Exposure and Human Health Risk Assessment of Pharmaceutical in. Drinking Water. BOJANA MURISIC B.S., DePaul University, 2008 THESIS

Transcription

1 Exposure and Human Health Risk Assessment of Pharmaceutical in Drinking Water BY BOJANA MURISIC B.S., DePaul University, 2008 THESIS Submitted as a fulfillment of the requirements for the degree of Master of Science in Public Health Sciences in the Graduate College of the University of Illinois at Chicago, 2011 Chicago, Illinois Defense Committee: Serap Erdal, Chair Steven Lacey, Advisor Michael Cailas

2 There are a few people who have played a part in my upbringing and my undergraduate years and who particularly supported me and cared for me throughout my graduate studies at the University of Illinois at Chicago (UIC). First of all, those few people were my mother Nevenka and my father Ljubomir. Though they live in Serbia, our yearly encounters and our long-distance phone calls helped me transform into a mature person. In my parents I saw my idols. I saw how life should be lived and fought for and how obstacles ought to be hurdled. In the beginning of my graduate work at UIC when I felt disoriented and scared, my brother Nebojsa, who had similar experiences in his graduate studies, shared his wisdom and prompted me to be more committed to my studies. I started competing with him with an aim to get one step closer to the academic accomplishments of my brother- my second idol. My two best friends, Anaïs and Carmen, whom I lived together with in the beginning of my master studies at UIC, welcomed me so generously to their modest and small Skokie apartment. They extended their warm hands of friendship, compassion, and understanding towards me both in the moments when I doubted I will ever complete my graduate work and in those other moments of bliss and satisfaction when I was sliding through my assignments and exams. I use this opportunity to thank them for providing me with a warm home and their two gentle hearts. Finally, I would like to express my gratefulness to my fiancé Marco. Although we met two years ago, the friendship and love between us developed rapidly. Marco taught me how to be more responsible and more dedicated to my studies, but, he showed me how to do these two things through joy and laughter. Above all, Marco s unconditional support and care for me helped me become a better person. I dedicate this thesis to my mother Nevenka, my father Ljubomir, my brother Nebojsa, my two best friends Anaïs and Carmen, and my fiancé Marco. I thank all of them for keeping me in their thoughts and for keeping me close to their hearts in my childhood, my undergraduate years, and my graduate years. iii

3 ACKNOWLEDGEMENTS I would like to thank my thesis committee members, Dr. Serap Erdal, Dr. Steven Lacey, and Dr. Michael Cailas, for their guidance and assistance. I am especially grateful to the chair of my committee Dr.Erdal. In my two years at UIC she helped me stay afloat in the moments when I was losing my optimism and motivation. With her kindness and her subtle way of inspiring me, she was my guide in those moments when the light in that long tunnel seemed invisible and unreachable. I would also like to acknowledge a USGS scientist, Patrick Phillips, who was very helpful to me during the data collection process. His dataset on the concentrations of human pharmaceuticals in the US surface water was an important contribution in the conduct of my thesis. BM iv

4 TABLE OF CONTENTS CHAPTER PAGE I. INTRODUCTION A. Background B. Statement of the Problem... 1 C. Purpose of the Study II. CONCEPTUAL FRAMEWORK AND RELATED LITERATURE... 6 A. Conceptual Framework... 6 B. Review of Related Literature... 8 III. RISK ASSESSMENT PROCESS A. Planning Stage B. Hazard Identification Stage C. Dose-Response Assessment Stage D. Exposure Assessment Stage IV. RISK CHARACTERIZATION PROCESS A. Assumptions B. Uncertainties C. Conclusions D. Recommendations CITED LITEARTURE VITA 39 v

5 LIST OF TABLES TABLE PAGE I. PHARMACEUTICALS INCLUDED IN THIS RESEARCH II. HAZARD IDENTIFICATION STAGE III. TOXICITY BENCHMARK VALUES IN DOSE-RESPONSE ASSESSMENT STAGE IV. EXPOSURE PARAMETERS USED IN RISK ANALYSIS V. MEC, ADI REFERENCE, AND ADI CALCULATED VALUES IN EXPOSURE ASSESSMENT STAGE.. 23 VI. HQ AND HI VALUES IN EXPOSURE ASSESSMENT STAGE VII. PNEC VALUES AND MEC/PNEC RATIOS IN EXPOSURE ASSESSMENT STAGE VIII. TD/DWI AND I70 VALUES IN EXPOSURE ASSESSMENT STAGE IX. TOXICITY, PERSISTENCE, AND BIOACCUMULATION IN EXPOSURE ASSESSMENT STAGE vi

6 LIST OF FIGURES FIGURE PAGE 1. Comparison between the reference and calculated ADI values vii

7 LIST OF ABBREVIATIONS ADI CALCULATED ADI DW ADI DW+FISH ADI FISH ADI REFERENCE AT BCF BW DWI ED EF EPA GABA HI HQ I 70 IR DW IR DW+FISH IR FISH LD 50 LOAEL MEC Acceptable Daily Intake-Calculated Value Acceptable Daily Intake for Drinking Water Acceptable Daily Intake for Drinking Water and Fish Combined Acceptable Daily Intake for Fish Acceptable Daily Intake-Reference Value Averaging Time Bioconcentration Factor Body Weight Drinking Water Intake Exposure Duration Exposure Frequency Environmental Protection Agency Gamma Amino Butyric Acid Hazard Index Hazard Quotient Indirect Ingestion of Drinking Water Assuming Lifetime Exposure Ingestion Rate of Drinking Water Ingestion Rate of Drinking Water and Fish Ingestion Rate of Fish Lethal Dose 50-dose that causes death in 50% of tested animals Lowest Observable Adverse Effect Level Measured Environmental Concentrations viii

8 NOAEL NSAID PEC PhATE PNEC PNEC DW PNECDW DW+F PNEC FISH QAC RfD SSRI TD TD 50 USGS No Observable Adverse Effect Level Non-Steroidal Anti-Inflammatory Drug Predicted Environmental Concentrations using the PhATE model Pharmaceutical Assessment and Transport Evaluation Predicted No Effect Concentration Predicted No Effect Concentration for Drinking Water Predicted No Effect Concentration for Drinking Water and Fish Combined Predicted No Effect Concentration for Fish Quaternary Ammonium Compounds Reference Dose Selective Serotonin Reuptake Inhibitor Therapeutic Dose Tumor Dose 50-dose that induces tumors in 50% of tested animals United States Geological Survey ix

9 SUMMARY This thesis investigates the concentration of human pharmaceuticals in US drinking water by conducting human health risk assessment for 35 pharmaceuticals, belonging to 17 drug classes, for which the US environmental monitoring data is accessible. Using the measured environmental concentrations (MECs) in surface water from the U.S. Geological Survey (USGS) dataset and related papers, human exposure was assessed by calculating acceptable daily intakes (ADIs) for each one of the 35 pharmaceuticals. Calculated ADIs, which are an indication of potential exposure, are then compared to reference ADIs. Reference ADI values exist to protect the general population, as well as the sensitive subgroups of general population, likely to be exposed to human pharmaceuticals through two indirect exposure pathways: drinking water and fish. Predicted no effect concentrations (PNECs) for children and adults are then calculated using potential exposure values and compared to MECs. Both the ratios of potential exposure to reference exposure as well as the ratios of MECs to PNECs were very low. For all 35 compounds, these two low ratios show that there is no considerable human health risk from the presence of trace amounts of these pharmaceuticals in US drinking water and fish. x

10 I. INTRODUCTION A. Background Since the middle of the 19 th century, research on pharmaceuticals has been on the forefront of scientific advancement. A revolution in the preparation and use of pharmaceuticals, as well as an increase in their number, their effectiveness, and the range of illnesses and diseases that they provide relief for or they cure, gave people a reason to marvel at them. Pharmaceuticals have improved the quality of human life and have helped people live longer. But, this is only the case when pharmaceuticals are taken directly. When they are disposed of unused or when they are excreted by humans, pharmaceuticals indirectly reach drinking water as unmetabolized and/or metabolized products. Gradually, they enter fish and human bodies as biologically active compounds. As such compounds, pharmaceuticals could cause adverse health effects both in humans and in fish. B. Statement of the Problem Although pharmaceuticals enter bodies of fish and humans indirectly as biologically active compounds, and this could be dangerous for both groups of organisms, researchers have not started assessing human health risk associated with indirect exposure to fish and drinking water until the beginning of the 21 st century. They assessed ecological risk much earlier than they assessed human health risk. Two potential challenges have contributed to this outcome. Firstly, human health risk should not be assessed for general population only. It should be estimated for sensitive subpopulations as well. This group includes children, pregnant women or women of child-bearing age, and older people suffering from respiratory or cardiac problems. Assessing health risks for both the general population and the sensitive subgroups is a difficult task. Secondly, pharmaceuticals are present in our drinking water as a mixture, never as single compounds. Mixture of pharmaceuticals may have additive or even multiplicative adverse effects on humans. Assessing a human health risk of a mixture of pharmaceuticals is another challenge. 1

11 2 Researchers did not assess human health risk from indirect exposure of pharmaceuticals in drinking water and fish until the beginning of the 21 st century. As a result of the lack of knowledge and lack of concern, or perhaps because of the beneficial effects of pharmaceuticals in curing illnesses and diseases, the U.S. Environmental Protection Agency (EPA) does not have a single mandatory regulation for pharmaceuticals in drinking water. The EPA has established and enforces 170 drinking water standards, but, none of them are for pharmaceuticals (Halling-Sørensen et al., 1998). In many instances, this drinking water contains antimicrobial agents from antibiotics and quaternary ammonium compounds (QACs) from disinfectants. Antimicrobial agents and QACs allow bacteria to develop resistance to microbial agents and to disturb denitryifying bacteria which are constituents of water treatment plants (Hallig-Sørensen et al., 1998). C. Purpose of the Study The purpose of this study is to evaluate whether indirect exposure to 35 pharmaceuticals through consumption of drinking water and fish results in any adverse human health effects. This evaluation is done using the measured concentrations of 35 pharmaceuticals most often present in the U.S. surface water. It is essential to point out that the concentration of pharmaceuticals in surface water is higher than in drinking water. In the water purifying facilities, surface water goes through several processes to ensure that the drinking water has only the trace amount of pharmaceuticals present. Thus, we look at the worst case scenario of pharmaceuticals in surface water with a certainty that in drinking water the concentration of pharmaceuticals can only be lower. The risk connected to exposure is assessed in 5 stages in accordance with the human health risk paradigm established by the National Research Council in 1983 (NRC, 1983) and practiced by the U.S. EPA (1991, 1998): planning stage, hazard identification stage, dose-response assessment stage, exposure assessment stage, and risk characterization stage. In this thesis, risk assessment is separated from risk characterization. While risk assessment consists of the following four stages: planning, hazard

12 3 identification, dose-response assessment, and exposure assessment; risk characterization is a section on its own including the following four parts: assumptions, uncertainties, conclusions, and recommendations. In the planning stage, hazards of concern associated with 35 pharmaceuticals are explained, along with origins of hazards and time-scale for these hazards to cause toxic effects. In the hazard identification stage, toxicodynamics and the toxicokinetics processes are discussed. Toxicodynamics process explains what health problems drugs may cause and toxicokinetics process explains what happens once the pharmaceuticals reach human body. In this thesis, planning stage is joined with the hazard identification stage and summarized in Tables I and II. In the dose-response assessment stage, the probability of people experiencing adverse health problems at different exposure levels is documented for different end-points. The toxicity benchmark values such as lethal dose 50 (LD 50 ), no observable adverse effect level (NOAEL), and acceptable daily intake (ADI REFERENCE ) are also compiled and presented. For carcinogenic pharmaceuticals, the doses that induce tumors in half of the tested animals (TD 50 ) are also presented. Exposure assessment stage consisted of three parts. In the first part of the exposure assessment stage, MECs which are measurements of 35 pharmaceuticals in surface water, along with appropriate exposure parameters (ingestion rate, exposure frequency, exposure duration, etc.) were used to calculate average daily intake via fish consumption for adults and children (i.e., ADI DW, FISH, DW+FISH, CHILDREN and ADI DW, FISH, DW+FISH, ADULTS ). ADIs are protective values of daily indirect human exposure to pharmaceuticals in drinking water and fish. Besides the calculated ADI values, reference ADI values are also considered in this analysis. ADI REFERENCE values, that come from the literature and that researchers calculated using the toxicity benchmark values (either NOAEL or LOAEL) and the uncertainty factors used to extrapolate from animal to human assessment, are divided by the their respective calculated ADI DW CHILDREN, ADI FISH CHILDREN, and ADI DW+FISH CHILDREN values in order to obtain an estimate of hazard quotient values (HQ DW CHILDREN, HQ FISH CHILDREN, and HQ DW+FISH CHILDREN ). In every risk assessment, there is no chance of adverse health effect if this quotient is less than 1. In our analysis, HQ of less than 1 means that the daily intake of each one of 35 pharmaceuticals assessed in drinking water and fish is much

13 4 lower than a reference acceptable intake calculated using NOAEL and LOAEL values obtained from lab experiments. Later, hazard quotient for the two pathways of concern, i.e., drinking water consumption and fish consumption, (HQ DW CHILDREN and HQ FISH CHILDREN ) are added to calculate the hazard index (HI). HI values in most cases match the HQ DW+FISH CHILDREN values. Finally, as a last step of the first stage of the exposure assessment process, we turn to a worst case scenario- a human health risk had a mixture of all 35 pharmaceuticals been present in surface water. So, we add the HI values for all 35 pharmaceuticals and obtain a HI. In the second part, predicted no effect concentrations (PNECs) for both the children and the adults and for all the exposure pathways of drinking water consumption only, fish consumption only, and drinking water and fish consumption combined are presented. PNECs look at the calculated acceptable daily intakes of drinking water and fish and the other parameters such as the body weights, ingestion rates, etc to derive an exposure value that is predicted not to cause any adverse outcome in humans. So, if humans are exposed indirectly to PNECs or the values below the PNECs through the ingestion of pharmaceuticals in fish and drinking water, they will not suffer any adverse health outcomes. PNECs for children only are divided with the MEC values of pharmaceuticals in surface water. If the ratio of MEC/ PNEC CHILDREN is smaller than 1, sensitive population as well as the general population is not likely to suffer from pharmaceuticals present in surface water and fish. In this second part of the exposure assessment stage, we pay a special attention to children because human health risk assessment has an objective to protect first and foremost the sensitive subgroups and then the general population. If the risk is assessed correctly, the measures taken to protect the children will protect adults as well. In the third part of the exposure assessment stage, MECs are multiplied by 2 l/day which are adult water intake values. This is done to get the drinking water intake values of 35 pharmaceuticals in mg/day (DWIs). DWIs are then divided by the therapeutic doses (TD) recommended for 35 pharmaceuticals and obtained from the literature review. TDs exist to indicate the daily amount of pharmaceuticals that are effective in treating illnesses, but that do not pose a threat to human health. TD/DWI ratio is desired to be much higher than 1 which would show that the daily efficacious therapeutic doses of pharmaceuticals

14 5 taken directly are higher than the level of pharmaceuticals that humans take indirectly via drinking water. DWIs are then multiplied with 70 (lifetime exposure duration=number of years in an average human lifetime) and 365 (number of days in one year) to estimate the lifetime indirect consumption of pharmaceuticals in drinking water in micrograms for 70 years (I 70 ). The ratio of TD and I 70 provides daily ingestion (I 70 daily dose). So, in this last part of the exposure assessment stage, a comparison is made between two I 70 values and a TD. The first I 70 value is a lifetime indirect exposure to pharmaceuticals via drinking water. The second I 70 value, which is unitless, is a daily indirect exposure to pharmaceuticals via drinking water. The unitless I 70 value is expressed in terms of daily TD and is desired to be lower than 1 which would mean that the daily indirect exposure to pharmaceuticals via drinking water is less than the daily therapeutic dose of pharmaceuticals humans are allowed to take directly. In the final stage of this risk assessment, the risk characterization stage, the assumptions, uncertainties, and conclusions are presented along with risk assessment results. Additionally, recommendations are provided to the members of general public on how to safely and properly dispose unused pharmaceuticals. Furthermore, open issues for future research are discussed along with guidance to regulatory bodies that aim at protecting the public from potential adverse health effects associated with exposure to pharmaceuticals in the environment.

15 II. CONCEPTUAL FRAMEWORK AND RELATED LITERATURE A. Conceptual Framework Most of the studies that assessed human health risk to pharmaceuticals in the environment involved indirect exposures through consumption of drinking water and fish and had a goal to obtain ADI values using MECs in surface water as well as appropriate exposure parameters (e.g., IR, EF, ED, BW). Once ADI values were calculated, they were compared to the reference ADI values. Further, HQs were calculated as ratios of ADI CALCULATED and ADI REFERENCE. Having an HQ lower than unity means that noncancer adverse effects in humans are unlikely. Some of these studies went further by calculating ADI values separately for fish consumption and for drinking water consumption, by using the fish and water ingestion rates respectively. Once these values were obtained, researchers could simply add the two ADI values to obtain a hazard index (HI). HI lower than 1 means that no non-cancer adverse health effects are anticipated in humans. Gradually, researchers studying human health risk assessment leaped two steps further. In the first step, researchers estimated PNEC values from calculated ADI and standard exposure parameters (IR, EF, ED, BW) (i.e., back-ward risk assessment process). Most often, PNEC values are calculated both for children and adults. For children, ADI values are lower because children have lower body weight and intake rate. PNEC calculation does not only involve two age groups (adults and children) in the population, but, it is also performed for three likely indirect human exposure pathways: ingestion of drinking water, ingestion of fish, and ingestion of both combined. So, most of the newer studies calculate six PNEC values for: oral intake of drinking water by children; oral intake of drinking water by adults; fish consumption by children; fish consumption by adults; oral intake of drinking water and consumption of fish pathways combined for children; and 6

16 7 oral intake of drinking water and consumption of fish pathways combined for adults. Although calculated, most often PNEC values for adults are not used in further analysis. Rather, PNEC values for children are used since they are lower and, thus, protective for the adults as well. Of relevance here is that the exposure parameters involved in the calculation of PNEC for children and adults differ. For instance, IR of drinking water for adults is 2 l/day and for children it is only 1 l/day; children consume less fish; AT for adults is much longer because they have lived longer than children, ED is naturally longer for adults than for children, and the BW of adults is larger than that of children. Once PNECs for children, i.e., safe concentrations, are calculated, they are compared to MEC values to ascertain whether unacceptable health risk may be present for the exposed population. This provides an indirect method for assessing health risk, i.e., by calculating the ratio of measured environmental concentration (MEC) to safe concentration (i.e., PNEC or risk-based concentration). It is desired to have the MEC/PNEC ratios for all three exposure pathways much lower than 1 (i.e., measured concentration less than the safe concentration). This low ratio is protective even for sensitive populations. Having MEC/PNEC ratio bigger than 1 or close to 1 would be very concerning since it would mean that the measured concentrations of pharmaceuticals in the drinking water and fish are very likely to cause adverse effects in humans. In the second step in these newer studies, researchers focus on estimating DWI values in mg/day. They do this by multiplying the MECs (in mg/l) by drinking water intake rate, which is 2 l/day. Therapeutic doses of pharmaceuticals, which humans are prescribed by their physicians and take directly, are compiled and then, are compared to estimated DWI of pharmaceuticals. It is desired that the ratio of TD/DWI is much lower than 1. I 70 values based on average human lifetime of 70 years are also calculated by multiplying DWI with 70 (years) and 365 (days). I 70 values are first calculated in micrograms and then, the ratio of TD/I 70 (µg) is estimated to obtain a unitless I 70 daily dose. It is hoped that the I 70 daily dose is lower than 1.

17 8 B. Review of Related Literature Humans are complex species. We have different genetic backgrounds, are exposed to different environmental hazards, belong to different age groups and sexes, and vary in our susceptibility to diseases. All these factors play a role in the risk assessment process. A host of factors determine how likely individuals are to suffer from adverse effects when consuming different levels of pharmaceuticals found indirectly in fish and drinking water. To add to this, humans are more likely to suffer from the mixture of pharmaceuticals found in our drinking water than are aquatic species. On some humans, a mixture of pharmaceuticals may have additive effects, multiplicative effects, or maybe no effects at all. Predicting what type of effect a mixture of pharmaceuticals will have on a particular human being is extremely challenging. Two separate studies that assessed the human risk associated with oral exposure to trace amounts of pharmaceuticals indirectly through drinking water and fish, did the analysis by following two different approaches. In the first study, Schwab et al. (2005) presented human health risk assessment for 26 pharmaceuticals often found in US surface waters. They first gathered the measured environmental concentrations from the peer-reviewed literature (Kolpin et al., 2002, Hilton and Thomas, 2003, Bratton et al., 2003, Calamari et al., 2003, Hirsch et al., 1996, Kolpin et al., 2004, Snyder et al., 2001, Ternes et al., 1998, Vanderford et al., 2003, Weigel et al., 2004, and Zuccato et al., 2000). They derived the ADI values for each one of these compounds by using the MECs. Subsequently, they used these ADI CALCULATED values to obtain the PNEC values for drinking water, fish, and drinking water and fish combined. They further compared the PNEC values with the MEC values. The research presented in this thesis followed the same approach in risk analysis. In their paper, Schwab et al. also compared the PNEC values to the predicted environmental concentrations (PECs) obtained from the PhATE model, which assumes low river flow and no depletion occurring during metabolism (Schwab et al., 2005). Schwab et al. showed that both the ratios of MECs to PNECs and the ratios of PECs to PNECs are comparably low. For all 26 pharmaceuticals that were analyzed, these low ratios signal that the presence of trace amounts

18 9 of these pharmaceuticals in surface water and drinking water poses no appreciable human health risk (Schwab et al., 2005). In the second study, Webb et al. (2003) were similarly concerned about potential threats that the presence of pharmaceuticals or their metabolites may pose to human health. They made a comparison between the MECs present in German drinking water and the therapeutic doses (TDs) for each one of these pharmaceuticals. The therapeutic doses in this paper were obtained from Ternes (2001) and Ternes (2001b). Their comparison showed that for every pharmaceutical they examined, the TDs were at least three orders of magnitude higher than the MECs found in drinking water (Webb et al., 2003). A similar approach was utilized in this thesis as an additional approach to risk analysis of pharmaceuticals in the environment. For some compounds, Webb et al. went a step further by comparing the benchmark exposure to ADI values involved in meat and food residues following veterinary use. Webb et al. concluded that the potential benchmark exposure was less than the ADI values for each compound they analyzed, implying that there are no substantial concerns with regard to indirect exposure via drinking water (Webb et al., 2003). In the third study by Snyder and Snyder on the toxicological significance of endocrine disruptors in drinking water, the authors assessed a human health risk associated with compounds that are likely to show endocrine disrupting properties. They took surface water samples from 17 participating water facilities throughout the United States. 14 pharmaceuticals that they involved in their analysis are also present in this thesis. The authors calculated the ADI REFERENCE and DWEL (drinking water equivalent level which are the same as DWI in my thesis), classified drugs into special drug categories, evaluated their carcinogenic properties, and measured their concentrations in surface water. The authors did an interesting comparison in their analysis. They created a chart showing the comparison of lifetime risks that are associated with different events or situations in human life. According to them, the risk of developing cancer from environmental agents (ie indirect consumption of pharmaceuticals in drinking water and fish) is higher than the risk involved in riding a bicycle (5 in 1,000,000) but lower than the risk of developing a skin cancer (1 in 5) (Snyder& Snyder,

19 ). Also, according to these 2 authors and the World Health Organization, chemical contaminants found in drinking water supplies include many other organic and inorganic chemicals. Researchers calculate the guideline values for individual chemicals only without considering the likelihood of the interaction between substances present in this mixture. There is a large margin of uncertainty when assessing human health risk associated with indirect exposure to pharmaceuticals via drinking water. According to the authors, this margin is worrisome (Snyder& Snyder, 2008). In the fourth source, Benotti et al. analyzed the concentrations of human pharmaceuticals in source, finished, and distribution water systems. Out of 51 compounds targeted in their analysis, Benotti et al. detected most of the pharmaceuticals in source waters, while only 3 of them were detected in finished or distribution systems. Among those three, meprobamate and phenytoin are analyzed this thesis. The reason why the Benotti et al. found pharmaceuticals to be often present in source water but seldom present in drinking water is because of the chlorine or ozone oxidation which remove these pharmaceuticals before they get to the drinking water (Benotti et al., 2009). As their analysis show, meprobamate and phenytoin are resistant to ozone and chlorine oxidation. In the fifth source, Kolpin et al. from the U.S. Geological Survey used 5 analytical methods to measure the concentrations of 95 organic compounds in streams across 30 states during 1999 and Their analysis shows that for most compounds, the measured concentrations rarely exceeded drinking water guidelines (Kolpin et al., 2002). However, the authors say that little is known about the interactive effects (such as synergistic effects) which could occur from mixtures of pharmaceuticals present in the environment. They further encourage the researchers to obtain data on metabolites of pharmaceuticals in order to understand not only the transport of pharmaceuticals in the hydrologic system but also the cumulated effect on both the pharmaceuticals and their metabolites on human health (Kolpin et al., 2002).

20 III. RISK ASSESSMENT PROCESS A. Planning Stage In this first stage of human risk assessment, the first task was to gather pharmaceuticals and their metabolites that are most often detected in US surface waters. Based on a comprehensive literature review (Webb et al., Schwab et al., Snyder& Snyder, Benotti et al., and Kolpin et al.), 35 compounds are identified as potential chemicals of concern and these compounds are classified into 17 different groups based on their primarily role in therapy. Table I summarizes the compounds examined for human health risk analysis in this research. Information for this table was found using three websites (rxlist.com, drugs.com, medicinenet.com). Humans are indirectly exposed to pharmaceuticals through drinking water and fish. But, how do trace amounts of pharmaceuticals get to drinking water? The answer to this question is given in this first stage of risk assessment. The chain begins with the pharmaceutical companies which manufacture these drugs, humans who take them, excrete them through urine and feces, and throw the left over drugs to trash. Human excretions and left over medicines, along with the residues from pharmaceutical manufacturing and hospital use, go to sewers. Humans get exposed to contaminated water either by drinking it, bathing with it, or by consuming fish that has pharmaceutical residues. Thus, the main human exposure routes to all 35 pharmaceuticals included in this analysis are ingestion of drinking water; consumption of contaminated fish; and dermal exposure during bathing. It is assumed that the dermal exposure is minimal and is thus not taken into account in cumulative risk assessment. Toxic effects in humans may occur either from a single toxic dose, from repeated ingestion of larger doses, or from chronic ingestion of smaller doses of these pharmaceuticals. 11

21 12 B. Hazard Identification Stage In the hazard identification stage, the second stage of risk assessment process, hazard potential of the chemical of concern and the nature, strength and extent of this hazard are ascertained once the chemical enters the human body. The process of a drug or a mixture of drugs being absorbed, distributed, metabolized, and excreted by human body is the primary focus of toxicokinetics. Toxicodynamics explains what effects a particular drug or a mixture of them has/have on our bodies. Information regarding the toxicodynamics and toxicokinetics processes were obtained using the above mentioned websites (rxlist.com, drugs.com, and medicinenet.com). Table II presented below gathers information on the name of 35 pharmaceuticals, their molecular formulae, indications, dosage, side effects one might suffer when taking them, pharmacodynamic, and pharmacokinetic properties.

22 13 TABLE I PHARMACEUTICALS INCLUDED IN THIS RESEARCH a,b,c Number Pharmaceutical Name Pharmaceutical Class a,b,c 1 Acetaminophen Antipyretic 2 Albuterol Antiasthmatic 3 Atenolol Beta-blocker 4 Atorvastatin Antilipidemic 5 Carbamazepine Anticonvulsant 6 Chlortetracycline Antibiotic 7 Cimetidine Antacid 8 Ciprofloxacin Antibiotic 9 Diazepam Antianxiety agent 10 Diclofenac Anti-inflammatory 11 Dehydronifedipine Antianginal 12 Digoxigenin Against heart failure 13 Digoxin Against heart failure 14 Diltiazem Antihypertensive 15 Doxycycline Antibiotic 16 Enalaprilat Antihypertensive 17 Erythromycin-H 2 O Antibiotic 18 Fluoxetine Antidepressant 19 Gemfibrozil Antilipidemic 20 Ibuprofen Anti-inflammatory 21 Lincomycin Antibiotic 22 Meprobamate Antianxietyagent 23 Metformin Antidiabetic 24 Naproxen Anti-inflammatory 25 Norfloxacin Antibiotic 26 Oxytetracycline Antibiotic 27 Paroxetine metabolite Antidepressant 28 Phenytoin Anticonvulsant 29 Ranitidine Ulcer agent 30 Risperidone Antipsychotic 31 Sulfamethoxazole Antibiotic 32 Sulfathiazole Antibiotic 33 Tetracycline Antibiotic 34 Trimethoprim Antibiotic 35 Warfarin Antibiotic a rxlist.com b drugs.com c medicinenet.com

23 14 TABLE II HAZARD IDENTIFICATION STAGE a,b,c Pharmaceutical Molecular Indications Dosage Side Pharmcodyn. Pharmacokin. Name Formula (mg/day) Effects Acetaminophen C 8 H 9 NO 2 pain, fever 4000 stomach pain elevates pain rapid absorption threshold from GI tract Albuterol C 13 H 21 NO 3 broncho- - fine tremor - - spasm Atenolol C 14 H 22 N 2 O 3 hyper- 50 slow/uneven blocks beta1- absorbed rapidly tension heartbeat receptor excreted renally a. pectoris feet & ankle swelling Atorvastatin X d * 3H 2 O elevated muscle pain affects liver eliminated in cholesterol bile Carbamazepine C 15 H 12 N 2 O convulsions - low white depresses bound to epilepsy blood cell polysynaptic plasma proteins count reflexes metabolized in mood changes liver Chlortetracycline C 22 H 24 N 2 O 8 respiratory 1500 fever chills inhibits protein well absorbed infections pale skin synthesis Cimetidine C 10 H 16 N 6 S duodenal 800 trouble inhibits action absorbed rapidly ulcer breathing of histamine excreted in urine bruising Ciprofloxacin X e * H 2 O urinary 500 joint pain inhibits absorbed rapidly infections confusion topoisomerase distrib. in body Diazepam X f anxiety 20 depressed acts on well absorbed disorders mood neurotransm. excreted in urine GABA Diclofenac X g arthritis stroke & inhibits well absorbed infarction risk prostaglandin excreted biliary sythesis Dehydro- C 17 H 16 N 2 O 6 hypertens nifedipine Digoxigenin C 23 H 34 O Digoxin C 41 H 64 O 14 heart - - decreases heart eliminated by conditions function liver Diltiazem C 22 H 26 N 2 O 4 S hypertens swelling/rapid inhibits influx well absorbed weight gain calcium ions Doxicycline C 22 H 24 N 2 O 8 syphilis - vision problems inhibits protein eliminated by chlamydia synthesis liver acne malaria Enalaprilat X h * 2H 2 O hypertens. 6 congestive heart increases serum renally excreted failure potassium Erythromycin- C 43 H 75 NO 16 respiratory 1600 uneven heartbeat inhibits protein well absorbed infections synthesis biliary excreted

24 15 TABLE II (CONT.) HAZARD IDENTIFICATION STAGE a,b,c Pharmaceutical Molecular Indications Dosage Side Pharmcodyn. Pharmacokin. Name Formula (mg/day) Effects Fluoxetine X i * HCl depression insomnia inhibits hepatic bulimia mood changes serotonin metabolism uptake Gemfibrozil C 15 H 22 O 3 high trigly sharp stomach decreases hepatic well absorbed ceride levels pain extraction of free fatty acids Ibuprofen C 13 H 18 O 2 fever 3200 severe stiffness inhibits prostag- eliminated in inflammation landin synthesis urine Lincomycin X j * H 2 O serious 600 chills/flu - biliary excretion infections symptoms Meprobamate C 9 H 18 N 2 O 4 anxiety 2400 drowsiness acts on thalamus - disorders dizziness & limbic system Metformin C 4 H 11 N 5 HCl high blood 1000 weakness decreases hepatic urine excretion glucose slow heart rate glucose product. Naproxen C 14 H 14 O 3 arthritis 500 heartburn inhibits prosta- urine excretion gastric ulcers glandin synthesis Norfloxacin C 16 H 18 FN 3 O 3 urinary tract 800 seizures promotes DNA bliary excretion infections hallucinations breakage Oxytetracycline C 22 H 24 N 2 O 9 typhus fever 250 burning eye inhibits protein biliary excretion pneumonia sensation synthesis Paroxetine inhibits serotonin - metabolite uptake Phenytoin C 15 H 12 N 2 O 2 seizures 300 depression inhibits seizure biliary excretion hostility activity Ranitidine X k * HCl duodenal 300 fatigue inhibits urine excretion ulcer insomnia histamine action Risperidone C 23 H 27 FN 4 O 2 schizophrenia - confusion affects dopamine urine excretion bipolar disorder sweating & serotonin Sulfametho- C 10 H 11 N 3 O 3 S urinary inhibits bacterial urine excretion xazole infections synthesis malaria Sulfathiazole C 9 H 9 N S 2 common infections Tetracycline C 22 H 24 N 2 O 8 pneumonia - tiredness - - genital infections Trimethoprim C 14 H 18 N 4 O 3 urinary tract 200 skin eruptions inhibits bacterial urine excretion infections synthesis Warfarin C 19 H 16 O heart - pale skin - - loss of hair a rxlist.com g X = C 14 H 10 Cl 2 NNaO 2 b drugs.com h X = C 18 H 24 N 2 O 5 c medicinenet.com i X = C 17 H 18 F 3 NO d X = (C33H34FN 2 O 5 ) 2 Ca j X = C 18 H 34 N 2 O 6 S * HCl e X = C 17 H 18 FN 3 O 3 * HCl k X = C 13 H 22 N 4 O 3 S f X = C 16 H 13 ClN 2 O

25 16 C. Dose-Response Assessment Stage In dose-response assessment step, the probability of experiencing adverse health effects when exposed to different levels of each pharmaceutical under study is ascertained. The benchmark values derived from toxicity assessment (i.e., dose-response assessment) studies in animal models (i.e., LD50 and NOAEL values) for each compound of interest is obtained from the literature. Most of the data comes from bioassays performed on rats and mice. LD50 is a lethal dose that causes mortality in at least 50% of experimental animals. The rodents are given very large amounts of compounds, significantly higher than the amounts humans are exposed to on a daily basis through fish and drinking water intake. The animal studies usually involve 50 animals for dose and sex group. Small sample size in animal toxicity studies requires usage of high doses in experiments to reduce the possibility of missing an effect due to low statistical power. This practice allows for successful extrapolation from animal exposure in high doses to human exposure in low doses. Table III shows toxicity benchmark values derived from mice studies. Based on the literature search, three pharmaceuticals are identified to exert carcinogenic effects. For those that are not yet shown to induce carcinogenic effects, a linear dose-assessment is applied, relying on a threshold hypothesis. This hypothesis signifies that there is a range of exposures from zero to a value that can be tolerated without any chances of toxic effect expression. This value is called noobservable-adverse-effect-level or NOAEL. NOAEL values are presented in Table III.. For carcinogenic pharmaceuticals, TD 50 values, which are the doses that induce tumors in 50% of the tested animals, are also included in Table III. NOAEL values allow calculation of ADI REFERENCE values - estimates of maximum daily oral exposures unlikely to cause any adverse health effects as a result of a lifetime exposure to a compound for all exposed individuals including the sensitive subpopulations (also shown in Table III). In Schwab et al. 2005, these values are calculated by dividing the 5 uncertainty factors with NOAEL values. Uncertainty factors account for extrapolation from animal to human studies, intraspecies differences, quality of the data and the study, and the exposure duration.

26 17 TABLE III TOXICITY BENCHMARK VALUES IN DOSE-RESPONSE STAGE Number Pharmaceutical LD 50 NOAEL ADI REFERENCE TD 50 (mg/kg) a (mg/kg-day) b (µg/kg-day) (mg/kg-day) d 1 Acetaminophen mcg/kg-day c Albuterol b / 3 Atenolol > b / 4 Atorvastatin b NED 5 Carbamazepine b NED 6 Chlortetracycline b / 7 Cimetidine mcg/kg-day c / 8 Ciprofloxacin >2000 not available 29 b / 9 Diazepam b / 10 Diclofenac b / 11 Dehydronifedipine 4000 not available 100 b / 12 Digoxigenin b / 13 Digoxin b / 14 Diltiazem mcg/kg-day c / 15 Doxycycline b / 16 Enalaprilat mcg/kg-day c / 17 Erythromycin-H 2 O b / 18 Fluoxetine b / 19 Gemfibrozil b Ibuprofen mcg/kg-day c / 21 Lincomycin > mcg/kg-day c / 22 Meprobamate b / 23 Metformin mcg/kg-day c / 24 Naproxen b / 25 Norfloxacin > b / 26 Oxytetracycline mcg/kg-day c / 27 Paroxetine metabolite b / 28 Phenytoin b Ranitidine > not available / 30 Risperidone b NED 31 Sulfamethoxazole b / 32 Sulfathiazole 1320 not available 50 b / 33 Tetracycline b / 34 Trimethoprim b / 35 Warfarin b / a msdsonline.com b Snyder & Snyder (2008). c Schwab et al. (2005). d e NED- no established dose for carcinogens

27 18 D. Exposure Assessment Stage In the exposure assessment step of risk assessment, how people come into contact with the chemical of concern is one of the most important questions answered. Here, three exposure pathways are considered: ingestion of drinking water; consumption of fish; and consumption of drinking water and fish combined. In this step, average daily intake (or dose) (ADI) for each pathway is estimated using exposure concentration in the exposure medium of interest (e.g., drinking water) and a number of exposure factors that are subject-specific (e.g., intake rate, body weight) and/or indicate time-scale of exposure (e.g., exposure time, frequency and duration). In order to estimate ADI for drinking water ingestion pathway, MECs in surface water as well as body weight, exposure frequency and duration, ingestion rate, bioconcentration factor, and averaging time are needed. Similarly, ADI calculations for each pathway are performed for both the children and adult receptors (i.e., ADI DW CHILDREN, ADI DW ADULTS, ADI FISH CHILDREN, ADI FISH ADULTS, ADI DW+F CHILDREN, ADI DW+F ADULTS ) using the equations given in Equations 1-6 below. These equations are commonly used by the US. EPA for developing risk-based concentrations and/or exposure limits/standards to protect the general public from pollutants in the environment (Schwab et al. 2005). ADI ADI ADI DWCHILDREN DW ADULTS FISHCHILDREN MEC IRDW CHILDREN EF EDCHILDREN CF = (1) BWCHILDREN ATCHILDREN MEC IRDW ADULTS EF EDADULTS CF = (2) BWADULTS ATADULTS MEC IRFISH CHILDREN EF EDCHILDREN CF = (3) BWCHILDREN ATCHILDREN ADI FISHADULTS MEC IRFISH ADULTS EF EDADULTS CF = (4) BWADULTS ATADULTS

28 19 ADI DW + FISHCHILDREN = MEC ( IR FISH CHILDREN BCF + IR BWCHILDREN AT DW CHILDREN ) CHILDREN EF ED CHILDREN CF (5) ADI DW + FISHADULTS = MEC ( IR FISH ADULTS BCF + IR BWADULTS AT DW ADULTS ) ADULTS EF ED ADULTS CF (6) Here, MEC, IR for children and adults, EF, ED for children and adults, CF, BW for both age groups, and AT for both age groups are all given in Table IV. TABLE IV EXPOSURE PARAMETERS USED IN RISK ANALYSIS IR DW IR FISH EF ED BW AT CF a (1/day) a (kg/day) a (days/year) a (years) a (kg) a (days) a Children e -6 Adults e -6 a Schwab et al. (2005) Table V summarizes the results for ADI DW CHILDREN, ADI FISH CHILDREN, ADI DW+FISH CHILDREN. The MECs and ADI REFERENCE values are tabulated for ease of referencing. Because all of ADI DW ADULTS, ADI FISH ADULTS,

29 20 and ADI DW+FISH ADULTS values are lower than those for children, they were not included in Table V. Comparison between the ADI REFERENCE values and the ADI calculated values for children and for three exposure pathways is also illustrated in Figure 1. Looking at Table V and Figure 1, one can see how for each one of 35 compounds analyzed in this thesis, the reference ADI values are much higher than the calculated ADI values for drinking water and fish ingestion. Even when one combines the water and fish ingestion and looks at them together, ADI reference values are still much higher than the ADI calculated values. This large margin between the 2 ADI values shows that the indirect ingestion of pharmaceuticals through fish and drinking does not show a concern for human health. The calculated ADI values for children and for all three exposure pathways are then compared to reference ADI values. The ratio of ADI CALCULATED to ADI REFERENCE is a hazard quotient or HQ for each pathway. Further, HQ values for children for each pathway and each pharmaceutical individually are added to obtain the hazard index (HI). It is interesting to note that HQ DW+FISH CHILDREN values are equal to HI values. Looking at Table VI, where all HQ children values are presented along with the HI values, one can see that for every pharmaceutical HQ values for each exposure pathway are lower than 1. Even when HQ values are added to get HI, those HI values are again lower than 1. HQ and HI values lower than 1 indicate that the risk to humans from indirect exposure to pharmaceuticals via drinking water and fish is negligible. Even in the worst case scenario, when HI values for every pharmaceutical are summed, one gets a HI of 2.06e -4. So, provided that we assume the worst case scenario in which the concentration of 35 pharmaceuticals in drinking water is as high as it is in surface water, and provided that all 35 pharmaceuticals are present in the mixture, hazard index is still much lower than 1. HI is also summarized in Table VI. As an alternative method to traditional exposure characterization by estimating the ADI, we predicted no effect concentration (PNEC) values for both the children as well as adults. These values, which serve as safe concentrations in the environment, were calculated using ADI CALCULATED presented in Table V, BW, IR, AT, ED, EF, and BCF (only for the fish consumption pathway). The PNEC calculation is performed not only for two groups of humans (adults and children), but, also for three pathways which

30 resemble the most likely exposures throughout human lifetime. People have to drink water and most people eat fish. Therefore, we calculated six PNEC values using the equations shown below: 21 PNEC PNEC PNEC DWCHILDREN DWADULT FISHCHILDREN ADIDWCHILDREN BWCHILDREN ATCHILDREN CF = (7) IRDWCHILDREN EF EDCHILDREN ADIDWADULT BWADULT ATADULT CF = (8) IRDWADULT EF EDADULT ADIFISHCHILDREN BWCHILDREN ATCHILDREN CF = (9) IRFISHCHILDREN EF EDCHILDREN PNEC FISHADULT ADIFISHADULT BWADULT ATADULT CF = (10) IRFISHADULT EF EDADULT PNEC PNEC DW + FISHCHILDREN DW + FISHADULT ADI = ( IR ADI = ( IR DW + FISHCHILDREN DWCHILDREN DW + FISHADULT DWADULT BW + BCF IR BW + BCF IR CHILDREN FISHCHILDREN ADULT FISHADULT ATCHILDREN CF ) EF ED ATADULT CF ) EF ED ADULT CHILDREN (11) (12) where all the ADIs were calculated using Equations 7-12 and the conversion factor is The exposure parameters such as BW, AT, IR, EF, and ED, tabulated in Table IV formed the basis of these calculations. The calculated PNECs for children are compared to MEC values reported in the literature to determine whether measured concentrations of pharmaceuticals in drinking water exceeded their respective safe concentration (PNEC). It is desired to have the MEC/PNEC ratios for all three exposure pathways much lower than one. This low ratio is deemed to be protective even for sensitive populations. Having MEC/PNEC ratios larger than unity or close to one would be very concerning since it would signify that the measured concentrations of pharmaceuticals in the drinking water and fish will likely cause adverse effects in humans, potentially adverse effects that are synergistic given the fact that pharmaceuticals are present in the environment in a mixture. Table VII summarizes these calculations. Specifically, it show the results for PNEC DW_CHILDREN, PNEC DW_ADULTS, PNEC FISH_CHILDREN, PNEC FISH_ADULTS,, PNEC DW+FISH_CHILDREN, PNEC DW+FISH_ADULTS, along with the following three ratios:

31 22 PNEC DW_CHILDREN /MEC; PNEC FISH_CHILDREN /MEC; PNEC DW+FISH_CHILDREN /MEC. Equations 7-12 are general average daily intake or dose estimation equations consistent with those used by US.EPA (Schwab et al. 2005) for developing exposure concentration limits to protect the public from hazardous exposure to pollutant in various environmental compartments.

32 23 TABLE V MEC, ADI REFERENCE AND ADI CALCULATED VALUES IN EXPOSURE ASSESSMENT STAGE Pharmaceutical MEC ADI REFERENCE ADI DWCHILDREN ADI FISHCHILDREN ADI DW+FISHCHILDREN (ng/l) a (mg/kg-day) a (mg/kg-day) (mg/kg-day) (mg/kg-day) Acetaminophen e e -5 7e -4 Albuterol e e e -6 Atenolol e e e -6 Atorvastatin e e e -8 Carbamazepine e e e -6 Chlortetracycline e e e -5 Cimetidine e e e -5 Ciprofloxacin e e e -6 Diazepam e e e -8 Diclofenac e e e -8 Dehydronifedipine e e e -6 Digoxigenin e e e -7 Digoxin e e e -6 Diltiazem e e e -6 Doxycycline e e e -6 Enalaprilat e e e -6 Erythromycin-H 2 O e e e -4 Fluoxetine e e e -6 Gemfibrozil e e e -4 Ibuprofen e e e -4 Lincomycin e e e -5 Meprobamate e e e -5 Metformin e e e -5 Naproxen e e e -6 Norfloxacin e e e -6 Oxytetracycline e e e -5 Paroxetine metabolite e e e -6 Phenytoin e e e -6 Ranitidine e e e -6 Risperidone e e e -7 Sulfamethoxazole e e e -4 Sulfathiazole e e e -6 Tetracycline e e e -5 Trimethoprim e e e -5 Warfarin e e e -8 a Kolpin et al. (2002) & Benotti et al. (2009) & USGS ( )

33 E-05 Acetaminophen Atenolol Carbamazepine Cimetidine Diazepam Dehydronifedipine Digoxin Doxycycline Erythromycin-H2O Gemfibrozil Lincomycin Metformin Norfloxacin Paroxetine metabolite Ranitidine Sulfamethoxazole Tetracycline Warfarin E-08 ADI reference values ADI dw children values 3E-09 ADI fish children values 1E-10 ADI values (mg/kg-day) ADI dw +fish children values Figure 1. Comparison between the reference and calculated ADI values