceuticals in the environment. The Structure of the Report The first chapter includes a short, general introduction to the background and objectives of the report. In chapter 2, environmental occurrence studies conducted in Korea are summarized. Overviews of existing regulatory systems for pharmaceuticals in the environment are provided. In addition, an overview is given on computerized exposure models for human and veterinary pharmaceuticals focusing on models estimating PECs in surface water. Among those models, PhATE TM and SWMM are identified as appropriate exposure models for this study. Chapter 3 describes the estimation of PECs of human pharmaceuticals in the Han River using PhATE TM. The model predicted concentrations are compared to measured concentrations to help evaluate the reliability of PECs derived using PhATE TM.

27 Chapter 1. Introduction 7 Chapter 4 describes expanding SWMM for estimating veterinary pharmaceuticals in a sub urban area of the Kyungahn stream, which is an upper stream of the Han River. Chapter 5 examines the application of the PhATE TM SWMM combined model for predicting environmental concentrations of pharmaceuticals where human and veterinary pharmaceuticals coexist. In chapter 6, environmental risk assessment is carried out for selected pharmaceuticals. Hazard Quotients using the PECs and MECs are compared to each other to understand the relative merit and value of the modeling and monitoring approach. The potential benefit of using molecular level biomarkers to assess pharmaceutical toxicity is also discussed and methods are presented. Finally, in chapter 7 the overall conclusions and recommendations are presented.

28 8 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Chapter 2. Review and Background Myung Hyun Kim This introductory chapter will provide a review of existing research and regulations on pharmaceuticals in light of environmental risk. First and foremost, a comprehensive summary of the results from environmental occurrence studies conducted in Korea is presented. Until recently, research and monitoring data on the occurrence of pharmaceuticals in waters of Korea have been limited to studies focused on a few selected compounds in localized areas, mostly in the Han River. Secondly, regulations relating to the environmental risk assessment of pharmaceuticals are briefly described (Koschorreck, 2006). In the European Union, the United Sates, Canada, Australia, and Japan environmental protection is regulated. A specific guideline has been implemented in the United States and EU for human pharmaceuticals, and it is being developed in Canada and Australia, whereas Japan has fewer regulations in this area. However, a specific guideline for veterinary pharmaceuticals has been implemented in the EU, the United States, Australia, and Japan. Thirdly, an overview is given on computerized exposure models for human and veterinary pharmaceuticals focusing on the models estimating PECs in surface water. The chapter concludes that PhATE TM and SWMM, among those models, are appropriate exposure models for this study. 1. Occurrence Studies of Pharmaceuticals in the Korean Aquatic Environment A few studies are available in Korea regarding the occurrence of pharmaceutical substances in the aquatic environment. Although not a complete literature review, this chapter provides a summary of key studies conducted in Korea.

29 Chapter 2. Review and Background 9 An Approach for Developing Aquatic Environmental Risk Assessment Framework for Pharmaceuticals in Korea. KEI, 2006 This study aimed to prioritize pharmaceutical substances for aquatic environment risk assessment (ERA) and to provide a basis for developing the most proper and sensible strategy for pharmaceutical ERA in Korea. Since it may be nearly impossible to monitor and evaluate the occurrences and potential environmental risks of all active pharmaceutical ingredients, it is prudent to attempt to identify and develop a list of pharmaceuticals that deserve more immediate attention. A preliminary list of the primary pharmaceuticals of concern was developed from an inventory of human and veterinary pharmaceuticals manufactured in Korea. For human pharmaceuticals, the net amount of active ingredient produced was determined and used to rank its potential to exist in the environment. For veterinary pharmaceuticals, drug production amount along with its potential to enter the environment were considered in determining priority. As a part of ERA, the environmental concentrations of three antibiotics, roxithromycin, trimethoprim, and chloramphenicol, were empirically measured at a few sites in the Han River basin. Antibiotics were more frequently detected at increased levels in effluent samples, and in samples collected during the low flow season. For trimethoprim and chloramphenicol, the levels observed in surface water during the low flow season were on average 108 and 31 ng/l, respectively. These levels were comparable to the levels measured in the municipal effluents, i.e., on average 80 and 37 ng/l, respectively, suggesting the presence of other sources upstream, e.g., livestock wastes. For roxithromycin, surface water levels were about an order of magnitude lower than the effluent levels. The study results indicated that the flow rate conditions in the body of water sampled appear to have an immediate impact on the concentrations. Water samples collected from STP influent or effluent water showed higher levels of the test compounds than the levels observed in surface water samples.

30 10 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Monitoring Pharmaceuticals in the Four Major River Basins in Korea. NIER, 2006 Most of the research and monitoring data on the occurrence of pharmaceuticals in Korea has focused on limited areas, mostly in the Han River area. Recently, however, researchers with the NIER s surveillance study have developed analytical methods for 17 pharmaceuticals, including many commonly reported human and veterinary pharmaceuticals, and have conducted a nationwide survey. The study covered the four major river basins in Korea: the Han River, the Nakdong River, the Kum River, and the Youngsan River. Water samples were collected from surface water, wastewater treatment plant influents and effluents, and livestock wastewater treatment facilities, where accessible. In this study, 16 of the 17 pharmaceuticals studied were detected. Notably, lincomycin, ibuprofen, and sulfamethazine were detected at concentrations greater than 1 μg/l in some samples. The study also raised concerns that pharmaceuticals are only partially eliminated through the wastewater treatment process and, therefore, residual amounts could reach ambient waters. Ecotoxicological Risk of Pharmaceuticals from Wastewater Treatment Plants in Korea: Occurrence and Toxicity to Daphnia Magna. Han et al., 2006 This study investigated the ecotoxicological risk of seven pharmaceuticals detected in wastewater treatment plants. The overall ecotoxicological effect of pharmaceutically active compounds (PhACs) detected in the effluents of wastewater treatment plants (WWTPs) to Daphnia magna was investigated using biological and chemical analyses. The bioassay results showed median lethal concentrations and no observed effect concentrations ranging from a few to tens of ppm levels for nine PhACs in 48 h acute and 21 d chronic tests. The effects of mixtures of pharmaceuticals also were examined in additional acute and chronic tests, however, these studies showed no significant increase in toxicity except for a slightly increased combined effect of approximately twofold. The residual concentrations of nine PhACs were detected at

31 Chapter 2. Review and Background 11 concentrations ranging from 10 ng/l to 89 mg/l in the influents, and from 10 ng/l to 11 mg/l in the effluents from four metropolitan cities in South Korea between January and November of Through repeated investigations of the influents and the effluents from different WWTPs, relatively higher removal efficiencies ( %), compared to those of previous surveys performed in other countries, were observed for most pharmaceuticals, with the exception of acetaminophen (8.7 %). The present study showed no significant risk effects from the effluents from WWTPs containing pharmaceuticals (i.e., hazard quotients were <1) even at the 95th percentile concentration range, although a risk assessment factor of 1,000 was applied. Therefore, it can be concluded that the potential risk of environmentally detected pharmaceuticals should be monitored carefully with additional bioassay data, because many uncertainties still exist in the determination and toxicity of metabolites in water environments. No significant risk was observed, however, from the selected PhACs in the effluents from WWTPs discharged into surface waters. The Effects on Aquatic Organisms Induced by Pharmaceutical Residues in the Han River. Kim et al., Kim et al. ( ) have conducted two separate investigations in the Han River basin in 2005 and In their 2005 study (Kim et al., 2005), they analyzed 12 pharmaceutical substances from surface water samples and WWTPs of the Han River basin. The substances analyzed were acetaminophen, 1,7 dimethylxanthine, caffeine, carbamazepine, diltiazem, cimetidine, and 6 sulphonamide antibiotics including sulfamethoxazole, trimethoprim, sulfachloropyridazine, sulfathiazole, sulfamethazine, and sulfadimethoxine. They reported a significant elimination of acetaminophen in the effluents from STPs in the Han River. The occurrences of caffeine in the effluents were at similar levels compared to the levels reported in other countries. Carbamazepine in STP effluents was slightly lower than found in Canada or Germany. Regarding the sulphonamide antibiotics, sulfachloropyridazine was not detected in Canada while it was measured from the effluents from STPs in Korea. Sulfadimethoxine was detected at a higher level than in the USA while at a lower level than in Germany. Overall,

32 12 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin pharmaceutical levels detected in the Korean environment were not comparable to levels found in other countries. In the 2006 study, Kim et al. focused on 13 veterinary antibiotics that have a greater possibility of occurring in the Han River basin. They collected surfacewater samples at 18 sites from the Han River and the South and North upstream areas of the Han River. Samples were collected three times in May, August, and September These sampling events reflect normal, high, and low flow conditions, respectively. Influent and effluent waters from sewage treatment plants were also analyzed. In the north Han River, enrofloxacin was detected in all water samples during the normal flow season. In the south Han River, tetracycline was detected in high frequency in the high flow season. In the Han River, florfenicol was detected in all surfacewater during normal flow season. More veterinary antibiotics were detected at a greater frequency in upstream Han River compared to downstream, probably resulting from the proximity to concentrated animal feeding operations. Table 2 1, 2 2a, and 2 2b summarize the results presented in the key studies mentioned above. The majority of samples in these studies were analyzed with an HPLC MS MS system, however, GC MS was utilized in the study by Han et al (2006). Table 2 1 lists the limit of detection (LOD) for the pharmaceuticals analyzed in the studies.

33 Chapter 2. Review and Background 13 Table 2-1. Limits of detection (LOD) for the pharmaceuticals analyzed Unit: µg/l NIER, 2006 Han et al., 2006 Kim et al., 2005 Kim et al., 2006 Park, 2006 Park, 2007* 1,7 Dimethylxanthine Acetaminophen Caffeine Carbadox Carbamazepine Chloramphenicol Chlortetracycline Cimetidine Ciprofloxacin Clofibric acid 0.01 Diclofenac Diltiazem Enrofloxacin Erythromycin Florfenicol Gemfibrozil 0.5 Ibuprofen Lincomycin Naproxen Oxytetracycline Roxithromycin Salicylic acid 0.1 Sulfachloropyridazine Sulfadimethoxine Sulfamethazine Sulfamethoxazole Sulfathiazole Tetracycline Trimethoprim Tylosin Virginiamycin *: Analyzed in the current study.

34 14 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Table 2 2a and 2 2b list the concentrations of pharmaceuticals detected in samples collected from the Korean aquatic environment. Table 2 2a includes samples from influents and effluents of WWTP and surfacewater. Thirty one pharmaceuticals were detected at various levels. Table 2 2b presents the concentrations of pharmaceuticals detected in samples related to livestock wastewater treatment plants or surfacewater adjacent to discharges from livestock WWTPs. In calculating the mean concentration value for each pharmaceutical, a factor of one half of LOD, or LOD/2, was applied to samples with concentration levels below their respective LOD, as listed in Table 2 1.

35 Chapter 2. Review and Background 15 Table 2-2a. Summary of occurrence for pharmaceuticals analyzed in Korea (WWTPs and surfacewater) Compounds Reference Wastewater treatment (µg/l) Surfacewater (µg/l) (Number of Influent Effluent samples*) Mean Min Max Mean Min Max Mean Min Max 1,7 Dimethylxanthine Kim et al., 2005 (a) NIER, 2006 (b) <LOD <LOD <LOD Acetaminophen Han et al., 2006 (c) <LOD <LOD Kim et al., Park, 2007 (d) Caffeine Kim et al., Carbadox NIER, 2006 <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Kim et al., 2006 (e) <LOD <LOD <LOD <LOD <LOD <LOD Carbamazepine Han et al., <LOD <LOD Kim et al., Chloramphenicol Park, 2006 (f) Chlortetracycline NIER, <LOD <LOD <LOD <LOD <LOD <LOD Kim et al., <LOD <LOD <LOD Cimetidine Kim et al., Ciprofloxacin NIER, Clofibric acid Han et al., <LOD Diclofenac NIER, Han et al., <LOD <LOD Diltiazem Kim et al., Park,

36 16 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Compounds Reference Wastewater treatment (µg/l) Surfacewater (µg/l) (Number of Influent Effluent samples*) Mean Min Max Mean Min Max Mean Min Max NIER, <LOD <LOD <LOD Enrofloxacin Park, Kim et al., Erythromycin NIER, <LOD <LOD <LOD Florfenicol Park, Kim et al., Gemfibrozil Han et al., 2006 <LOD <LOD <LOD <LOD <LOD <LOD Ibuprofen NIER, Han et al., <LOD <LOD Lincomycin NIER, Naproxen NIER, Oxytetracycline NIER, 2006 <LOD <LOD <LOD <LOD <LOD <LOD Kim et al., <LOD <LOD <LOD Roxithromycin NIER, 2006 <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Park, Salicylic acid Han et al., <LOD Sulfachloropyridazine Kim et al., <LOD <LOD <LOD Kim et al., Sulfadimethoxine Kim et al., Kim et al., 2006 <LOD <LOD <LOD <LOD <LOD <LOD

37 Chapter 2. Review and Background 17 Compounds Reference Wastewater treatment (µg/l) Surfacewater (µg/l) (Number of Influent Effluent samples*) Mean Min Max Mean Min Max Mean Min Max Sulfamethazine Sulfamethoxazole Sulfathiazole NIER, 2006 <LOD <LOD <LOD <LOD <LOD <LOD Kim et al., 2005 <LOD <LOD <LOD <LOD <LOD <LOD Kim et al., NIER, Kim et al., Park, Kim et al., NIER, Kim et al., <LOD <LOD <LOD <LOD <LOD <LOD Kim et al., Tetracycline Kim et al., <LOD <LOD <LOD Trimethoprim NIER, Kim et al., Park, Park, Kim et al., Tylosin NIER, 2006 <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Virginiamycin Kim et al., 2006 <LOD <LOD <LOD <LOD <LOD <LOD * Number of samples: a : influent (10), effluent (10), surfacewater (24), b : carbamazepine, salicylic acid, gemfibrozil, acetaminophen (16), diclofenac, ibuprofen, clofibric acid (17), c : influent (12), effluent (12), surfacewater (24, inclusion of Han River and Kyungahn stream concentration), d : influent (30), surfacewater (68, inclusion of Han River and Kyungahn stream concentration), e : influent (18), effluent (20), surfacewater (106, inclusion of Han River, South Han River, North Han River, and Kyungahn stream concentration), f : influent (16), effluent (12), surfacewater (67, inclusion of Han River and Kyungahn stream concentration).

38 18 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Table 2-2b. Summary of occurrence for pharmaceuticals analyzed in Korea (livestock WWTPs and neighboring surfacewater) Unit: µg/l Compounds Reference Livestock wastewater treatment Surfacewater (Number of Influent Effluent samples*) Mean Min Max Mean Min Max Mean Min Max Acetaminophen NIER, 2006 (a) Carbadox NIER, 2006 <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Kim et al., 2006 (b) Chlortetracycline NIER, <LOD <LOD <LOD <LOD <LOD <LOD Kim et al., Ciprofloxacin NIER, Diclofenac NIER, <LOD <LOD <LOD Enrofloxacin NIER, Kim et al., Erythromycin NIER, 2006 <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Florfenicol Kim et al., Ibuprofen NIER, Lincomycin NIER, Naproxen NIER, Oxytetracycline NIER, <LOD <LOD <LOD Kim et al., Roxithromycin NIER, 2006 <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Sulfachloropyridazine Kim et al.,

39 Chapter 2. Review and Background 19 Compounds Reference Livestock wastewater treatment Surfacewater (Number of Influent Effluent samples*) Mean Min Max Mean Min Max Mean Min Max Sulfadimethoxine Kim et al., Sulfamethazine NIER, Kim et al., Sulfamethoxazole NIER, Kim et al., Sulfathiazole NIER, Kim et al., Tetracycline Kim et al., Trimethoprim NIER, 2006 <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Kim et al., Tylosin NIER, <LOD <LOD <LOD <LOD <LOD <LOD Virginiamycin Kim et al., 2006 <LOD <LOD <LOD * Number of samples: a : Influent (10), effluent (10), surfacewater (16), b : 9

40 20 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin 2. Environmental Safety Regulations for Pharmaceuticals 2.1 The European Union Human Pharmaceuticals Legally, all new European drug approvals need to satisfy the requirement of Directive 2004/27/EC (European Union, 2004) regarding environmental risk assessment. As shown in Figure 2 1 (Adler et al., 2007), the legislation concerning the requirements for the registration of human pharmaceuticals in Europe was driven by the European Commission (EC), based on the relevant Directive 93/39/EEC (European Union, 1993) and its changes up to 2001/83/EC and 2004/27/EC (European Union, 2001 and 2004) (Reinhard et al., 2006) Guideline comes into force (EMEA/CHMP/SWP/4447/00) 2004 EU Pharma review: ERA obligatory for any new application (2004/27/EC, amending 2001/83/EC 2004/28/EC, amending 2001/82/EC) nd Draft Guideline st Draft Guideline Consolidation of Pharmaceutical legislation Discussion paper for ERA of HMPs (2001/82/EC, 2001/83/EC) 1993 Directive 93/39/EEC Introduced environmental safety Requirements for human medicines 1995 Guideline for medicinal products Containing or consisting of GMOs Figure 2-1. Progress of European Union (EU) legislation and technical guideline on ERA of human pharmaceuticals. The 1993 Directive vaguely indicated the importance of an environmental risk assessment by stating that the dossier should contain if applicable, reasons for any precautionary and safety measures to be taken for the storage of the medicinal product, its administration to patients and for the disposal of

41 Chapter 2. Review and Background 21 waste products, together with an indication of any potential risks presented by the medicinal product for the environment. The initial impact of the directive merely required notification of potential environmental risks in broad terms. The latest versions of the directive (European Union, 2001 and 2004) broadened the impact by specifying its goals in several articles: Article 1, 28(1) (assess) any risk of undesirable effects on the environment and Article 8 Evaluation of the potential environmental risks posed by the medicinal product. This impact shall be assessed and, on a case by case basis, specific arrangements to limit it shall be envisaged. (g) Reasons for any precautionary and safety measures to be taken for the storage of the medicinal product, its administration to patients and for the disposal of waste products, together with an indication of potential risks presented by the medicinal product for the environment. In short, the initial phrasing of the directive required only a notification of any environmental risk with no further specification concerning what possible actions may be taken if such risks are identified. The recent version of the directive clearly names the environmental risk assessment (ERA) as a goal itself, followed by risk management actions on a case by case basis, if needed. The risk management options are described vaguely as specific arrangements, which leaves flexibility for the regulators and manufacturer in proposing such arrangements. In order to fulfill the requirement of Directive 2004, the EMEA document CHMP/SWP/4447/00 has been adopted as a technical guideline (Figure 2 2).

42 22 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Figure 2-2. Guideline on the Environmental Risk Assessment of Medicinal Products for Human Use (EMEA, 2006a). The requirements of the guideline changed significantly over time from the first published drafts in 1994/1995 to the latest final version in 2006 (EMEA, 2006a). Major changes were:

43 Chapter 2. Review and Background 23 - Calculation of the predicted environmental concentration (PEC) by market penetration factor instead of the company s market predictions (2003). - Introduction of a two phased experimental strategy based on hazard characteristics (not risk characterization) (2003). - Bioaccumulation study in fish (Kow > 1,000) (2003). - Degradation in soil (Koc in sludge > 10,000) (2003). - Long term distribution and degradation study in aquatic sediments instead of screening tests for biodegradation and basic physicochemical data (2005). - Long term ecotoxicology studies in daphnids and fish instead of acute tests (2005). ERA procedures for pharmaceuticals in accordance with the EMEA regulatory guideline is a step wise, tiered procedure starting with rough estimates and progressing to more elaborate, refined methods if a potential risk cannot be excluded. The assessment may be terminated either when sufficient information is available, indicating that the product/compound is unlikely to represent an environmental risk, or when a risk has been identified and sufficiently characterized. A risk quotient (RQ) is usually calculated from a predicted (or measured) environmental concentration (PEC or MEC) and a predicted no effect concentration (PNEC). PEC can be calculated from a combination of estimates of the amount of consumption or sales, expected route of entry into the environment, and physico chemical properties. In higher tiers, fate in the environment (such as biodegradability, bioconcentration, and adsorption to soil or sediment) is taken into account. A PNEC is obtained by dividing the lowest no observed effect concentration (NOEC) for the most sensitive species with an appropriate safety factor. Accordingly, the outcome of an ERA is dependent on amount used, environmental fate, and the ecotoxicity of the compound in question. More detailed requirements and procedures can be found in the guideline.

44 24 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Veterinary Pharmaceuticals In the European Union, all veterinary medicinal products (VMPs) 1 require an environmental impact assessment (EIA) in accordance with Directive 2001/82/EC, as amended by Directive 2004/28/EC. The only VMPs excluded from this requirement are those applications that deal with generic drugs. The Directive includes a new definition of the risks related to the use of veterinary medicinal products and introduces a definition of a risk/benefit analysis (EMEA, 2006). For veterinary medicinal products, the Directive stated that the environmental assessment should be carried out in two phases. In the first phase the extent of environmental exposure should be estimated, while in the second phase the fate and effects of the active residue should be assessed. The basic framework provided by the Directive was further characterized by guidelines published by CVMP in 1997, providing guidance to both applicants and to the regulators on how the assessment of environmental safety should be carried out. The CVMP guidelines have in the meantime been replaced with VICH (International Cooperation on Harmonization of Technical Requirements for Registration of Veterinary Products) 2 guidelines. Environmental impact assessments (EIAs) for veterinary medicinal products (VMPs) Phase I (GL 6) was published in 2000, and the Phase II document (GL 38) came into force in October 2005 (EMEA, 2006). For drugs that qualify for a VICH Phase I assessment only, environmental fate and effects data are not required for evaluation of the potential risk of drug residues on nontarget species in the environment. In Phase I, if the EICaquatic value and PECsoil for the VMP entering the environment was less than 1 μg/l and 100 μg/kg, respectively, then the assessment may stop at Phase I (VICH, 2000). For products that require a full assessment (VICH Phase II), environmental fate and effects study data are required to assess the potential impact of veterinary drug residues (parent and metabolites, in 1 The term VMP is unique to the European Union. In the United States, the term veterinary drug is used. In the European Union, VMPs are regulated by Directorate General (DG) III of the European Commission. Feed additive growth promoters are authorized by a different regulatory authority, DG VI. In the United States, all drugs intended for use in animals are regulated by FDA/CVM. 2 VICH is a trilateral program aimed at harmonizing technical requirements for the registration of veterinary drugs among the regions that are a party to VICH.

45 Chapter 2. Review and Background 25 certain situations) on nontarget species in aquatic and terrestrial ecosystems. A Phase II assessment of a given veterinary drug residue includes studies of (1) physicochemical properties (e.g., water solubility, octanol/water partitioning), (2) binding to soils, (3) biodegradation in soil or aquatic test systems, and (4) effects of the drug residue on select aquatic (algae, daphnia, fish) and terrestrial species (soil microflora, plants, earthworms, and possibly dung beetles and flies). As of 2006, the VICH Phase I and II guidelines have been adopted officially in the European Union, United States, Japan, Australia, and New Zealand. 2.2 The United States The Food and Drug Administration (FDA) is the primary federal agency responsible for the regulation of pharmaceuticals and personal care products in the United States. The agencyʹs primary role is the assessment of applications for clinical investigation and the marketing of drugs and devices for human and animal use. As part of this assessment, the FDA evaluates the potential environmental impact of use of drugs in accordance with the US National Environmental Policy Act of 1969 (NEPA). This law requires the US government to consider the potential environmental impact of ʺactionsʺ that the government takes. Approval of a marketing application for a human or animal drug is considered to be an action under the FDA specific NEPA regulations, and therefore the environmental impact of the marketed product must be assessed. Human Pharmaceuticals The FDA implemented a guidance document on the ERA of human pharmaceuticals in Specific product types are specified in the Guidance for Industry: Environmental Assessment of Human Drug and Biologics Applications (US FDA, 1998) and, in particular, some pharmaceuticals, which anticipate no expected impact on the environment, are categorically excluded from assessment and data requirements (Park, 2006). There are a number of categorical exclusions described in the US Code of Federal Regulations (21 CFR Part 25.31). These include situations in which the manufacturer can demonstrate that: (1) there is no increase in use of the active

46 26 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin pharmaceutical moiety; (2) there is an increase in the use of the active moiety, but the release into the aquatic environment is less than 1 μg/l (or EIC); (3) the drug is a naturally occurring substance, and there is no significant change in concentration or distribution of the compound in the environment; or (4) the drug is in the investigational stage of clinical development. However, if extraordinary circumstances exist, then a categorical exclusion is not appropriate, and an environmental assessment should be submitted with the application. A brief overview of the FDAʹs environmental assessment (EA) procedure for human drugs can be found elsewhere (Park, 2005). Veterinary Pharmaceuticals The US statute that provides the legal basis for requiring ecotoxicity assessments for veterinary drugs is the National Environmental Policy Act. The center for Veterinary Medicine of the FDA regulates the manufacture and distribution of animal drugs and food additives. When an EA is needed for a veterinary drug, the information required for this document is to be decided in advance between the applicant and FDA/CVM in accordance with regulation 21 CFR part This regulation provides no guidance on how the EA is to be conducted (data requirements and decision making criteria), in contrast to the detailed information prescribed in the 1998 CVMP Note for Guidance for inclusion in the EIA. As noted, the US participated in the development of the VICH guidelines. The guidelines were split into two phases. For Phase I, environmental fate and effects studies are not required for the assessment. But, if suitable biodegradation data are available, then the Phase I assessment allows these data to be used to refine the PEC for soil exposure. A Phase I assessment should suffice for veterinary drugs for which the concern is deemed minor based on limited environmental exposure. In contrast, a full environmental fate and effects data package is required for a drug (or metabolite of concern) that advances to Phase II. The premise is that Phase II assessments are required when exposures are sufficient to generate potential concerns. Both phases have been completed and formally adopted. Phase I has been in force in the United States since 2000/2001. Phase II was completed, and the VICH formally signed off on the document, in September Since that

47 Chapter 2. Review and Background 27 time, the guideline has been adopted and implemented in the United States as well as in the European Union, Australia, and New Zealand. 2.3 Canada Human pharmaceuticals and veterinary medicines require an environmental risk assessment under the New Substances Notification Regulations (NSNR) of the Canadian Environmental Protection Act 1999 (CEPA 1999). The environmental risk assessment of substances in F&DA products started in September The Environmental Assessment Unit in Health Canada is responsible for these environmental risk assessments. At present, the evaluation is only related to the assessment of chemicals. Human Pharmaceuticals Health Canada requires the assessment of medicinal products regulated under the Food and Drug Act (F&DA) with respect to their potential effect on the environment (Park, 2006). If the substance is not included on the Domestic Substances List (DSL), it is considered ʺnewʺ and is subject to the New Substance Notification Regulations (NSNR) of the CEPA. The intention of these regulations is to implement a preventive approach to new substance management by requiring data on the substance before it can be manufactured or imported into Canada. Based on this data, Health Canada and Environment Canada conduct an assessment and determine: the risk of the substance to human health and the environment, whether controls may be warranted to mitigate risk, or whether additional information is required. The New Substances Notification process can ultimately result in the addition of a substance to the DSL. CEPA could be considered a ʺsafety netʺ in that it captures many of the substances not screened for environmental impacts under other federal acts or regulations. CEPA endeavors, however, to avoid legislative duplication by recognizing other federal acts and regulations that provide equivalent protection to the environment and human health. The original CEPA (CEPA, 1988) did not require notification of these new substances. In addition, at that time the Food and Drug Act (F&DA), the principal legislation regarding the safety and efficacy of foods and drugs in

48 28 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Canada, was considered to be exempt from CEPA. As a result, substances contained in commodities such as human drugs, biologics, veterinary drugs, cosmetics, novel foods, food additives, natural health products, and medical devices were not notifiable to the NSNR under CEPA 1988 (Andrew, 2007). Veterinary Pharmaceuticals The VICH guidelines are not in force in Canada. Health Canada is conducting a consultative process to determine what are the most appropriate regulations for these substances. Through the Veterinary Drugs Directorate (VDD), Health Canada evaluates and monitors the safety, quality and effectiveness, sets standards and promotes the prudent use of veterinary drugs administered to food producing and companion animals (Health Canada, 2007). 2.4 Australia Human Pharmaceuticals Australian authorities start asking for environmental risk assessments of human medicinal products. Veterinary Pharmaceuticals In Australia veterinary chemicals are assessed under the Agricultural and Veterinary Chemicals Act administered by the Australian Pesticides and Veterinary Medicines Authority (APVMA). The legal basis for product acceptance is that the use of the proposed product ʺwould not be likely to have an unintended effect that is harmful to animals, plants or things, or to the environmentʺ. If serious environmental risk of the veterinary chemical is identified, however, marketing authorization can be denied. Legislation does allow label restrictions, warning statements, etc., to mitigate risk (though this usually applies only to pesticides). Focus on new products, no update of ecotox dossier in renewals necessary, assessment of applications for extensions in use where significant increases in environmental exposure may result. The Department of Environment started risk assessments for both pesticides and veterinary medicines in late Part 7 (Environment) of Veterinary Manual of Data Requirements and Guidelines (MORAG) was

49 Chapter 2. Review and Background 29 released in 1997 by Australian registration authority APVMA. VICH Phase I came to into force in July 2001 (with some qualifications), VICH Phase II followed Not too many risk mitigation measures so far with the exception of the macrocyclic lactones with known effects on dung fauna. 2.5 Japan Human Pharmaceuticals At present, there is no legal requirement for an environmental risk assessment of human pharmaceuticals in Japan. However, the Ministry of Health, Labour and Welfare (MHLW) is investigating the fate and effects of pharmaceuticals in the environment. Veterinary Pharmaceuticals The regulation of VMPs in Japan is linked to the VICH. VMPs are regulated for environmental safety as described in the Environmental Impact Assessment for VMPs, Phase I in 2000 and phase II in Japan raised public opinion over VICH GL 38 (ecotoxicity phase II) in Currently, the regulation procedure for VMPs in Japan requests no data on ecotoxicity. The municipal law will be revised after the agreement reached in the meeting for VICH (Yoshioka, Yoshitada, 2007). The Ministry of Agriculture, Forestry and Fisheries releases regulation covering the description of the environmental risk assessment on veterinary pharmaceuticals. Only new products have to be assessed. It has not been decided yet if the new regulation will envisage risk mitigation measures. A serious risk for the environment can be a reason for the refusal of a market authorization. The implementation of VICH Phase II and I is on its way. A simple guideline for exposure estimation to go along with the VICH document will be developed. Table 2 3 summarizes the current status of pharmaceutical ERA regulations in the EU, the US, Canada, Australia, and Japan as described above.

50 30 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Table 2-3. Pharmaceutical ERA regulations in other countries EU USA Canada Australia Japan Human Pharmaceuticals EMEA (CHMP) Directive 2004/27/EC Guideline on the Environmental Risk Assessment of medicinal Products for Human Use, 2006 Action limit: 0.01 ppb (as PECsurfacewater) US FDA Guidance for Industry: Environmental Assessment of Human Drug and Biologics Applications, 1998 Action limit: 1 ppb (as EIC) F&DA New Substances Notification Regulations (NSNR) Final stage to implement ERA for human drugs Start asking for ERA of HMP No legal requirement for an ERA Investigating fate and effects of PIE in MHLW Veterinary Pharmaceuticals EMEA (CVMP) Directive 2004/28/EC VICH (International Cooperation on Harmonization of Technical Requirements for Region of Veterinary Medicinal Products) Guidelines: Environmental Impact Assessments (EIAs) for Veterinary Medicinal products Phase I (GL 6) and Phase II (GL 38) Phase I: 2000, Phase II: 2005 US FDA VICH Guidelines F&DA NSNR VICH Guidelines are not in forced, but expected to be effective in The Department of Environment started RA for both pesticides and VMP in late In 1997, Veterinary Manual of Data requirements and Guidelines (MORAG) was released. VICH Phase I: July, 2001 VICH Phase II: 2006 Ministry of Agriculture, Forestry and Fishers Link to the VICH, expected to be effective in 2007

51 Chapter 2. Review and Background Models for Predicting PECs of Pharmaceuticals 3.1 Models for Predicting PECs of Human Pharmaceuticals The detection of low levels of pharmaceuticals in rivers and streams, drinking water, and groundwater has raised questions as to whether these levels may affect human health. River catchment models, such as PhATE TM (US) and GREAT ER (Europe), have been developed recently and used for predicting the environmental concentrations of pharmaceuticals (Schowanek and Webb, 2002; Cunningham et al., 2004). a. PhATE TM model (Pharmaceutical Assessment and Transport Evaluation) In the US, the pharmaceutical industry trade association, PhRMA (Pharmaceutical Research and Manufacturers of America) developed a watershed specific model to predict environmental concentrations from patient use (Anderson et al., 2004). The PhATE model is a watershed based approach and was developed as a tool to more realistically estimate concentrations of active pharmaceutical ingredients (APIs) discharged to US surface waters through consumption of medicines (GSK online, 2007). The model may be used in screening mode using conservative input values or in a more realistic mode using available fate data. The model is based on 11 watersheds selected to be representative of most watersheds and hydrologic regions of the US (Figure 2 3). The PhATE model predicts concentrations at low (7Q10) and mean flow for WWTP effluents and surface waters. The current version of PhATE does not consider veterinary pharmaceuticals or septic system discharge because these releases are through pathways other than a WWTP.

52 32 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Source: Anderson et al., Figure 2-3. The locations of the 11 watersheds included in PhATE TM. The PhATE model is comprised of two modules the exposure module and the human health effects module. 3 The exposure module generates PECs and the human health effects module estimates PNECs on human health for drinking water and fish consumption. Overview of the Exposure Module The exposure module estimates concentrations of APIs in surface waters of a given watershed. PhATE begins with a mass of a compound distributed homogeneously across the population of the United States. The starting mass is assumed to be the annual sales volume (in kg/year) and the mass entering the watershed(s) included in PhATE is proportional to the number of people served by Publicly Owned Treatment Works (POTWs) in the watershed(s). A key aspect of the PhATE model is that it allows the parent compound to be transformed into an inactive form. The transformation (or 3 This content was extracted from Pharmaceutical Assessment and Transport Evaluation Model (PhATE) User s manual (PhRMA, 2005).

53 Chapter 2. Review and Background 33 loss) can be represented as either instantaneous loss or first order decay, depending upon where the loss is assumed to occur. Loss of a compound may occur at four places in the model. 1) Human use losses: These losses are assumed to occur before the parent compound reaches a POTW and include mechanisms such as metabolism. 2) POTW losses: These losses are assumed to occur inside the POTW during treatment and include mechanisms such as biological decay and partitioning to sludge. 3) In stream losses: These losses include all of the potential fate mechanisms that could affect a compound after it has been discharged from the POTW and before it enters a Drinking Water System (DWS). 4) DWS losses: These losses are assumed to occur inside the DWS during treatment. The current version of PhATE (Version 2.0) only allows for a one time instantaneous decay for the human, POTW, and DWS loss categories. However, a user may select either instantaneous or first order decay for instream loss. To enable instantaneous loss a user must specify the amount (fraction) of parent compound lost at one or all of the four places where PhATE assumes losses may occur. Estimation of the first order in stream loss is more complicated and requires user supplied information about the first order decay rate coefficient in a water body. The mass entering a POTW is divided by the average flow rate of the POTW to estimate the concentration of a compound in the influent of the POTW. The mass entering a POTW can either be conserved or some can be lost depending upon the compound and the type of treatment at each POTW. As described above, PhATE allows users to specify a fractional loss for each of the different POTW treatment types. The mass leaving a POTW is divided by the flow rate of the POTW to estimate the concentration of a compound in the effluent of the POTW. The mass of compound leaving a POTW then enters a surface water body. As described above, mass can be lost in surface water due to a variety of

54 34 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin physical and biological mechanisms and PhATE allows the user to specify whether those mechanisms are instantaneous or first order. If in stream loss is assumed to be instantaneous, the loss at each POTW is taken prior to the compound entering the surface water. If loss is assumed to be first order, then in stream loss is taken within each segment. Overview of Human Health Effects Module The human health effects module is the second part of version 2.0 of the PhATE model. The human health effects module estimates a predicted no effect concentration (PNEC) for only drinking water and surface water from which fish is consumed. The fish consumption PNEC does not refer to the concentration in fish tissue, but rather to the concentration in the water column that would result in an effect if humans consumed fish from that water. To derive PNECs the PhATE model uses the standard human health risk assessment formula that combines the acceptable daily intake (ADI) of a compound with the amount of water (or fish) humans are assumed to consume to derive an allowable concentration in water (or fish). That allowable concentration is the PNEC and is derived using Equation 2 1. ADI BW AT PNEC = (Eq. 2 1) CR EF ED Where, PNEC ADI BW AT CR EF ED = predicted no effect concentration (µg/l) = acceptable daily intake (mg/kg day) = body weight (kg/person) = averaging time (days) = consumption rate (l/person day for water; g/person day for fish) = exposure frequency (days/year) = exposure duration

55 Chapter 2. Review and Background 35 As written, Equation 2 1 calculates a PNEC for drinking water in units of µg/l and for fish consumption in units of µg/kg. ADI can be either obtained from another source or calculated from a no observed effect level (NOEL) or lowest observed effect level (LOEL) along with a series of uncertainty factors. To back calculate water concentrations for fish consumption in units of mg/l a bioconcentration factor (BCF) is added to the denominator, see Equation 2 2. PNEC fish ADI BW AT ( mg / L) = (Eq. 2 2) BCF IngRfish FSfsih EDfish Where, PNECfish ADI BW AT BCF IngRfish FSfish EDfish = predicted no effect concentration fish = acceptable daily intake (mg/kg day) = body weight (kg/person) = averaging time (days) = bioconcentration factor relating fish tissue concentration (mg/kg) to water concentration (mg/l). = ingestion rate of fish = fraction from source fish = exposure duration fish The PEC at the effluent of a DWS (estimated in the exposure module) can be compared to a PNEC in drinking water (developed in the human health effects module) to evaluate the potential risk associated with drinking the water. Overview of the Information Needs of PhATE TM To estimate PECs and PNECs, PhATE relies upon information stored in a database as well as information supplied by the user. Information about the 11 watersheds of the US (including hydrological, POTW and DWS data) is included in the databases that come with the model. PhATE requires the user to input compound specific information, including physical and chemical properties, toxicological information, the fraction of an API that may be lost during human use, in a POTW, in stream or in a DWS, and key

56 36 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin characteristics that determine the potential exposure of people using drinking water or consuming fish. Table 2 4 summarizes key inputs for executing PhATE. Table 2-4. Summary of key inputs for PhATE TM Model Compound use and characteristics Publicly owned treatment works (POTWs) Dams and reservoirs Segment detail Input parameter Physical and chemical properties, Toxicological information, Usage per capita, in stream first order loss, human loss (e.g., metabolism), POTW removal efficiency for each POTW treatment type Name, location, POTW treatment type, population served, flow rate Name, volume, surface area, length, depth System segment number, segment sequencing, mean flow, low flow (7 day, 10 year low flow; 7Q10), meanflow velocity, low flow velocity, length, depth, width Output PhATE results are provided as a print results, graph results, and map results (Figure 2 4 to Figure 2 6). In print results, PhATE allows users to display results in both graphic and tabular formats. Fifteen tabular reports are available: ADI input summary, chemical input summary, dose calculation inputs, drinking water concentrations, drinking water system summary, model run information summary, PNEC summary, POTW masses and concentrations, POTW system summary, surface water concentrations, surface water travel times, sedimentation loss in reservoirs, surface water concentrations by reservoir, surface water reservoir summary by watershed, and travel time in reservoirs.

57 Chapter 2. Review and Background 37 Figure 2-4. PhATE TM Model output (print results). Figure 2-5. PhATE TM Model output (graphic results).

58 38 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Figure 2-6. PhATE TM Model output (map results). PhATE TM Application and Corroboration In order to demonstrate the utility of PhATE TM as a predictive tool, PhATE TM developers compared model results with field data (Anderson et al., 2004). In the study, field data and model PECs for the same locations were compared on a point by point basis (Figure 2 7). In the next phase of model corroboration 4, the cumulative probability distributions of field measured concentrations were compared to the distributions of PECs for all segments included in PhATE TM. Simulations using three surrogate compounds showed that PECs generated by PhATE TM were generally within an order of magnitude of measured concentrations and that the cumulative probability distribution of PECs for all watersheds included in PhATE TM was consistent with the nationwide distribution of measured concentrations of the surrogate compounds. 4 Anderson et al. (2004) claimed that model verification or validation is impossible for environmental models and that some lesser degree of demonstrating the model s utility is all that can be achieved. Rather, corroboration is more appropriate as the term corroboration recognizes that a model can never be shown to be absolutely true (verified) or free of flaws (validated) but that by a variety of comparisons of model predictions with available data, confidence in the ability of the model to make useful predictions can be increased.

59 Chapter 2. Review and Background 39 Figure 2-7. Field-measured and PhATE TM -predicted concentrations of surrogate compounds are shown for each segment that had both a fieldmeasured and PhATE TM -predicted concentration. Sampling stations (or river mile for LAS) are identified on the x-axis, and concentration (in ng/l) is shown on the y-axis. Model PECs are represented as a square ( ). Field-measured concentrations above the MDL are presented as diamonds ( ). Estimated concentrations below the MDL are represented by a plus sign (+). If the surrogate compound was not detected, the MDL is represented by an X (x). (Source: Anderson et al., 2004)

60 40 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Figure 2-8. The cumulative probability distribution of all caffeine concentrations in U.S. surface waters reported by the USGS and PECs generated by PhATE TM for all model segments. (Source: Anderson et al., 2004) Cunningham et al. (2004) implemented PhATE TM for an environmental risk assessment of paroxetine. The PhATE TM model was used to estimate surface water concentrations across all 11 watersheds in the US. The results are shown in Figure 2 9 along with the calculated PNEC. Considering the range of segments between 1 and 99% (to eliminate the extremes most subject to bias), the PECs range from ng/l to 100 ng/l, giving PEC/PNEC ratios of to Since these ratios are significantly smaller than 1, paroxetine would not be expected to have an adverse impact on environmental organisms.

61 Chapter 2. Review and Background 41 Figure 2-9. Predicted surface water concentrations of Paroxetine using the PhATE TM model. The higher curve represents low flow conditions and the lower curve represents mean flow conditions. (Source: Cunningham et al., 2004) In the Schwab et al. study (2005), human health risk assessments for 26 active pharmaceutical ingredients (APIs) and/or their metabolites were carried out using PhATE TM. The PhATE TM model predictions are made under conservative assumptions of low river flow and no depletion (i.e., no metabolism, no removal during wastewater or drinking water treatment, and no instream depletion). Acceptable daily intakes (ADIs) were derived using the considerable data that were available for APIs. The resulting ADIs were designed to protect potentially exposed populations, including sensitive subpopulations. The ADIs were then used to estimate PNECs for two sources of potential human exposure: drinking water and fish ingestion. The PNECs are compared to measured environmental concentrations (MECs) from the published literature and to maximum PECs generated using the PhATE TM. Overall, these results demonstrate that PhATE TM may be a useful tool in predicting screening level concentrations of APIs and related compounds in the environment.

62 42 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin b. GREAT ER model (Geo referenced Regional Exposure Assessment Tool for European Rivers) The GREAT ER 5 model was developed by the European Centre for Ecotoxicology and Toxicology of Chemicals (ECETOC). 6 It is an aquatic chemical exposure prediction tool for use within environmental risk assessment schemes and river basin management. The GREAT ER software calculates the distribution of PECs of consumer chemicals in surfacewater (Schowanek and Webb, 2002). This model applies a stochastic technique (i.e. Monte Carlo simulation), which allows most input parameters to be described in terms of a distribution (normal, lognormal, or uniform distributions can be specified). The Monte Carlo approach generally requires about 1,000 runs for sufficient convergence to be obtained. Thus GREAT ER can produce a statistical distribution of PECs, as required for probabilistic risk assessment (ECETOC, 1994). ARCVIEW 3.0a or 3.1 software is required to run GREAT ER (Schowanek et al., 2001). This model has many features (GREAT ER online, 2007) such as; A higher tier support system designed for use at a post screening level in the EU Risk Assessment process, and in the EU Water Framework Directive. PC software which predicts concentrations of chemicals in rivers across Europe. A software system that combines a GIS with fate models to produce a 5 GREAT ER 1.0 was the first release of the model. The current technical standard is version ECETOC, European Centre for Ecotoxicology and Toxicology of Chemicals, was established in 1978 as a scientific, non profit, non commercial association. It is financed by 52 of the leading companies with interests in the manufacture and use of chemicals. A stand alone organization, it was established to provide a scientific forum through which the extensive specialist expertise in the European chemical industry could be harnessed to research, review, assess and publish studies on the ecotoxicology and toxicology of chemicals. The objective of the GREAT ER project is to develop and validate a powerful and accurate aquatic chemical exposure prediction tool for use within the EU environmental risk assessment schemes. Current techniques to estimate regional PECs use a generic multimedia unit worldʹ approach and do not account for spatial and temporal variability in landscape characteristics, river flows and/or chemical emissions. Hence, the results are merely applicable on a generic screening level since these models do not offer a realistic prediction of actual steady state background concentrations. The system uses Geographical Information Systems (GIS) for data storage and visualization, combined with simple mathematical models for prediction of chemical fate. Hydrological databases and models are used to determine flow and dilution data.

63 Chapter 2. Review and Background 43 simple and clear visualization of predicted chemical concentrations and water quality along a river. A tool to study the impact of chemicals emitted by point sources into rivers: GREAT ER allows calculating GIS based equivalents of PEC local and PEC regional for the aquatic environment. Free accessibility. The model has been implemented for a variety of river basins: 4 in the UK (Aire, Calder, Went, and Rother), 1 in Italy (Lambro), 4 in Germany (Itter, Unter Main, Main and Ruhr), 1 in Belgium (Rupel) and 1 in France (Mayenne). Several other river basins are under construction. GREAT ER Output (1) Color Coded River Maps GREAT ERʹs direct output provides PECs linked to a river network, which are visualized as color coded digital river maps using a GIS. To capture natural variability, the predicted concentrations are represented as statistical distributions. The GIS tools allow identification of any locations within a region where site specific PEC values may exceed the PNEC (i.e. ʹhot spotsʹ). (Figure 2 10, GREAT ER Online, 2007).

64 44 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Figure GREAT-ER model outputs user interface. (2) Concentration Profiles PECs through the studied catchment can be generated for any selected probability level (e.g. PECmean or PEC90%). Such profiles clearly illustrate chemical emissions and fate from a riverʹs headwaters down to its mouth, and can be used to compare model predictions with monitoring data (GREAT ER Online, 2007). (3) Aggregated PECs New concepts have been developed to aggregate geo referenced model results into a spatially averaged PEC, which is representative of the river basin under study. GREAT ER can generate PECinitial which comes from the distribution of concentrations in the river stretch below each emission point. This corresponds to the PEClocal concept used in the EU Technical Guidance Document. GREAT ER can also generate PECcatchment by incorporating the

65 Chapter 2. Review and Background 45 concentration distributions in each river stretch in the catchment. This corresponds to the PECregional concept used in the EU Technical Guidance Document (GREAT ER Online, 2007). (4) Application and Validation The GREAT ER system has been successfully applied to model emissions of chemicals 7 from domestic use (treated or direct discharge) and from industrial point sources (GREAT ER Online, 2007). To evaluate the accuracy of GREAT ER 1.0, 2 year monitoring programs were conducted in the four UK catchments (Yorkshire), in Italy (Lambro) and in Germany (Itter). Over 2000 river water and over 600 wastewater treatment effluent samples were analyzed for the surfactant LAS and for Boron. A monitoring program for the Rupel catchment (Belgium) was done in Most catchments released by the GREAT ER organizations are accompanied by a field validation. 3.2 Models for Predicting PECs of Veterinary Pharmaceuticals Potential models for predicting the water quality of non point sources are: SWMM (Storm Water Management Model), DR3M QUAL (Distributed Routing Rainfall Runoff Model), HSPF (Hydrological Simulation Program Fortran), and STORM (Storage, Treatment, Overflow, Runoff Model). These four models essentially comprise a group of the best choices of full scale simulation models for urban areas. 8 Meawhile VetPec and VetCalc are models for veterinary medicines solely. A version of the USGS Distributed Routing Rainfall Runoff Model, which includes quality simulation (DR3M QUAL), is available from that agency for general use (Alley and Smith, 1982). Runoff generation and subsequent routing use the kinematic wave method, and parameter estimation assistance is included in the model. Quality is simulated using buildup and washoff functions, with settling of solids in storage units dependent on a particle size 7 Detergent ingredients, Pharmaceuticals, Textile dying agents, Biocides, Phosphate, Plasticizers, Caffeine. 8 As follows 4 models were extracted from Modeling of Nonpoint Source Water Quality in Urban and Non urban Areas (EPA/600/3 91/039, June, 1991).

66 46 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin distribution. The model has been used in some of the NURP studies that were conducted by the USGS (Alley, 1986). No microcomputer version is available. The Hydrological Simulation Program Fortran (HSPF) is the culmination of hydrologic routines that originated with the Stanford Watershed Model in 1966 and eventually incorporated many nonpoint source modeling efforts of the EPA Athens laboratory (Johanson et al., 1984). This model has been widely used for non urban nonpoint source modeling. The first significant use of continuous simulation of urban hydrology came with the Storage, Treatment, Overflow, Runoff Model (STORM), developed by the Corps of Engineers Hydrologic Engineering Center (HEC) (Roesner et al., 1974) for application to the San Francisco master plan for CSO pollution abatement. The HEC provides application guidelines (Abbott, 1977) and the current version includes dry weather flow input for combined sewer simulation. STORM utilizes a simple runoff coefficient, SCS, and unit hydrograph methods for generation of hourly runoff depths from hourly rainfall inputs. From the group of potential models for estimating PECs for veterinary pharmaceuticals, SWMM, VetPec, and VetCalc are reviewed in more detail below. a. SWMM (Storm Water Management Model) The original version of the Storm Water Management Model (SWMM) was developed for the US EPA in 1971 as a single event model specifically for the analysis of combined sewer overflows (CSOs)(Metcalf and Eddy et al., 1971), but its scope has vastly broadened since its original release. Version 4 (Huber and Dickinson, 1988; Roesner et al., 1988) of the model performs both continuous and single event simulation throughout the whole model, can simulate backwater, surcharging, pressure flow and looped connections (by solving the complete dynamic wave equations) in its extran block, and has a variety of options for quality simulation, including traditional build up and wash off formulations as well as rating curves and regression techniques. The current edition, Version 5, is a complete re write of the previous release (Figure 2 11). Subsurface flow routing (constant quality) may be performed in the Runoff Block in addition to surface quantity and quality routing, and

67 Chapter 2. Review and Background 47 treatment devices may be simulated in the Storage/Treatment Block using removal functions and sedimentation theory. A hydraulic design routine is included for sizing of pipes, and a variety of regulator devices may be simulated, including orifices (fixed and variable), weirs, pumps, and storage. A bibliography of SWMM usage is available (Huber et al., 1986) that contains many references to case studies. Originally SWMM consisted of four functional program modules (namely RUNOFF, transport, EXTRAN, and storage/treatment modules) along with a coordination module (i.e., Executive module) and several service modules (i.e., rain, temperature, combined and statistic modules). The RUNOFF module simulates continuous runoff hydrographs and chemographs for each subcatchment in the drainage basin. Runoff hydrographs are predicted based on input hyetographs and the physical characteristics of the subcatchment: drainage area, average slope, degree of impermeability, overland flow resistance factor, surface storage and overland flow distance. The EXTRAN module is a hydrodynamic streamflow routing model that routes flow hydrographs through an open channel, computing the temporal courses of discharges and heads throughout the system. To simulate flow, SWMM uses the St. Venant equations for gradually varied, turbulent, unsteady flow (Metcalf and Eddy Inc., 1971). The St. Venant equations represent the principles of conservation of momentum (Eq. 2 3) and conservation of mass (Eq. 2 4). y v v x g x g v t = S o S f (Eq. 2 3) Q A + = 0 x t (Eq. 2 4) Where, y v = Depth of the water = Velocity

68 48 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin x t g So Sf A Q = Longitudinal distance = Time = Gravitational acceleration = Channel slope, dimensionless = Friction slope, dimensionless = Area of the flow cross section and is a function of y based upon the geometry of the conduit = Discharge and is equal to A v The components of the conservation of momentum (Eq. 2 3) represent hydrostatic pressure, convective acceleration, local acceleration, and gravity and frictional forces, respectively. Representing the effects of turbulence and viscosity, the friction slope (Sf) is calculated in SWMM using Manning s equation (Metcalf and Eddy Inc., 1971). 2 Q S f = (Eq. 2 5) 1 2 4/3 A R 2 n Where, n R A Q = Manning s roughness coefficient = Hydraulic radius = Area of the flow cross section and is a function of y based upon the geometry of the conduit = Discharge and is equal to A v Eq. 2 5 is substituted into Eq. 2 3, and the resulting equation is solved for Q. Q 1 n y x v v g x 1 v g t 2 / 3 1/ 2 = AR ( + + S o ) (Eq. 2 6) Solute transport is simulated by assuming complete and instantaneous mixing within each element of the sewer system. The instantaneous mixing assumption introduces artificial dispersion; however, as the number of

69 Chapter 2. Review and Background 49 conduit elements is increased within a system, solute transport becomes represented by pure advection (James et al., 2002). The overall transport of solutes through the system is executed through a mass balance calculation that incorporates decay. Concentrations of solutes are determined by solving the finite difference form of the continuity equation (Metcalf and Eddy Inc., 1971): (Vc) = Qi Ci QoCo kcv + S t (Eq. 2 7) Where, c V Qi, Qo Ci, Co k S t = Concentration in the mixed volume = Volume = Inflow (i) and outflow (o) rate = Concentration of the influent and effluent = Decay constant = Source (or sink) = Time Assuming complete mixing and applying a finite difference scheme, Eq. 2 8 becomes: C j+ 1 2 ( Cj( Vj( ( D1 + D2 )) Q = t ( V oj j+ 1 ) + ( C ij Qi ) + ( C ij+ 1 2 ( + ( D1 + D2 )) + Q t j Q oj+ 1 ij+ 1 ) ) + D (Eq. 2 8) S( V + V 2 j j+ 1 )) Where, j D1 D2 S c = Time step number = Decay constant = Growth constant = Maximum growth = Concentration in the mixed volume

70 50 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin V Qi, Qo Ci, Co t = Volume = Inflow (i) and outflow (o) rate = Concentration of the influent and effluent = Time The transport block is designed to estimate infiltration and dry weather flow and to calculate internal storage. Transport can be used to route flow through pipes, manholes, and surface channels during dry periods or during precipitation events. To compute average width, which affects the runoff lamination and temporal hydrological response, the expression as follows is used: W = ( 2 Sk ) L (Eq. 2 9) S k = ( A A1) / A (Eq. 2 10) 2 A = A 1 + A 2 A = W L Where, W = the average width (m) A = subcatchment area (m 2 ) = area located to the left and to the right of the channel, A1, A2 respectively (m 2 ) SK = skew factor L = channel length (m)

71 Chapter 2. Review and Background 51 SWMM Application Figure Output screen of SWMM. In the Tsihrintzis and Hamid (1998) study, runoff quality was predicted from small urban catchments using SWMM. The runoff block of SWMM was used to simulate the quantity and quality of urban storm water runoff from four relatively small sites in South Florida, each with a specific predominant land use (i.e. low density residential, high density residential, highway and commercial). The objectives of the study were to test the applicability of this model in small subtropical urban catchments and provide modelers with a way to select appropriate input parameters to be used in planning studies. A total of 58 storm events, measured by the US Geological Survey (USGS), provided hyetographs, hydrographs and pollutant loadings for biological oxygen demand (BOD5), total suspended solids (TSS), total Kjeldahl nitrogen (TKN) and lead (Pb), and were used for calibration of the model. Several other catchment characteristics, also measured or estimated by USGS, were used in model input preparation. Application of the model was done using the Green Ampt equation for infiltration loss computation, a pollutant accumulation equation using a power build up equation dependent on the number of dry days, and a power wash off equation dependent on the

72 52 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin predicted runoff rate. Calibrated quantity input parameters were presented and compared with suggested values in the literature. The impervious depression storage was generally found to be the most sensitive calibration parameter, followed by the Manningʹs roughness coefficients of conduit and overland flow, the Green Ampt infiltration parameters and, finally, the pervious depression storage. Calibrated quality input parameters were presented in the form of regression equations, as a function of rainfall depth and the number of antecedent dry days. A total of 16 independent rainfall events were used for verification of the model, which showed a good comparison with observed data for both hydrographs and pollutant loadings. Average model predictions for the four constituent concentrations from the verification runs also showed good agreement with NURP published values in Florida and US sites. A study from Northern Spain (Temprano et al., 2005) presented an application of SWMM for predicting the pollution in rainy weather in a combined sewer system catchment in Santander, Spain. Suspended solids (SS), chemical oxygen demand (COD) and total Kjeldahl nitrogen (TKN) were measured at the exit of the catchment and these parameters were used for the calibration and validation of the model. The process of hydraulic and quality calibration was described and the values of the adjusted parameters were presented, comparing them with those obtained from other studies. The calibrated model simulated accurately the hydrograph s shape and the time of presentation of the peak flows. The accuracy of adjustment of the volume was 96%. As for the quality validation, the accuracy of adjustment among the total simulated loads of SS, COD and TKN, and those measured at the end of the rainfall events were 93, 95 and 78%, respectively, confirming the relative accuracy of the model in the prediction mode. The phenomenon of the first flush was analyzed, and it was determined that 65, 57 and 54% of the polluting loads of COD, SS and TKN, respectively, were swept along by the first 30% of the volume in the rainfall events used for the calibration of the model. Recently in Korea, the SWMM was applied to plan development areas. In the Jang et al. (2007) study, SWMM was used as a tool for hydrologic impact

73 Chapter 2. Review and Background 53 assessment. Hydrologic impact assessment is necessary for designing retention storage for urban drainage systems in a planned development area to minimize the effect of urbanization and to assess pollutant loads from urban areas. For such assessment, a single model can be used or two hydrologic models can be paired for pre and post development conditions. Typical pairings are the use of synthetic hydrograph models for both pre and post conditions or use of a synthetic hydrograph for pre development and an urban hydrology model for post development condition. Using two synthetic hydrographs poses the problem of accounting for the drainage structure for post development conditions while the latter method of using two different models can produce erratic evaluation from errors due to different model conceptualizations and parameterizations. In order to overcome the aforementioned shortcomings, the use of SWMM for both pre and post development conditions is proposed in this study. The SWMM was applied to four planned development areas in Korea. A comparison of the results with previous assessments done for the same sites showed that the new approach can resolve the irrationalities that can occur when combining of two different models such as smaller peak flow and longer time to peak for the post development condition. It is thought that the proposed method of using the SWMM to assess both pre and post conditions improves the value of the hydrologic impact assessment on planned development areas. b. VetPec VetPec was developed by the WRC (Water Research Center) on behalf of the veterinary medicines directorate (VMD). VetPec combines a number of models such as the FEDESA and the Mackay fugacity model to estimate PEC in soil (including pore water), groundwater, and surface water. The model presumes that the animals are treated indoors (i.e., not out in the field) and that the manure is spread in the field. The VetPec model contains default scenarios that are site specific for England, but generic assessments can also be made (Wajsman and Rudén, 2006). Input data for VetPec can be categorized according to the following: 1) physicochemical data, 2) dosing information (and more general information)

74 54 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin for a selected species of animal, 3) choice of application scenario for manure, 4) choice of groundwater scenario, and 5) choice of surface water scenario. VetPec has since been replaced by VetCalc, which is described below. c. VetCalc VetCalc is a software simulation tool that calculates the PEC of veterinary medicines in the aquatic environment. This model was developed by Cambridge Environmental Assessment (CEA), using funding from Defra in the UK. The software can calculate PECs based on the spreading of manure from animals treated with veterinary medicines while in housing and also based on direct excretion from animals treated while in pasture. VetCalc estimates PECs for groundwater and surface water for 12 predefined scenarios which were chosen on the basis of: the size and importance of their livestock production and its diversity, the range of agricultural practice covered by the scenario, and the desire to cover three different European climate zones (Mediterranean, central and continental/scandinavian). Each of the scenarios has been ranked in terms of its applicability as a possible scenario for each livestock species. The model also includes, for each scenario, the typical manure management practices for the region on which the scenario is based (Table 2 5; EMEA, 2006; VMD, 2007). The model is intended for use by those who are involved in the preparation of the ecotoxicology section of Part III of a Marketing Authorization application dossier. It is also intended for use by assessors within competent authorities responsible for the assessment of environmental safety data. The model is designed specifically for use when preparing a Phase II environmental risk assessment (EMEA, 2006). A new version of VetCalc, v1.4, was introduced in August A second update of VetCalc, v1.5, was launched in February 2007 (VMD, 2007; Figure 2 12).

75 Chapter 2. Review and Background 55 Figure VetCalc. VetCalc can be split into four major modeling tasks as follows: 1) provision of input on dosage regime and chemical characteristics, 2) calculation of maximum/initial PEC in excreta and soil, 3) simulation of subsequent fate in soil (including potential for run off, leaching and degradation and estimation of PEC values in shallow groundwater), 4) simulation of subsequent fate in surface water (including potential for dilution/advection, degradation and partitioning and estimation of PEC values in the water column and sediment). The VetCalc model provides flexibility in the simulation of certain processes that may provide a degree of mitigation in more realistic exposure assessment, where warranted. As such, the model provides the user with the opportunity to define both a Basic set of environmental fate and physico chemical parameters for the compound under assessment (minimum dataset required at VICH Phase II) or an Advanced option including the opportunity to include: 1) more accurate simulation of behavior in water sediment systems, 2) metabolism, 3) degradation during storage, 4) ph dependant sorption, 5) field dissipation versus lab degradation rates (influence of soil moisture and temperature conditions), and 6) behavior of metabolites.

76 56 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Table 2-5. Summary of 12 scenarios in VetCalc Region Soil type Climate type Spain Andalucia Italy Emilia Romagna Netherlands Noord Brabant Sandy silt loam Potential to cover other regions Similarity to FOCUS scenarios Mediterranean Portugal None Clay loam Mediterranean None FOCUS SW R3 Sand Central Denmark Loamy sand Central France Bretagne UK Yorkshire France Mid Pyrennes Ireland Germany Brandenburg Belgium (Vlaanderen); Germany (Nord Rhein Westfalia) Germany (Schleswig Holstein) FOCUS SW D3 FOCUS SW D3 Sandy loam Central None None Sandy loam Central None None Sandy clay loam Sandy clay loam Sandy silt loam Central Central Central Spain (Cataluna and Aragon) UK (Northern Ireland) Poland; Czech Republic FOCUS SW R4 None None UK Wales Clay loam Central None None UK Cornwall & Devon Finland Etalae Suomi Source: EMEA, Clay Central None None Sandy loam Continental / Scandinavia Sweden FOCUS GW

77 Chapter 2. Review and Background Selecting Models for This Study There are a wide variety of models available for estimating the environmental concentrations of chemicals. However, there are few models applicable for pharmaceuticals. This report briefly reviewed a selection of models that are potentially applicable for modeling pharmaceuticals: 1) PhATE TM and GREAT ER for human pharmaceuticals, and 2) SWMM, VetPec, and VetCalc for veterinary pharmaceuticals. The characteristics of each model are presented in Table 2 6. PhATE TM (US) and GREAT ER (Europe), river catchment models, have been developed recently and used for predicting the environmental concentrations of some pharmaceuticals. GREAT ER is a stochastic model to allow for calculation of a realistic distribution of down the drain chemicals and their environmental concentrations for use in the EU risk assessment scheme. Sewage treatment removal is modeled and the load of the chemical is estimated from market consumption data. It is reported that GREAT ER appears to be an appropriate tool at a catchment scale, where the availability of chemical data is limited (Keller, 2006). In conjunction with GIS (geographic information system), both models also provide a spatial visualization of the results, along with an exceptionally accurate representation of the river network. Therefore, both PhATE TM and GREAT ER appear to be appropriate for modeling human pharmaceutical concentrations in the Han River, Korea. However, considering model accessibility and input data requirements, PhATE TM was selected for this study. VetPec (currently replaced by VetCalc) and VetCalc are suited for veterinary pharmaceutical assessment and both are widely accepted in the European countries. The calculations in VetCalc aid in the computation of the environmental risk assessments required in accordance with European Uniondirective 81/852/EEC, amended by directive 92/18/EC, that form the basis of the EMEA guideline for risk assessment of veterinary products. VetPec and VetCalc, however, contain default scenarios that are site specific for England and selected European countries, respectively. Although the authors claim that generic assessments can be made using these models, the models were not readily available for this study.

78 58 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Unlike either VetPec or VetCalc, SWMM is not designed exclusively for environmental veterinary pharmaceutical assessment. SWMM is a dynamic rainfall runoff simulation model used for single event or long term (continuous) simulation of runoff quantity and quality from, primarily, urban areas. The runoff component of SWMM operates on a collection of subcatchment areas that receive precipitation and generate runoff and nonpoint source pollutant loadings (which will be veterinary pharmaceuticals in this study). Drugs used in animal husbandry, along with the drug metabolites, are excreted in manure and farmers subsequently use this manure to fertilize fields. Thus the drug residues in the manure may most commonly reach surface water as runoff from the soil after heavy rain. Since the main route of veterinary pharmaceutical introduction into the environment is through soil run off and SWMM is primarily a rainfall runoff simulation model, SWMM was selected as the predicting model for veterinary pharmaceuticals in this study.

79 Chapter 2. Review and Background 59 Table 2-6. Information on the reviewed models Suitable for Developer Scope Characteristics Availability PhATE TM GREAT ER SWMM VetPec VetCalc Human Medical Products (HMPs) PhRMA (Pharmaceutical Research and Manufactures of America) Estimating PEC in surfacewater Default scenarios that are site specific for US Not available to the public. HMPs, Chemical ECETOC (European Centre for Ecotoxicology and Toxicology of Chemicals) Estimating PECs in surface waters due to point source Area specific assessments and requires a GIS system ECETOC Nonpoint source runoff quality simulation US EPA (Environmental Protection Agency) Single event or long term (continuous) simulation of runoff quantity and quality Dynamic rainfallrunoff simulation model US EPA ( gov). Veterinary Medical Products (VMPs) WRC (Water Research Center) in UK Estimating PEC in soil, groundwater, and surface water Site specific for England VMPs Cambridge Environmental Assessments (CEA) using funding from Defra in UK Calculating PEC in the aquatic and soil environment Site specific for European Countries VMD (

80 60 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM Jeongim Park, Myung Hyun Kim, Kyungho Choi, Young Hee Kim The objective of this chapter is to investigate the potential use of the PhATE TM model for assessing pharmaceutical levels in the Han River basin. The PhATE TM model was developed by the Pharmaceutical Research and Manufacturers of America (PhRMA) as a tool to estimate predicted environmental concentrations (PECs) of pharmaceuticals resulting from human use of medicine. PhATE TM has modeled PECs in 11 watersheds selected to be representative of most hydrologic regions of the United States. More details concerning the PhATE TM model are described in Chapter 2. In this study, sufficient relevant information on the Han River was input into PhATE TM, resulting in the inclusion of the Han River as the 12 th watershed available for simulation runs by the application (Figure 3 1). Figure 3-1. Main screen of PhATE TM including the Han River.

81 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM 61 Two categories of information are required to run PhATE TM : watersheds to run and chemicals to evaluate (Table 3 1). A database of 122 human pharmaceuticals of high production in Korea (Park, 2006) was populated along with their chemical properties in the chemicals to evaluate category. Once the necessary data was entered, PhATE TM was prepared to run simulations for the Han River. The PECs generated by the PhATE TM model were then compared to the measured concentrations of selected pharmaceuticals, in order to demonstrate the utility of PhATE TM as a predictive tool. Table 3-1. Key inputs for PhATE TM model to run for the Han River Data Key inputs Data sources Watershed Data Chemical Data STPs Dams and reservoirs Segment Detail Name, Location, POTW treatment type, Population served, Flow rate Name, Volume, Surface area, Length, Depth Mean flow, Low flow, Mean flow velocity, Low flow velocity, Length, Depth, Width Chemical properties (Kow, Kd, BCF, Photo degradation rate, Hydrolysis rate, pka, water solubility etc.) MOE 1) MOCT 2) Hydrological Annual Report in Korea, KRC 3), K water 4) Literature review, EPI Suite ver. 2.0 Human use (Usage per capita) Park, 2006 Human Loss (metabolism) Literature review, Park, 2006 POTW removal efficiency Literature review, EPI Suite ver ) MOE: Ministry of Environment, 2) MOCT: Ministry of Construction & Transportation, 3) KRC: Korea Rural community & Agriculture corporation, 4) K water: Korea Water Resources Corporation.

82 62 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin 1. Constructing PhATE TM for the Han River There are several critical factors which must be taken into consideration when evaluating the Han River with the PhATE TM simulation model. The Han River runs through the heart of the Seoul Metropolitan Area where approximately 22.5 million people, or 46% of the Korean population, live. The population density of this area is almost four times greater than the national average. Many concentrated animal feeding operations are located in the upriver district, and these sites may also serve as additional source points for unwanted pharmaceutical residues to enter the nearby environment, including the river s tributaries. The Han River s flow condition could be an important factor influencing the variations in pharmaceutical occurrence. Unlike major rivers in other countries, Korean rivers, including the Han River, are characterized by extremely high coefficients of river regime (~300:1), which is the ratio of a riverʹs maximum and minimum flow volume. The high regime characteristic of the Han River is due to the occurrence of hot and dry periods followed by frequent rainy spells and typhoons, which arise in a 2 3 week period during the mid summer season. Approximately 60% of the total annual precipitation occurs during this season. Therefore, occurrences and levels of pharmaceuticals in the river water are expected to vary considerably depending on the river s flow conditions. The Han River is the confluence of the Namhan River (South Han River) and the Bukhan River (North Han River). The River flows through Seoul and then merges with the Imjin River shortly before it flows into the West Sea of the Korean peninsula. The total length of the Han River is 514 km. Although it is not a long river, it is the largest river in Korea considering the basin area and river discharge. The basin area of the Han River is 26,018 km 2, or 26% of the nation s total area, and the annual runoff volume is 27.7 billion m 3, 28% of the nation s total runoff. The Han River consists of a total of 705 tributaries, including 15 national rivers, 12 provincial rivers and 678 local rivers. 1.1 Watershed Data for the Han River For this study, one portion of the Han River was selected. The watershed

83 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM 63 extending from the Paldang Dam, where the Namhan River and Bukhan River meet, to the Imjin River was defined as the study area (Figure 3 2). The length is 78.3 km and the area of the watershed is 2,364 km 2. The watershed represents about 10% of the total basin area of the Han River. Nearly 16 million people live in the selected area, which contains 7 sewage treatment plants (STPs). Han River Nanji STP Jungnang Stream Wangsuk Stream Guri STP Jungnang STP Seonam STP Anyang Stream Sucksoo/ Bakdal STP Tancheon STP Tancheon Stream Paldang Dam Figure 3-2. PhATE TM study area and the locations of STPs. There are two tabs on the Watershed Data screen in the PhATE TM application, for Drinking water systems and POTW systems. The information entered for the 7 STPs and 11 segments was entered under POTW systems, and Drinking water systems were not considered in this study. Information was added to the model database to describe the STPs in the watershed area as follows: STP name: Name of the selected STP STP type: Treatment performed at the selected STP. Primary,

84 64 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin secondary, or advanced treatment types are shown in this study 9 Population served (number): The total population contributing waste to the selected STP Flow rate: the average daily effluent flow Next segment distance: The length from the selected STP to the end of the segment on which it is located in the watershed Table 3 2 presents the information on each STP included in this study and Figure 3 3 shows an input screen capture of POTW (or STP) information. Table 3-2. Summary of key information on seven STPs included in this study STP Name Capacity (10,000 m 3 /d) 1) Treatment Volume (10,000 m 3 /d) 1) Population served ( 1,000) 2) Treatment Process 1) Guri CNR (Cilium Nutrient Removal) Tancheon ,720 Activated sludge Jungnang ,798 Activated sludge Anyang 3) ,010 Activated sludge Nanji ,795 Activated sludge Seonam ,363 Activated sludge 1) Ministry of Environment Statistics of Sewerage. 2) Korea National Statistical Office, Resident registration record based population in ) Anyang STP includes both Sucksoo STP and Bakdal STP. 9 The seven POTW treatment types shown in the PhATE TM model are: No discharge, Raw discharge, Primary: BOD (Biological oxygen Demand)> 45 mg/l, Advanced primary: 30<BOD 45 mg/l, Secondary: 25 BOD 30 mg/l, Advanced treatment I: 10 BOD<25 mg/l and/or nutrient removal, and Advanced treatment II: BOD<10 mg/l and/or nutrient removal.

85 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM 65 Figure 3-3. STPs data screen in PhATE TM for the Han River. Figure 3 4 is the segmentation diagram of the Han River included in this study. The study portion of the Han River was divided into eleven segments, which were partitioned based on either STP location or tributary inlet points. Hydrological inputs (i.e., mean flows, travel times, etc.) for the segments were obtained from the reports by KRC and K water.

86 66 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin 11 Seonam STP Nanji STP Guri STP 2 Jungnang STP Tancheon STP Paldang Dam Sucksoo/Bakdal STP Figure 3-4. Segmentation scheme of the Han River for PhATE TM. 1.2 Target Compounds and Collection of Compound Specific Data The Chemical Data section in PhATE TM contains five tabs: 1) Chemical properties, 2) Human Use/Loss, 3) In Stream, 4) Treatment, and 5) Toxicology. Figure 3 5 shows an example of a physicochemical properties screen within the Chemical Data section. For this study, a database with 122 human pharmaceutical compounds was input into PhATE TM. The selected pharmaceuticals were identified to have a production volume greater than 7,000 kg per annum in Korea. Pharmaceutical compounds produced in a quantity of greater than 7,000 kg/year are predicted to potentially have an environmental introduction concentration of 1 µg/l (Park, 2006). Since the US FDA specified the environmental concentration action level for human pharmaceuticals as 1 µg/l, pharmaceuticals produced in excess of 7,000 kg per annum are those most likely to be of environmental concern. Physicochemical properties of the 122 pharmaceutical compounds were obtained, if available, from published information on the compounds or estimated using EPI Suite ver The comprehensive list of the 122 compounds along with physico chemical data can be found in Appendix 1.

87 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM 67 Figure 3-5. Chemical properties screen in Chemical data for PhATE TM. Figure 3 6 shows the Human use/loss screen and Treatment screen in the Chemical Data section in PhATE TM. Human use information was obtained from the production volume data presented in a previous study (Park, 2006), while 48,000,000 was used for the population of Korea (KNSO, 2006). Utilizing the production volume data and population, per Capita Usage (kg/capita year) was calculated. Fractional loss prior to POTW in the Human Loss field means the fraction or proportion of chemical lost before it reaches the POTW. The compound specific excretion rate (%) was supplied, when available. Treatment loss value, or STP removal rate (%) of the compound, on the Treatment screen was obtained from published information or estimated using EPI Suite ver

88 68 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Figure 3-6. Human Use/Loss screen and Treatment screen in Chemical data for PhATE TM. 2. Modeling Human Pharmaceuticals in the Han River Watershed with PhATE TM Predicting the PECs of human pharmaceuticals using PhATE TM simulation was performed for four compounds, including, acetaminophen, cimetidine, roxithromycin, and chloramphenicol. These pharmaceuticals were selected for the simulation study based on the availability of monitoring data from published information. Table 3 3 briefly describes the physicochemical properties and includes the PECsinitial (µg/l) of the four compounds. Pharmaceuticals that are also used in veterinary drug products, that occur naturally, or that have uses other than merely human pharmaceutical use were not selected for modeling purposes, because these compounds have pathways of entry into the environment that are not accounted for in the PhATE TM model.

89 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM 69 Table 3-3. Physicochemical properties of selected pharmaceuticals Chemicals Acetaminophen Cimetidine Roxithromycin Chloramphenicol Use (kg/year, 2003) 710, ,985 44,778 7,512 Therapeutic category Anti Inflammatory Anti ulcerant Antibiotic Antibiotic M.W CAS No Molecular structure Log Kow ) 0.4 2) ) ) Log BCF T1/2 (Hr) ,320 1,440 Excretion rate of parent compound (%) ) >65 30 STP removal (%) ) 4 9 PEC initial (µg /L), Korea ) Sangster, 1994., 2) Hansch et al., 1995., 3) oral administration, 4) calculated with EPI Suite ver

90 70 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin For this simulation study it was assumed that the four compounds of interest were discharged to surface waters only via STP effluents, thus diffuse sources, such as leaks or agricultural contamination were not accounted for in the model. The PhATE TM model predictions are made using the conservative assumptions of no depletion (i.e., no metabolism, no removal during wastewater or drinking water treatment, and no instream depletion). The modeling condition for flowrate was set at the mean flow, due to the current data availability. Therefore, for the simulation performed in this study, two main assumptions were made; 1) there was no in stream depletion, and 2) the flow condition was set to the mean flow rate. From the main screen of PhATE TM the user can select a result format, i.e., Graph Results or Map Results. Figure 3 7 to Figure 3 10 present the modeling results in graph and map format for the four study compounds, acetaminophen, cimetidine, roxithromycin and chloramphenicol, respectively.

91 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM 71 (a) 1.0e ug/l (b) Figure 3-7. PhATE TM modeled-segmental concentrations of Acetaminophen in the Han River, (a) graph results with comparison to the PECinitial, (b) map results (some segments were intentionally magnified).

92 72 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin (a) 2.33 ug/l (b) Figure 3-8. PhATE TM modeled-segmental concentrations of Cimetidine in the Han River, (a) graph results with the PECinitial, (b) map results (some segments were intentionally magnified).

93 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM 73 (a) 0.64 ug/l (b) Figure 3-9. PhATE TM modeled-segmental concentrations of Roxithromycin in the Han River, (a) graph results with the PECinitial, (b) map results (some segments were intentionally magnified).

94 74 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin (a) 0.11 ug/l (b) Figure PhATE TM modeled-segmental concentrations of Chloramphenicol in the Han River, (a) graph results with the PECinitial, (b) map results (some segments were intentionally magnified).

95 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM Point by Point Comparisons of Model Data to Field Data Demonstrating the utility of a model as a predictive tool is a process that is typically called verification or validation. However, use of the terms verification or validation is not applicable for environmental models (Oreskes et al., 1994) since an environmental model can never be proven to be absolutely true (verified) or free of flaws (validated). However, confidence in the model output can be improved by comparing model prediction data with available data. Therefore, in this study, a point by point comparison will be used to evaluate the prediction model. A study is defined as screening when limited calibration and validation data are available and the uncertainty associated with the predicted results is comparatively large, somewhere near an order of magnitude (US EPA, 1987). Therefore, in this study corroboration of the PhATE TM model is directed at determining the performance of the model as a screening tool. PhATE TM model corroboration in this study consisted of comparing fieldmeasured data with PhATE TM simulated PECs for the same locations on a point by point basis. To conduct the point by point comparisons, fieldmeasured concentrations reported by Kim et al. (2005) and Park (2006) were used. Point by point comparison data is presented in Figure 3 12.

96 76 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin 3.1 Field Measurement Data Hongjae Stream Joongnang Stream Wangsuk Stream Ahnyang Stream Sadang Stream Tancheon Stream Figure Sampling sites of human pharmaceuticals in this study.

97 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM 77 Table 3-4. Field measurement data generated during this study Chemicals Joongnang Stream* Ahnyang Stream* Wangsuk Stream* Tancheon Stream* Sadang Stream* Unit: ng/l Hongjae Stream* Acetaminophen <LOQ <LOQ <LOD <LOD Trimethoprim <LOQ <LOQ <LOQ <LOQ Enrofloxacin 20.0 <LOQ <LOQ 30.0 <LOQ <LOQ <LOD <LOQ <LOD <LOD <LOQ <LOQ Florfenicol <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Sulfamethoxazole <LOD <LOD <LOD <LOD Diltiazem <LOD <LOD <LOD <LOD <LOD <LOD * Duplicate samples taken at the same location.

98 78 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin 3.2 Point by Point Comparisons Point by point comparisons of modeling data to the field measurements are presented in Figure The PhATE TM model predicted concentrations for acetaminophen and cimetidine are in good agreement with actual measured concentrations. For Roxithromycin, PhATE TM predicted higher concentrations than reported by Park (2006). This difference cannot be explained by an overestimate of annual use but it could be explained in two other ways. First, the difference suggests that the STP removal rate for roxithromycin may be greater than 4%, which was the rate employed in the PhATE TM model in this study (Table 3 3). Another possible explanation could be that the excretion rate of 65% applied in the model for roxithromycin was too conservative. If a 30% excretion rate is applied, the PhATE TM predicted PEC would be decreased by approximately 50% (=30/65) and thus would be more in line with the actual measured values for roxithromycin concentration.

99 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM 79 Table 3-5. Field-measured and PhATE TM -predicted concentrations of four human pharmaceuticals for each segment Unit: ng/l Segment Acetaminophen Cimetidine Roxithromycin Chloramphenicol PEC_PhATE MEC 1) PEC_PhATE MEC 1) PEC_PhATE MEC 2) PEC_PhATE MEC 2) <LOD 0.05 <LOD 0.18 <LOD, <LOD, ), 210 3) 1, , , <LOD 18 <LOD 6.55 <LOD, <LOD, ), 180 3) 4,270 1, , 6, , 459, 1, , <LOD ), 150 3) 5,900 5,381 2, , 26, <LOD, 38, , <LOD, <LOD 3) 9,060 3, , 31, <LOD, 233, , <LOD , 31, <LOD, 233, , <LOD , 31, <LOD, 233, , <LOD * sampling sites of the segments without STPs: 3 Jamsil, 5 Hannam, 7 Mapo, 9, 10, 11 Haengjoo 1) Kim et al., 2005., 2) Park, ) Park, 2007.

100 80 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin

101 Chapter 3. Exposure Simulation for Human Pharmaceuticals in the Han River with PhATE TM 81 Figure Field measured (represented as a solid dot) and PhATE TM - predicted concentrations (represented as a line) for each segment. Field measured concentrations were three measurements at each segment from Kim et al. (2005), two measurements from Park (2006).

102 82 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream Young Hee Kim, Kyungho Choi, Min Young Kim 1. Introduction Pharmaceuticals in surfacewater have become of greater concern among the general population as well as environmental health experts since the work of Kolpin et al. (2002). They reported the presence of dozens of pharmaceutical residues in 139 streams in the US, and raised a concern for the potential consequences of such pollution. The levels detected in sewage treatment effluents were one to two orders of magnitude higher than in surfacewater (Ternes, 1998; Heberer, 2002; Kim et al., 2006). Pharmaceuticals have been detected in soil as well, especially in agricultural areas or on livestock farmland. Often the levels found in soil are greater than those in water (Halling Sørensen et al., 1998). Potential pathways of pharmaceuticals into the environment include: discharge from factories, effluent from municipal wastewater treatment plants, direct inflow from aquatic fish farms and treatment of agricultural land with manure. Among these pathways, sewage treatment plant (STP effluents are one of the most prominent sources of human pharmaceutical contamination in the environment. Estimation models for environmental levels of human pharmaceuticals from STPs are not complex and have been relatively well established as acceptable prediction tools by several key agencies like the European Agency for the Evaluation of Medicinal Products (EMEA) of the European Union (EU) and the US Food and Drug Administration (FDA). More refined simulation models for estimating human pharmaceuticals in the environment have been developed by the Pharmaceutical Research and Manufacturers of America (PhRMA) or the

103 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 83 European Centre for Ecotoxicology and Toxicology of Chemical (ECETOC). These models estimate environmental levels of pharmaceuticals with reasonable accuracy, often within an order of magnitude error range (Schowanek and Webb, 2002). Veterinary medicines are different from human pharmaceuticals in that this group of compounds is mainly discharged into the rivers through runoffs from the agricultural area treated with manure or products that might be contaminated with pharmaceuticals. After being administered to cure diseases or to improve health, veterinary pharmaceuticals are excreted in either unmetabolized or metabolized form. Therefore, animal manure generally contains pharmaceutical residues. Hence it is probable that during rainfall events, the runoff from agricultural or livestock areas will deposit pharmaceutical residues into neighboring water supplies. Based on this pathway, veterinary pharmaceuticals are considered as non point source contaminants. An attempt to estimate the environmental concentration of veterinary medicine through the use of exposure modeling software (VetCalc ver. 1.2, available on has been performed, however, the global applicability of this program is limited, since the model is only relevant for twelve European exposure scenarios. In this study we applied the Storm Water Management Model (SWMM), a non point source model developed by the US EPA, to simulate environmental loadings of veterinary pharmaceuticals through runoff from livestock farming areas. SWMM is reported to be a model that is more suitable for municipal area simulations, however recent studies indicated that SWMM was appropriate for rural and sub urban areas as well (Jang et al., 2007; Kim, 2003; Yoon et al., 2001). In the present study, we attempted, for the first time, to apply SWMM to simulate the levels of pharmaceutical residues in water during rainfall events. Along with models that predict the environmental levels of human pharmaceuticals, e.g., PhATE TM by PhRMA, the present model can be utilized to provide improved estimations of the environmental levels of veterinary pharmaceuticals for risk assessment purposes.

104 84 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin 2. Materials and Methods 2.1 Study Area Kyungahn stream is located in Gyeonggi province, southeast of Seoul, Korea, and it flows into the Han River. This stream is an important drinking water source for populations in the Seoul metropolitan area (Figure 4 1). The catchment under consideration in this study includes the two cities of Yongin and Gwangju and has a drainage area of 446 km 2. The area is covered by 60% forest, 16.7% agricultural land and 2.6% livestock farms, that quarter a total number of 15,317 cows, 281,928 pigs and 2,974,868 chickens (Table 4 1). In addition, two livestock waste treatment plants (WWTPs) are located upstream and middle stream. However since effluents from these two WWTPs were considered as point sources, they were excluded as pollutant sources in the SWMM model, since this particular model is applicable only for non point pollution sources. Table 4-1. Total number of major livestock animals in the study area City Cow Pig Chicken Total Yongin 10, ,353 2,316,290 2,597,148 Gwangju 4,812 11, , ,965

105 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 85 Figure 4-1. Study area of the Kyungahn stream for SWMM. Figure 4-2. Chicken manure composting facility as non-point sources of veterinary pharmaceuticals in the study area. 2.2 Selection of Target Pharmaceuticals Widely used and highly ranked veterinary antibiotics, based on annual market production in Korea (Park 2006), were selected as the target pharmaceuticals for this study. These compounds include: enrofloxacin, florfenicol, sulfamethoxazole and trimethoprim. Florfenicol and enrofloxacin ranked fourth and fourteenth, respectively, in annual sales in 2005 in Korea.

106 86 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin In addition, florfenicol and enrofloxacin have been reportedly detected at high ppt levels in Korean rivers (Kim et al., 2006) and sulfamethoxazole and trimethoprim are reputedly the most detected pharmaceuticals found in Korean rivers. Enrofloxacin and florfenicol are veterinary antibiotics used solely for the treatment of infection and pneumonia, principally in poultry. Sulfamethoxazole is a drug used for the treatment of malaria, conjunctivitis, toxoplasmosis and urinary tract infections, and trimethoprim is most often used in combination with sulfamethoxazole. These latter two pharmaceuticals are not very highly ranked in annual usage but they were included in the study mainly because they are used in both human and animal applications. One of the goals of this study is to evaluate environmental pharmaceutical concentration predictions using a combination of animal and human modeling tools (see Chapter 5), so it is necessary to obtain data for some pharmaceuticals that have human and veterinary usage. In this chapter the environmental concentration contributed only by veterinary usage of sulfamethoxazole and trimethoprim were simulated by SWMM. The general characteristics of the four selected antibiotics are summarized in Table 4 2. Table 4-2 General characteristics of selected antibiotics Chemicals Classification Domestic usage** (kg/year) Enrofloxacin Quinolones 313,031 Florfenicol Chloramphenicols 883,945 Trimethoprim Sulfonamides 14,791 Sulfamethoxazole Aminopyrimidines 7,572 Treatment Bacterial infection Bacterial infection, pneumonia (in pigs) Disease control Disease control * Estimated by Estimation Program Interface (EPI) Suite TM ver (US EPA). ** Only for veterinary use. (Kim et al., 2006). *** Schering Plough Animal Health Target Poultry, pig, cow Fish farm, cow, pig Poultry, pig, cow Poultry, pig, cow Log Kow* ***

107 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream Application of SWMM SWMM has been generally accepted as a representative model for runoff evaluation studies in urban areas. However, more recent studies have provided evidence indicating the applicability of SWMM in non urban areas alike (Kim, 2003; Yoon et al., 2001). The process of urbanization of rural areas includes a transitional stage that has characteristics of both urban and rural environments, so it can thus be assumed that SWMM may be an appropriate modeling tool for these transition areas as well. The study area examined in this report consists of both rural areas (upper Kyungahn stream) and suburban areas (downstream). In previous studies of this area, models suited solely to either urban or rural non point conditions were employed to simulate the water quality of the Kyungahn stream: Shin et al. (2004) used the STORM model, which is an urban non point source model; Kim et al. (2002) used the AGNPS model, which is a rural non point source model (Shin et al., 2004; Kim et al., 2002). However, in the current study, use of the SWMM application will be employed because of its capability for assessing both urban and sub urban areas alike. SWMM consists of four computational blocks and one executive block. The executive block serves to link results from the computational blocks and also controls copy processes. The computational block consists of four separate blocks: runoff, transport, extran and storage/treatment. Their individual roles within the model are to simulate: washoff from land surface, transport loss in channel and treatment loss. The runoff block simulates the quantity of runoff into a drainage basin and the routing of the runoff flows into the major conduits. The transport block is a hydro and polluto graph model for channel system and receives hydrograph input from the runoff block. The Extran and storage/treatment blocks simulate the hydraulic model and control treatment of flow and pollutant in the drainage system (Figure 4 3). This study used two computational blocks, the runoff and transport blocks, to simulate runoff flowrate and pollutant concentration. The information data set for SWMM is listed in Table 4 3.

108 88 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Figure 4-3. Schematic configuration of SWMM blocks (Huber and Dickinson, 1988).

109 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 89 Table 4-3. Information requirements for SWMM Data area Name Model variables Area area Subcatchment Average slope % slope Impervious area % imperv Catchment Watershed width width Conduit inlet depth inlet offset Stream Conduit outlet depth outlet offset Basin width bottom width Manning coeff. for impervious area N imperv Simulation Flowrate Water quality Manning Manning coeff. for pervious coefficient area N perv Manning coeff. for stream Flowrate channel roughness Detention Detention in impervious area Dstore imperv Detention in pervious area Dstore perv Max. infil. Rate max. infil. Rate Infiltra Horton Min. infil. Rate min. infil. Rate tion eq. Decay constant decay constant Climatory Rainfall Rain gage Max. buildup max. buildup Power linear Rate constant rate constant Con Buildup eq. power/sat. tami nant Power/sat. constant constant Washoff Coefficient coefficient Exponent exponent Data Collection In order to run area simulations, SWMM first requires the input of specific information into three major data sets, as presented in Table 4 4. The climatory category needs the input of rainfall intensity information for the runoff simulation. The pollutant discharge unit, entered in the pollutant

110 90 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin category, is the key variables for simulation of the buildup and washoff of chemicals and the geomorphological characteristics of the study area are included in the watershed portion and used for runoff flow and pollutant level simulation. Table 4-4. Data sources for SWMM Category List Source SWMM input data Climatory Watershed Pollutant Rainfall (intensity, previous dry day) Watershed (subcatchment area, slope, soil type, perviousness) Stream (stream length, flowrate, riverbasin width) Land use Livestock facilities The amount of pharmaceutical used Korea meteorological administration Korea national geographic information institute Korea ministry of environment Korea national statistical office Korean environmental institute Rain gauge Land use. Subcatchment area, slope, width, % impervious area Discharge unit (rate constant) a. Watershed characteristics data The geological information required by SWMM, describing the Kyungahn watershed area being studied, was created using the Geographic Information System (GIS). The eleven subcatchments identified in the study were derived from standardized watershed catchment information suggested by the Korea ministry of environment (Figure 4 4). Land use information was obtained from the examination of cadastral maps of the cities of Yongin and Gwnagju. The other variables required, such as Manning s coefficient, were determined based on previous research on the specific soil type. Detailed subcatchment information, as entered into SWMM, is summarized in Table 4 5.

111 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 91 Figure 4-4. Subcatchments of the study area. Solid circles indicate the measuring points of pharmaceuticals during rainfalls. Solid squares indicate meteorological observatories; 1 is Jiwol and 2 is Silchon.

112 92 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Table 4-5. Characteristics of subcatchments in the study area Subcatchment Residential (%) Agricultural (%) Forest (%) Livestock farm (%) Other (%) Area (ha) %IMP.* Slope (%) Sub Sub Sub Sub Sub Sub Sub Sub Sub Sub Sub * Values obtained by multiplying residential area percentage times impervious percent (assumed residential medium density with an impervious factor of 0.6), (Arnold et al., 2002a).

113 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 93 b. Meteorological data From among the weather stations located in the Kyungahn watershed, rainfall intensity data from the Jiwol and Silchon observatories was entered for the Kyungahn stream and Gonjiam stream, respectively (Figure 4 4). Among the four rainfall events examined in this study, data from May 24 th was used only for verification of the flowrate simulation model (Table 4 6) and data from the other events, May 12 th, May 16 th and June 21 st, was used for pollutant concentration simulation exercises. Table 4-6. Rainfall events included in the study Rainfall Jiwol Silchon Date 2007, May , May. 12 Total Rainfall 14.5 mm 13 mm 1 st event** Duration 9 hrs 6 hrs Dry day* 3 d 3 d Date 2007, May , May. 16 Total Rainfall 58 mm 49 mm 2 nd event** Duration 9 hrs 11 hrs Dry day* 3 d 3 d Date 2007, May , May. 24 Total Rainfall 60.5 mm 3 rd event Duration 10 hrs Dry day* 5 d Date 2007, June , June. 21 Total Rainfall 23.5 mm 21 mm 4 th event** Duration 18 hrs 13 hrs Dry day* 27 d 27 d * Dry day is defined as a day having less than 10 mm of total rainfall. ** Pharmaceutical concentration was measured during rainfall.

114 94 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin c. Hydraulic data In the study area, water level observation posts are installed at Kyungahn 1 st Bridge, Chowol Topyung station and Kwangdong Bridge. However, rating curves of Kwangdong Bridge and Chowol Topyung stations were not established. Therefore, flowrate data was obtained only from the Kyungahn 1 st Bridge station (see Appendix 2). d. Water quality data Environmental concentrations of acetaminophen, diltiazem, sulfamethoxazole, trimethoprim, enrofloxacin and florfenicol were measured during rainfall at three sampling sites on the 12 th and 16 th of May and the 21 st of June in 2007 (Figure 4 4). Among the six pharmaceuticals evaluated, acetaminophen and diltiazem showed frequent detection (Tables 4 7 to 4 9). The prevalence of acetaminophen and diltiazem, both human pharmaceuticals, may be the result of an overflow from STPs because the Kyungahn stream watershed contains sewage treatment systems. These human pharmaceuticals were also found at a higher concentration downstream, where the population is denser than upstream, where the area is more rural. However, since acetaminophen and diltiazem are human pharmaceuticals they were not included in SWMM data set Among the three rainfall events studied, enrofloxacin and florfenicol exhibited a relatively higher concentration in the first rainfall event among three rainfalls.

115 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 95 Table 4-7. Pharmaceutical concentrations during rainfall on May 12, 2007 Site time 07:00 08:00 09:00 09:30 10:00 10:30 11:00 Sindae Bridge Kyungahn 1 st bridge Chowol Topyung Acetaminophen <LOQ <LOQ <LOQ Unit: ng/l Trimethoprim <LOQ <LOQ <LOQ <LOQ <LOQ 10 <LOQ Enrofloxacin Florfenicol 17 <LOQ <LOQ Sulfamethoxazole <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ Diltiazem 10 <LOQ 20 <LOQ <LOD <LOD <LOQ time 8:30 9:30 10:00 10:30 11:00 11:30 Acetaminophen Trimethoprim 10 <LOQ <LOQ <LOQ <LOQ <LOQ Enrofloxacin <LOQ Florfenicol <LOD <LOD <LOD <LOD ND <LOD Sulfamethoxazole Diltiazem <LOQ <LOQ 10 <LOQ time 8:00 9:00 9:30 10:00 10:30 11:00 11:30 Acetaminophen Trimethoprim 10 <LOQ 12 <LOQ <LOQ <LOQ <LOQ Enrofloxacin Florfenicol <LOD <LOD <LOD <LOD <LOD <LOD <LOD Sulfamethoxazole Diltiazem

116 96 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Table 4-8. Pharmaceutical concentrations during rainfall on May 16, 2007 Unit: ng/l Site time 14:35 15:35 16:05 16:35 17:05 17:45 18:15 19:40 20:50 21:50 22:50 23:50 Acetaminophen Sindae Bridge Kyungahn 1 st bridge Chowol Topyung Trimethoprim <LOQ <LOQ <LOQ <LOQ <LOD <LOD <LOD <LOQ <LOD <LOQ <LOD <LOD Enrofloxacin <LOQ <LOD <LOQ <LOQ <LOD <LOQ <LOQ <LOQ <LOQ Florfenicol <LOD <LOQ <LOQ <LOQ <LOD <LOD <LOD <LOD <LOD <LOQ <LOD <LOD Sulfamethoxazole 10 <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ Diltiazem <LOQ <LOQ <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD time 14:00 15:00 16:00 16:30 17:00 17:30 18:00 18:00 18:30 19:00 20:00 21:00 22:00 23:00 Acetaminophen Trimethoprim <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ <LOD Enrofloxacin <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Florfenicol <LOQ <LOD <LOD <LOQ 213 <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ Sulfamethoxazole Diltiazem 10 <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ <LOD <LOQ <LOQ <LOD <LOD time 14:35 15:35 16:05 16:35 17:05 17:45 18:10 18:40 19:40 20:40 21:40 22:40 23:40 Acetaminophen <LOQ <LOQ 14 <LOQ Trimethoprim <LOQ <LOQ <LOQ <LOQ <LOQ <LOQ Enrofloxacin <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Florfenicol <LOQ <LOQ <LOQ 23 <LOQ <LOQ <LOD <LOD <LOD Sulfamethoxazole Diltiazem <LOQ 10 <LOQ <LOQ <LOQ <LOQ <LOD 10 <LOQ <LOQ 10

117 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 97 Table 4-9. Pharmaceutical concentrations during rainfall on June 21, 2007 Unit: ng/l Site time 12:15 15:15 17:15 19:15 21:15 23:15 01:15 02:15 03:15 Sindae Bridge Kyungahn 1 st bridge Chowol Topyung Acetaminophen <LOD <LOQ 13 <LOD 10 <LOQ Trimethoprim <LOQ <LOD <LOD <LOD <LOD <LOD <LOQ <LOD <LOD Enrofloxacin <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Florfenicol <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Sulfamethoxazole <LOQ <LOQ <LOQ <LOD 17 <LOQ Diltiazem <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD time 12:15 15:15 17:15 19:15 21:15 23:15 01:15 02:15 03:15 Acetaminophen <LOD <LOQ <LOQ Trimethoprim <LOQ <LOQ <LOQ <LOQ Enrofloxacin <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Florfenicol <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Sulfamethoxazole Diltiazem 10 <LOQ <LOQ <LOQ <LOQ <LOD <LOD <LOD <LOD time 12:15 15:15 17:15 19:15 21:15 23:15 01:15 02:15 03:15 Acetaminophen <LOD Trimethoprim <LOQ <LOQ 17 <LOQ <LOD <LOQ Enrofloxacin <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Florfenicol <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOD Sulfamethoxazole Diltiazem <LOD <LOD <LOD <LOD <LOD <LOD <LOD <LOQ 40

118 98 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin 3. Results and Discussion 3.1 Simulation of Runoff Flowrate a. Calibration and verification Only two of the four computational blocks RUNOFF and TRANSPORT were examined in the simulations studied in this report. Using the knowledge of the geographic information system (GIS) and hydraulic data system the representative dataset was created for SWMM. The SWMM system for this study consisted of 11 subcatchments and 16 conduits. For all of the subcatchments the area slope, hydrologic width, Manning s roughness coefficient and infiltration parameters, such as maximum filtration rate and decay constant, were determined based on literature and experimental data. In addition, for the 16 conduits, the channel slope and Manning s roughness coefficient were set to baseline values. Table Baseline values for variable sensitivity analysis in SWMM Variables Baseline value Reference Catchment surface slope (%) 20 Average surface slope Impervious area (%) 16.7 Percent of residential area Hydraulic width (m) 8263 Cross sectional width Maximum infiltration (mm) 20 Lee et al., 1996 n for impervious area 0.13 Huber, 1992 n for open channel 0.05 Chow et al., 1998 Detention flow (mm) 0.75 ASCE, 1992 To calibrate the SWMM model for the study area, a sensitivity analysis using the variables listed in Table 4 10 was performed. Sensitivity analysis is not only the most uncomplicated way to calibrate the model, but sensitivity analysis also allows the user to determine which calibration variable should be adjusted first to optimize model simulation accuracy. During the sensitivity analysis, as the value of each variable was increased or decreased by 40%, the simulated peak flow rate and total flow rate graphs were

119 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 99 compared to the corresponding graphs constructed from the baseline values of variables (Figure 4 5, 4 6). Maximum infiltration volume was found to be the most influential variables to both peak flow and total flow. When the maximum infiltration volume increased by 40% the peak flow and total flow decreased upto 20%. The forty percent increased Manning s coefficient for surface roughness of pervious area decreased about 10% of peak flow. The imperviousness of surface area affected to peak flow and total flow with a relatively lesser degree. On the contrary, hydraulic width showed a positive relationship to both peak flow and total flow. Figure 4-5. Sensitivity analysis at Kyungahn 1 st bridge (peak flow).

120 100 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin roughness %solpe max.volume width %imperv dstore_perv N-perv 10 Q T (%) Validation width (%) Figure 4-6. Sensitivity analysis at Kyungahn 1 st bridge (total flow). Therefore, after the sensitivity analysis the optimal values for each variable for the calibrated model were obtained by trial and error using actual measured data from the May 16 th rainfall event as the comparison event. The model calibrated using the May 16 th data was then verified by comparing a simulation with measured data, using the rainfall data from the May 24 th event (Figure 4 7). The fitness of the calibrated model was confirmed by calculation of an acceptable regression coefficient (0.84~0.91) from comparison of the verification model output and the measured data (Figure 4 9). The final variable values determined during the SWMM runoff flow model calibration procedure can be found in Table 4 11.

121 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream Rainfall (mm) Flow rate (m 3 /s) measured simulated precipitation :00: :00: :00: :00: :00:00 Time (hr) Figure 4-7. Calibration on SWMM parameters at Kyungahn 1 st bridge using the 2007/05/16 event (runoff flow) Rainfall (mm) Flow rate (m 3 /s) measured sim ulated precipitation :00: :00: :00:00 Time (hr) Figure 4-8. Calibration on SWMM parameters at Kyungahn 1 st bridge using the 2007/05/24 event (runoff flow).

122 102 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin 200 a) y = x R 2 = Simulated Flow rate (m 3 /s) Measured Flow rate (m 3 /s) Simulated Flow rate (m 3 /s) b) y = 1.049x R 2 = Measured Flow rate (m 3 /s) Figure 4-9. Regression curve between simulated and measured data (2007/05/24), a) curve based on the data from the left side of the peak and b) similar curve based on the data from the right side.

123 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 103 Table Values of each variable as determined during runoff model calibration Variable Values Reference values Sources N imperv (smooth asphalt) Huber, 1992 N perv ~0.32 (natural) Huber, 1992 D store imperv ~0.2 inch (impervious surfaces) ASCE, 1992 Dstore perv. 7 (mm) 0.3 inch (forest litter) ASCE, 1992 Max. infiltration rate * 200 (mm/hr) 203 Lee et al., 1996 Min. infiltration rate 10 (mm/hr) 12.7 Lee et al., 1996 Decay constant (1/hr) Lee et al., 1996 Roughness (natural stream) Chow et al., 1998 * NRCS soil type B (see Table 4 12). Table Infiltration coefficient of the Horton equation (Lee et al., 1996) Parameters Min. filtration volume (mm/hr) Max. filtration volume (mm/hr) NRCS soil type A B C D Decay constant (1/hr) A: Low runoff potential. Soils having high infiltration rates even when thoroughly wetted and consisting chiefly of deep, well to excessively drained sands or gravels. B: Soils having moderate infiltration rates when thoroughly wetted and consisting chiefly of moderately deep to deep, moderately well to well drained soils with moderately fine to moderately coarse textures, e.g., shallow loess, sandy loam. C: Soils having slow infiltration rates when thoroughly wetted and consisting chiefly of soils with a layer that impedes downward movement of water, or soils with moderately fine to fine textures, e.g., clay loams, shallow sandy loam. D: High runoff potential. Soils having very slow infiltration rates when thoroughly wetted and consisting chiefly of clay soils with a high swelling potential, soils with a permanent high water table, soils with a clay pan or clay layer at or near the surface, and shallow soils over nearly impervious material.

124 104 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin 3.2 Simulation of Runoff Quality a. Estimation of unit discharge To estimate unit discharge three assumptions were made. First, the total amount of domestic sales market is equally distributed to each livestock unit (only cow, pig, and chicken were targeted in this study). The rationale for this assumption was partially supported by the equivalence of animal species distribution in the nation and the study area (Figure 4 10). Second, all excreted manure from targeted animals was composted and spread nationwide evenly over agricultural areas. Third, composted manure is applied all year round. Based on these assumptions we calculated the unit discharge of each selected pharmaceutical. Each pharmaceutical s total domestic sales market was obtained from Park s report (2006) and the total number of animals was obtained from the National Statistical Office ( To determine the usage per animal, the total domestic market (total usage in Korea) was divided by the total number of animals. Multiplying the usage per animal and the total number of animals in the study area resulted in the total usage in the study area. The unit of discharge of each chemical was then obtained by dividing the total usage in the study area by the total agricultural area of the target watershed. The results are summarized in Table a) study area b) nation wide 15,317 (0.5%) 281,928 (8.6%) 9,720,853 (2%) 36,792,570 (7%) cow pig chicken 2,974,868 (90.9%) 506,672,399 (91%) Figure Cow, pig, and chicken distribution in the study area.

125 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 105 Table Unit discharge of target chemicals Chemical Total usage in Korea 1) (kg/year) Total number of animals in Korea 2) (thousands) Total number of animals in the study area (thousands) 3) Total usage in the study area (kg/year) Agricultural land in the study area (ha) 3) Unit discharge (mg/m 2 day) Farm, orchard Livestock farm 4) Enrofloxacin 24, Florfenicol 19, ,211 3, Sulfamethoxazole 14, Trimethoprim 7, ) Park, ) Korea National Statistical Office ( included only cow, pig and chicken. 3) Statistic report of sewage and veterinary wastewater in ) Multiply two weighted values to the unit discharge of the farm.

126 106 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin b. Calibration and verification Pollutant buildup and washoff from subcatchment areas are determined by the land uses assigned to those areas. Pollutant buildup that accumulates within a land use category is described (or normalized ) as a mass per unit of subcatchment area. The amount of buildup is a function of the number of preceding dry weather days and is computed using an exponential function. Pollutant washoff occurs during rainfall and is estimated by the following equation; W = q C c2 i B (Eq. 4 1) Where, W C1 C2 q B = wash load = washoff coefficient = washoff exponent = runoff rate per unit area (inches/hour or mm/hour) = unit discharge, pollutant buildup in mass (lbs or kg) per unit area The water quality model was calibrated for each study pharmaceutical using the same trial and error method that was used to calibrate the runoff flow rate model. In the calibration, the washoff coefficient and washoff exponent variables were adjusted and reset based on a comparison of simulated results and measured data (Figure 4 12). The storm event used for parameter calibration occurred on May 16 th Korea has clearly defined wet and dry seasons. In October, when the rainy season ends, the dry season begins and reaches its peak in March. April and May signal the beginning of the rainy season (Figure 4 11). Therefore, in Korea, May to June provides a good timeframe to calibrate runoff quality parameters, because the areas tested can be considered dry at the start of the test, since they are just emerging from a significant dry spell. However, as shown in Figure 4 13 and Figure 4 14, the values simulated using the calibrated model do not seem to explain the pattern of discharge observed. This is especially true for florfenicol, which was predicted to be

127 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 107 present in concentrations up to 60 ppt, but was not even detected during the June 21 st storm event. In the simulation prepared for the May 12 th rainfall event, the concentration of enrofloxacin was underestimated in the simulation by over six fold compared to the actual measured amount. Composted manure containing pharmaceuticals is generally applied more heavily during the dry season than during the rainy season. Possibly, the reason that some of the veterinary pharmaceuticals exhibited a higher thanpredicted concentration on the May 12 th event is because this storm event occurred relatively early in the wet season and much of the pollutant which had accumulated during the dry season was washed off in this early season rainfall event. The sulfamethoxazole and trimethoprim concentrations were slightly underestimated in the simulation of the May 12 th and June 21 st events. This can be partially explained by the possible contribution from human use to the measured concentration because these pharmaceuticals are used in both human and veterinary applications. Overall however, the SWMMpredicted environmental concentration of the veterinary pharmaceuticals evaluated in this study during rainfall events agreed to within one order of magnitude with the actual measured concentration. Figure Storm events in 2007.

128 108 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin a) Enrofloxacin simulated <LOQ Concentration (ug/l) :00: :00: :00: :00: :00:00 Time b) Forfenicol Florfenicol simulated measured < LOQ Concenration (ug/l) LOQ (0.01ug/L) :00: :00: :00: :00: :00:00 Time

129 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 109 c) Sulfamethoxazole simulated measured Concentration (ug/l) LOQ (0.01ug/L) :00: :00: :00: :00: :00:00 Time d) Trimethoprim simulated measured < LOQ Concentration (ug/l) LOQ (0.01ug/L) :00: :00: :00: :00: :00:00 Time Figure Comparison of measured and simulated values with SWMM at Kyungahn 1 st bridge for the 2007/05/16 event.

130 110 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin a) Enrofloxacin 0.14 simulated measured < LOQ 0.12 Concentration (ug/l) LOQ (0.01ug/L) :00: :00: :00: :00:00 Time b) Florfenicol simulated < LOQ Concenration (ug/l) LOQ (0.01ug/L) :00: :00: :00: :00:00 Time

131 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 111 c) Sulfamethoxazole simulated measured Concentration (ug/l) LOQ (0.01ug/L) :00: :00: :00: :00:00 Time d) Trimethoprim 0.14 simulated measured < LOQ 0.12 Concentration (ug/l) LOQ (0.01ug/L) :00: :00: :00: :00:00 Time Figure Comparison of measured and simulated values with SWMM at Kyungahn 1 st bridge for the 2007/05/12 event.

132 112 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin a) Enrofloxacin simulated <LOQ Concentration (ug/l) LOQ (0.01ug/L) :00: :00: :00: :00: :00: :00 Time b) Florfenicol simulated < LOQ Concentration (ug/l) LOQ (0.01ug/L) :00: :00: :00: :00: :00: :00 Time

133 Chapter 4. Predicting Environmental Exposure of Veterinary Pharmaceuticals with SWMM model in the Kyungahn Stream 113 c) Sulfamethoxazole 0.14 simulated measured 0.12 Concentration (ug/l) LOQ (0.01ug/L) :00: :00: :00: :00: :00: :00 Time d) Trimethoprim simulated measured < LOQ Concentration (ug/l) LOQ (0.01ug/L) :00: :00: :00: :00: :00: :00 Time Figure Comparison of measured and simulated values with SWMM at Kyungahn 1 st bridge for the 2007/06/21 event.

134 114 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin 4. Conclusions The current study illustrates the potential application of SWMM in simulating the environmental concentration of pharmaceuticals. The study also demonstrated that SWMM is well suited for the evaluation of nonurbanized watershed areas. Simulations of runoff flow rate yielded results that were in good agreement to measured data. Water quality simulations, although not as superior as the flowrate simulations, nevertheless were capable of predicting pollutant concentrations to within one order of magnitude of the actual measured concentrations. Probably one of the main sources of error in the water quality simulation model is that in this study the model is being used to predict pharmaceutical rather than general contaminant concentrations. Pharmaceuticals have different buildup and washoff patterns compared to general contaminants such as BOD and SS. In general BOD and SS have a continuous accumulation pattern because they originate from unspecified contaminant sources. On the other hand, since pharmaceuticals in animal manure are applied to agricultural areas only in the dry season, accumulation is not replenished after washoff by rainfall. Another reason for discrepancy in the water quality simulation is that the contribution of WWTP effluents is not considered in the model. In this study, two WWTPS are located in the study area but their effluents were not considered in SWMM input data. This factor probably plays a considerable role in the discrepancy noted in the simulation of pharmaceuticals used in both human and veterinary applications. However, the simulated concentration level agreed to within one order of magnitude with the measured data. Therefore, SWMM could prove beneficial for estimating veterinary pharmaceutical concentrations on a screening basis for the purpose of prioritizing pharmaceuticals that may require further risk assessment evaluation.

135 Chapter 5. Environmental Exposure Modeling of Pharmaceuticals with the PhATE TM -SWMM Combined Model in the Kyungahn Stream 115 Chpater 5. Environmental Exposure Modeling of Pharmaceuticals with the PhATE TM SWMM Combined Model in the Kyungahn Stream Jeongim Park, Myung Hyun Kim, Kyungho Choi, Young Hee Kim, Min Young Kim In this chapter we attempt to estimate the integrated environmental concentration of pharmaceuticals used for both human and veterinary applications by using a PhATE TM SWMM combined model. Sulfamethoxazole and trimethoprim were the representative pharmaceuticals evaluated in the study since they are both used in human and animal medication. We simulated the environmental concentration of these chemicals using PhATE TM (in Chapter 3) and SWMM (in Chapter 4), but each of the two models underestimated the actual measured environmental concentration of sulfamethoxazole and trimethoprim. Therefore, it is apparent that a new approach, that will reduce the gap between simulated and measured values, is desired. 1. Materials and Methods 1.1 Target Pharmaceuticals As evidenced in Chapters 3 and 4, it is difficult to estimate environmental concentrations using a simulation model that accounts for only human or only veterinary pharmaceuticals in areas where pharmaceuticals used for both humans and animals have pathways into the environment. Therefore, in this integration study the target pharmaceuticals, sulfamethoxazole and trimethoprim, were chosen because they have applications as both human and veterinary pharmaceuticals. Furthermore, sulfamethoxazole and trimethoprim are frequently detected in Korean rivers and are ranked relatively highly in annual usage. Although not part of the model integration

136 116 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin study, acetaminophen and cimetidine were used for PhATE TM model validation in the Kyungahn stream. Chemical characteristics of the selected study pharmaceuticals are referenced in Chapters 3 and Model Application Using the PhATE TM model and the SWMM calibrated at Kyungahn stream, the simulation procedure for determining the integrated concentration of pharmaceuticals is outlined in Figure 5 1. Pharmaceuticals both for human and animals Total amount used for animals Total amount used for human Simulation of environmental concentration using SWMM Simulation of environmental concentration using PhATE Integrated environmental concentration Figure 5-1. Diagram for integrated environmental concentrations of pharmaceuticals used for both human and animals. Setting up the PhATE TM model for the Kyungahn stream The Kyungahn stream watershed has nine STPs and 440 thousand population. The nine STPs cover 80 percent of the populated area and the sewer system covers 65 percent. However Kyungahn STP and Kwangdong STP are located out of the study area, so they were excluded in PhATE TM model for Kyungahn stream (Figure 5 2). PhATE TM data for the Kyungahn stream was entered for 16 segments (Figure 5 3) in the same manner as done for the Han River data in Chapter 3 and resulted in the inclusion of the Kyungahn watershed in the PhATE TM model as seen in Figure 5 4.

137 Chapter 5. Environmental Exposure Modeling of Pharmaceuticals with the PhATE TM -SWMM Combined Model in the Kyungahn Stream 117 Figure 5-2. Site map of STPs in the Kyungahn stream watershed. STPs indicated with Gray solid circles (Kwangdond and Kyungahn STPs) were excluded in this study.

138 118 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin Table 5-1. Major sewage treatment plants (STPs) in the Kyungahn stream Name Flowrate (m 3 /d) Treatment type Population (thousands) Kyungahn* PID 71,119 Gwangju Activated sludge +HBR 71,119 Maesan 500 Long term aeration 300 Gonjiam Lagoons 50,122 Dochuk 2000 Lagoons 7,975 Mansun 150 Long term aeration 500 Opo 7000 Lagoons 26,775 Yongin B 132,550 Gwangdong* 1250 Long term aeration +HBR 12,302 * excluded in PhATE TM Gwangju STP 8 13 Gonjiam STP 7 Opo STP Maesan STP 5 Dochuk STP Mansun STP 2 Yongin STP 1 Figure 5-3. Segmentation scheme for PhATE TM in the Kyungahn stream.

139 Chapter 5. Environmental Exposure Modeling of Pharmaceuticals with the PhATE TM -SWMM Combined Model in the Kyungahn Stream 119 Figure 5-4. Input data setup in PhATE TM for the Kyungahn stream. 2. Results 2.1 PhATE TM model in the Kyungahn stream The environmental concentration of acetaminophen and cimetidine simulated with PhATE TM in the Kyungahn stream agreed well with the measured concentration. However sulfamethoxazole and trimethoprim were underestimated, as was expected. This data confirmed that sulfamethoxazole concentration in the Kyungahn stream is certainly influenced by veterinary usage. A similar phenomenon supported the detection of a lower concentration of acetaminophen and cimetidine in the upper Gonjiam stream where concentrated livestock farms and agricultural areas are located (Figure 5 5). Figure 5 7 underlined the contribution of sources; it presented higher concentration in the Gonjiam stream than in upper region of the Kyungahn stream. High concentrations in the beginning of the Kyungahn stream could be due to the effluents from Yongin STP which treated livestock sewage.

140 120 Environmental Risk Assessment of Pharmaceuticals: Model Application for Estimating Pharmaceutical Exposures in the Han River Basin (a) (b)

141 Chapter 5. Environmental Exposure Modeling of Pharmaceuticals with the PhATE TM -SWMM Combined Model in the Kyungahn Stream 121 (c) (d) Figure 5-5. PhATE TM modeled-segmental concentrations of pharmaceuticals, (a) Acetaminophen, (b) Cimetidine, (c) Trimethoprim, (d) Sulfamethoxazole