Quantitative Structure Analysis Relationships for Predicting the Fates of

Future Contaminants in Indirect Potable Reuse Systems

by

Seung Lim

A Dissertation Presented in Partial Fulfillment

of the Requirements for the Degree

Doctor of Philosophy

Approved March 2011 by the

Graduate Supervisory Committee:

Peter Fox, Chair

Morteza Abbaszadegan

Rolf Halden

ARIZONA STATE UNIVERSITY

May 2011

i

ABSTRACT

The objective of this research was to predict the persistence of potential

future contaminants in indirect potable reuse systems. In order to accurately

estimate the fates of future contaminants in indirect potable reuse systems, results

describing persistence from EPI Suite were modified to include sorption and

oxidation. The target future contaminants studied were the approximately 2000

pharmaceuticals currently undergoing testing by United States Food and Drug

Administration (US FDA). Specific organic substances such as analgesics,

antibiotics, and pesticides were used to verify the predicted half-lives by

comparing with reported values in the literature. During sub-surface transport, an

important component of indirect potable reuse systems, the effects of sorption and

oxidation are important mechanisms. These mechanisms are not considered by the

quantitative structure activity relationship (QSAR) model predictions for half-

lives from EPI Suite. Modifying the predictions from EPI Suite to include the

effects of sorption and oxidation greatly improved the accuracy of predictions in

the sub-surface environment. During validation, the error was reduced by over

50% when the predictions were modified to include sorption and oxidation.

Molecular weight (MW) is an important criteria for estimating the persistence of

chemicals in the sub-surface environment. EPI Suite predicts that high MW

compounds are persistent since the QSAR model assumes steric hindrances will

prevent transformations. Therefore, results from EPI Suite can be very misleading

ii

for high MW compounds. Persistence was affected by the total number of halogen

atoms in chemicals more than the sum of N-heterocyclic aromatics in chemicals.

Most contaminants (over 90%) were non-persistent in the sub-surface

environment suggesting that the target future drugs do not pose a significant risk

to potable reuse systems. Another important finding is that the percentage of

compounds produced from the biotechnology industry is increasing rapidly and

should dominate the future production of pharmaceuticals. In turn,

pharmaceuticals should become less persistent in the future. An evaluation of

indirect potable reuse systems that use reverse osmosis (RO) for potential

rejection of the target contaminants was performed by statistical analysis. Most

target compounds (over 95%) can be removed by RO based on size rejection and

other removal mechanisms.

iii

TABLE OF CONTENTS

Page

LIST OF TABLES ..................................................................................................... vii

LIST OF FIGURES ..................................................................................................... x

CHAPTER

1 INTRODUCTION .................................................................................. 1

Backgroud ........................................................................................... 1

Research Objectives ............................................................................ 4

2 HYPOTHESES USED IN THIS STUDY ........................................... 5

3 LITERATURE REVIEW ..................................................................... 6

Hammett Equation Early QSAR ..................................................... 6

QSARs ................................................................................................ 8

QSAR Models ........................................................................ 9

EPI Suite .................................................................. 10

BIOWIN ...................................................... 12

AOPWIN ..................................................... 17

Shortcomings of EPI Suite ...................................... 18

Fugacity Models ...................................................... 22

Level I Model .............................................. 23

Level II Model ............................................. 25

Level III Model ........................................... 27

iv

Page

Pharmaceutical Drugs ....................................................................... 31

Development History of Pharmaceutical Drugs .................. 32

Usages ................................................................................... 33

Drug Metabolism in the Body ............................................. 36

Occurrence and the fates of Pharmaceutical Drugs in the

Environment ......................................................................... 38

FDAs Drug Review Process ............................................... 41

Key Parameters Affecting the Fates of Pharmaceutical Drugs and

Other Organic Chemicals ................................................................. 43

Biodegradation ..................................................................... 43

Biodegradation of Xenobiotic Compounds ............ 48

Biodegradation of Pharmaceutical Drugs in the

Environment ............................................................. 49

Sorption ................................................................................ 51

Cometabolism ....................................................................... 53

Water Solubility ................................................................... 59

Regulations of Persistent Organic Pollutants ................................... 61

4 METHODOLOGY .............................................................................. 64

Overview of Existing and Future Compounds Studied ................... 64

Modification of EPI Suite for Persistence Prediction ...................... 68

v

Page

5 VALIDATION OF THE MODIFIED HALF LIFE EQUATION .... 78

6 ESTIMATION OF THE FATES OF FUTURE CONTAMINANTS

IN THE SUB-SURFACE ENVIRONMENT .............................. 94

Introduction ....................................................................................... 94

Persistence of Target Pharmaceutical Drugs ................................... 97

Relationship between Molecular Weight and Half Life of Target

Drugs ............................................................................................... 103

Effects of Sortion on Half Life of Target Drugs ............................ 109

Comparison of Parameters Affecting the Persistence of Target

Pharmaceutical Drugs ..................................................................... 122

Effects of Halogenation in Target Drugs on Persistence ............... 126

Effects of Carbon Oxidation State on Half Life ............................ 134

7 EVALUATION FOR POTABLE REUSE BY REJECTION OF

FUTURE CONTAMINANTS THROUGH REVERSE

OSMOSIS MEMBRANES ........................................................ 138

8 OVERALL PERFORMANCE OF PREDICTIVE MODEL ........... 147

REFERENCES ...................................................................................................... 153

APPENDIX

I HALF-LIVES AND STRUCTURES OF TARGET DRUGS IN

ORAL CHEMO AGENTS .......................................................... 182

vi

Page

II HALF-LIVES AND STRUCTURES OF TARGET DRUGS IN

NMEs............................................................................................ 191

III HALF-LIVES AND STRUCTURES OF TARGET DRUGS IN

LATE PHASE III ......................................................................... 208

IV HALF-LIVES AND STRUCTURES OF TARGET DRUGS IN

MID TO LATE PHASE III .......................................................... 215

V HALF-LIVES AND STRUCTURES OF TARGET DRUGS IN

EARLY TO MID PHASE III ...................................................... 251

VI HALF-LIVES AND STRUCTURES OF TARGET DRUGS IN

LATE TO MID PHASE II ........................................................... 255

VII HALF-LIVES AND STRUCTURES OF TARGET DRUGS IN

EARLY PHASE II-LATE PHASE I ........................................... 315

VIII HALF-LIVES AND STRUCTURES OF TARGET DRUGS IN

PHASE I ....................................................................................... 323

vii

LIST OF TABLES

Table Page

1. Conditions Used in Fugacity Models .................................................. 22

2. Equations for Phase Z Values Used in Level I and II and the Bulk

Phase Used in Level III ..................................................................... 24

3. Relationship between BIOWIN3 Model and Converted Half Life .... 30

4. Reaction and Examples for Pharmaceutical Metabolism ................... 38

5. Occurrence and the Fates of Pharmaceutical Drugs in the

Environment ....................................................................................... 39

6. Several Functional Groups Attached to Aliphatic Carbon or

Aromatic Carbon Affecting Biodegradation ....................................... 47

7. Correlations between log Koc and log Kow for Several Chemical

Groups ................................................................................................ 53

8. Correlations between log Water Solubility and log Kow for Several

Chemical Groups ............................................................................... 60

9. Regulatory Programs and Criteria for POPs ....................................... 63

10. Average Physicochemical Properties for Each Category ................... 67

11. Numbers of Drugs not Suitable for Estimating the Fates in Each

Category ............................................................................................. 67

12. Percentages of Drugs not Suitable Estimating the Fates in Each

Category ............................................................................................. 68

viii

Table Page

13. Standard Coefficient Beta and Significance of Each Independent

Variable Analyzing with log Koc...................................................... 80

14. Standard Coefficient Beta and Significance of Each Independent

Variable Analyzing with log Kow .................................................... 80

15. Validation of the tso Using Several Organic Substances ..................... 83

16. Residual Sum of the Squares (RSS) for the Correlations in Each

Half Life ............................................................................................ 91

17. Persistent Target Pharmaceutical Drugs in Each Category (tso > 70

days) ................................................................................................... 99

18. Sub-persistent Target Pharmaceutical Drugs in Each Category

(50 days < tso 70 days) .................................................................. 101

19. Biodegradation Reaction Rate (k1), Oxidation Rate Constant (k3),

and Difference between k1 and k3 for Each Category ....................... 108

20. Effects of Total Numbers of Halogenated Aromatics on Persistence

of Target Drugs ................................................................................. 126

21. The Amount of Halogenated Aromatics in Each Category ............... 127

22. The Amount of Halogenated Aliphatics in Each Category ............... 128

23. Comparison of Persistence between Fluoromethane and Ethane ...... 129

24. Average Carbon Oxidation States of Non-persistent, Sub-persistent,

and Persistent Target Pharmaceutical Drugs .................................. 137

ix

Table Page

25. Average Physicochemical Properties of Drugs Less Than

MW 150 g/mol in Each Category .................................................... 141

26. Standard Coefficient Beta and Significance of Each Independent

Variable for Potable Reuse by Potential Rejection of Future

Contaminants through Reverse Osmosis Membranes .................... 143

27. Numbers of Drugs in Each Category which Can Be Completely

Removed by RO Membranes .......................................................... 145

x

LIST OF FIGURES

Figure Page

1. Reactive Intermediates Generated from P450 Cycle .......................... 58

2. Schematic Diagram of a Soil Aquifer Treatment ................................ 69

3. Plots of Half-Lives Predicted by BIOWIN versus the Observed

Half-Lives ............................................................................................ 90

3(a). Plot of the Half-Lives Predicted from BIOWIN (t1/2) versus the

Observed Half-Lives .......................................................................... 89

3(b). Plot of the Half-Lives Modified for Oxidation (to) versus the

Observed Half-Lives .......................................................................... 89

3(c). Plot of the Half-Lives Modified for Sorption (ts) versus the

Observed Half-Lives .......................................................................... 90

3(d). Plot of the Half-Lives including both the Effects of Oxidation and

Sorption (tso) versus the Observed Half-Lives .................................. 90

4. Variation of the Ratio of Observed Half Life to tso as a Function of

log Koc .............................................................................................. 92

5. Effects of Chlorine Atom on tso .......................................................... 93

6. Number of Chlorine Atoms of Each Organic Substance used for

Validation ........................................................................................... 93

7. Percentage of Persistent Drug in Each Category .............................. 103

8. Effects of Molecular Weight on Half-Lives of Target Drugs .......... 107

xi

Figure Page

8(a). Predicted Half Life from BIOWIN .................................................... 107

8(b). Ratio of tso to Predicted Half Life (Circle: Ratio of tso to Predicted

Half Life, Square: Sigmoidal Curve) ............................................... 107

9. Effect of Hydrophobicity on Half Life in the Sub-Surface

Environment (Circle: Ratio of tso to Predicted Half Life, Square:

Sigmoidal Curve) .............................................................................. 111

10. Effect of log Koc or MW on log Water Solubility of Each Target

Drug .................................................................................................. 113

10(a). Effect of log Koc on log Water Solubility ......................................... 113

10(b). Effect of MW on log Water Solubility ............................................. 113

11. Correlation between log Koc and Molecular Weight ...................... 115

12. Effects of Sorption and/or Oxidation on Half Life .......................... 119

12(a). Half Life (8.7 days)............................................................................ 117

12(b). Half Life (15 days) ............................................................................ 117

12(c). Half Life (37.5 days) ......................................................................... 118

12(d). Half Life (60 days) ............................................................................ 118

12(e). Half Life (180 days) .......................................................................... 119

13. Relationships between Atomic Ratio and Koc .................................. 122

13(a). Relationship between log H/C and log Koc ...................................... 121

13(b). Relationship between log O/C and log Koc ..................................... 121

xii

Figure Page

13(c). Relationship between log H/O and log Koc ..................................... 122

14. Comparison of Persistence Parameters between a Normalized

Total Number of Halogen Atoms in Chemicals and a Normalized

Sum of N-Heterocyclic Aromatics in Chemicals ............................ 125

14(a). Non-persistent Target Drugs (tso 50 days) ..................................... 125

14(b). Persistent Target Drugs (tso > 70 days) ............................................. 125

15. Effects of Halogenated Atoms on Persistence in Each Category .... 133

15(a). Effect of Total Numbers of Halogenated Aromatics in the

Potential Future Contaminants on Persistence (Resulted from

Half Life Prediction from BIOWIN) .............................................. 132

15(b). Effect of Total Numbers of Halogenated Aromatics in the

Potential Future Contaminants on Persistence (Resulted from

Modified Half Life Prediction, tso) .................................................. 132

15(c). Effect of Total Numbers of Halogenated Aliphatics in the

Potential Future Contaminants on Persistence (Resulted from

Half Life Prediction from BIOWIN) .............................................. 133

15(d). Effect of Total Numbers of Halogenated Aliphatiics in the

Potential Future Contaminants on Persistence (Resulted from

Modified Half Life Prediction, tso) .................................................. 133

xiii

Figure Page

16. Effects of Carbon Oxidation State on Half Life ................................. 135

16(a). Predicted Half Life from BIOWIN ................................................... 135

16(b). Ratio of tso to Predicted Half Life ..................................................... 135

17. Percentages of Drugs of Each Category which Can Be Completely

Removed by RO Membranes .......................................................... 146

1

Chapter 1

INTRODUCTION

Background

Many natural and synthetic toxic substances have been detected and

studied in the air, aquatic, soil, and sediment environments since the publication

of Silent Spring by Carson (1962). Researchers and organizations such as the

Organization for Economic Cooperation management and Development (OECD),

the United Nations Environment Programme (UNEP), the United States

Environmental Protection Agency (US EPA), and the European Union (EU) have

developed estimation tools for Persistence (P), Bioaccumulation (B), and Toxicity

(T) of organic chemicals in the environment (OECD, 2009; Schowanek and Webb,

2002; US EPA, 2009; Wegmann et al., 2009). Kier and Hall (1985) reported that

if the structural properties of a chemical are similar to those of another, most

physicochemical and biochemical properties of it can be inferred from those of

the other because the properties of each chemical are closely related to the steric

consequences of the structure. For many decades, quantitative structure activity

relationships (QSARs) were used to estimate the fate and transport of various

toxic contaminants and common organic substances, but the development of

QSAR methodology and their applications for practical uses have not been

sufficiently discussed (Moudgal et al., 2008).

2

Since the 1980s, in spite of the continuous discharge of micropollutants

such as pharmaceuticals and personal care products (PPCPs), endocrine disrupting

compounds (EDCs), and pesticides/herbicides at low levels from sewage

treatment plants (STPs), few researchers have been interested in the

environmental risks for human health and ecosystems (Kmmerer, 2001b; Tixier

et al., 2003). Specifically, despite the sharp increment of drug usage and disposal

into the environment, little interesting was given to persistence (P),

bioaccumulation (B), and toxicity (T) regarding pharmaceutical drugs until the

1990s (Daughton and Ternes, 1999; Dietrich et al., 2002; Halling-Srensen et al.,

1998; Heberer, 2002b; Kmmerer, 2001a). Most reports were focused on

detecting each pharmaceutical drug in receiving river or sediment conditions

(Abuin et al., 2006; Anderson et al., 2004; Loraine and Pettigrove, 2006; Pedersen

et al., 2005; Ternes et al., 2001; Yoon et al., 2010; Zuehlke et al., 2004). In

addition, the occurrence of pharmaceuticals in the aquatic environment has been

reported in most countries (Kolpin et al., 2002; Loraine and Pettigrove, 2006;

Stumpf et al., 1999; Ternes, 1998; Yoon et al., 2010). Pharmaceutical drugs in the

aquatic environment have the potential to persist after percolation into an aquifer

(Heberer, 2002a), however, many compounds are transformed during sub-surface

transport as compared to surface water transport.

Recent studies on the fates of specific pharmaceutical compounds by soil

column studies, field studies, and computer simulations have been completed.

3

However, the test results have not been consistent due to differences in

methodology, hypotheses, and test conditions. In particular, specific information

about the soils and aquifer types in most field studies was not generally provided

(Cordy et al., 2004; Drewes et al., 2003; Snyder et al., 2004). In contrast to

column studies or computer simulations, it is not possible to control

physicochemical/biochemical parameters such as soil type, spatial variations of

soil type, variations in temperature/dissolved oxygen/pH/redox potential,

groundwater flow velocity, biomass, and the activity of biomass. Thus, it is

necessary that inferential analyses be used in field studies. However, most data

obtained from field studies were directly used to understand and estimate the fates

of target organic chemicals. It is time-consuming, laborious, and costly to obtain

accurate data from field studies. In addition, data from field studies cannot be

obtained with a complete set of parameters. This suggests that column studies or a

computer modeling can be a useful alternative. The results from field studies can

be used for correlating results obtained from column studies or verifying the

hypotheses used in predictive models.

4

Research Objectives

In this study, the four major research objectives are classified as follows:

Modification of existing half life relationships for estimating the fates of

future contaminants in the sub-surface environment.

In order to estimate the fates of future contaminants in the sub-surface

environment, the aqueous half life relationship from EPI Suite will be

modified to include the effects of sorption and co-metabolic oxidation.

Verification of the modified half life relationship using specific organic

compounds for which independent studies have been completed to

measure their persistence.

Estimating the fates of target contaminants in the sub-surface environment

and investigating the influencing factors on the persistence of each target

compound.

It will be shown which physicochemical/biochemical parameters have the

greatest impact on the persistence of each chemical.

Evaluation for indirect potable reuse by reverse osmosis (RO). It is

assumed that RO is the primary removal mechanism for future

contaminants.

The rejection of future contaminants by reverse osmosis membranes will

be evaluated.

5

Chapter 2

HYPOTHESES USED IN THIS STUDY

Existing QSAR models can be modified to predict the persistence of

organic chemicals in an indirect potable reuse system where sub-surface

transport occurs. Modifications would include factors such as sorption and

co-metabolism.

Existing QSAR models do not accurately predict the persistence and/or

biodegradation of high molecular weight compounds.

A modified QSAR model can be used to assess persistence trends in the

development of future emerging contaminants of concern, in particular the

approximately 2000 chemical entities that are currently undergoing the

United States Food and Drug Administration (US FDA) testing.

The removal of future emerging contaminants of concern may also be

evaluated for potable reuse systems that use reverse osmosis membranes

for removal (simply use molecular weight, log Kow, pKa, etc.).

6

Chapter 3

LITERATURE REVIEW

Hammett Equation Early QSAR

Hammett elucidated which factors contributed to biochemical and

biological reactions in 1935. He demonstrated the hypothesis that similar

changes in structure produce similar changes in reactivity by testing with

benzoic acids. In other words, the log pKa of meta- or para-substituted benzoic

acids was proportional to the log pKa of meta- or para-substituted phenylacetic

acids, respectively. Hammett defined the parameter as follows:

(1)

where,

= Hammett constant

kx = the ionization constant of meta- or para-substituted benzoic acid in water at

25oC

kH = the ionization constant of parent benzoic acid in water at 25oC

A positive implies an electron withdrawing effect of substituent from aromatic

rings. The Hammett equation can also be used to describe changes in reaction rate

or equilibrium constants by equation 2.

7

(2)

where,

kx = equilibrium constant or rate constant

kH = corresponding constant for the parent compound, unsubstituted compound

= sensitivity of a reaction to the electronic effect of the substituent X

However, the Hammett equation has some drawbacks (Hansch and Leo,

1995). The Hammett equation is actually best fit to aromatic compounds similar

to benzoic acids. One of the most difficult aspects is that the Hammett equation is

applied to structure-activity relationships (SARs) with both steric and

hydrophobic parameters. A Hammett constant () does not consider the geometry

of a target chemical (i.e., assuming a fixed plane the same as that of an aromatic

ring). As a matter of fact, the observed Hammett constants are different from the

calculated due to the conjugation or twisting of each functional group. For

example, a methyl group and chlorine atom are matched with each other, while

the twist occurs between aromatics and N(CH3)2 group (Hansch et al., 1973).

This steric limitation of the Hammett constant was modified by developing a

correlation between meta- and para- constants as presented in equation 3

(Hansch et al., 1991).

8

(3)

QSARs

SARs are can be complicated relationships between the physicochemical

properties and biochemical activity of a chemical. In other words, SARs can be a

function of both chemical structure (steric characteristics) and its associated

degradability in the environment. The degradation of many organic chemicals in

aqueous environments mainly results from biochemical reactions (e.g., enzymes

of microorganisms). In addition, organic chemicals can be removed by

physicochemical depletion processes such as UV oxidation, oxidation by free

radicals, sorption, and/or evaporation. Both Qualitative SARs and Quantitative

SARs are usually referred to as QSARs. Qualitative SARs result from non-

continuous data such as positive or negative results, whereas Quantitative SARs

are obtained from continuous data like reaction rate constants and half-lives. The

half life of a chemical can represent its persistence in the environment and half-

lives are commonly used to present the results of QSARs.

In order to obtain sufficient information on a specific organic chemical,

QSAR tools require some physicochemical data with regard to the chemical.

These data can be obtained from handbooks, free Internet sources, commercial

9

Internet sources, or constantly evolving databases. Some useful handbooks

including the Physical-chemical properties and environmental fate for organic

chemicals, Handbook on physical properties of organic chemicals, and CRC

Handbook of chemistry and physics are available. The Internet provides excellent

sources including databases such as PubChem; http://pubchem.ncbi.nlm.nih.gov/,

Chemical book; http://www.chemicalbook.com/ProductIndex_EN.aspx,

ChemSpider; http://www.chemspider.com/, Drugbank; www.drugbank.ca, QSAR;

www.qsar.org, and Chemweb; www.chemweb.com. In addition, there are several

commercial sites such as Chemical Abstract Service (www.cas.org), Prous

Science (www.prous.com), and Derwent (www.derwent.com).

QSAR Models

In general, modeling of a chemical is based on its molecular structure,

constitution, and charge distribution. It suggests that chemical properties are a

function of primarily steric characteristics and molecular constitutions. The

OECD proposed some QSAR modeling tools such as the OECD Pov & LRTP

Screening Tool and the QSAR Application Toolbox. The QSAR Application

Toolbox was released in 2008. This model is used to estimate the environmental

fate and toxicity of organic chemicals using metrics of overall persistence and

long-range transport potential (OECD, 2009). There are several motivations to

http://pubchem.ncbi.nlm.nih.gov/http://www.chemicalbook.com/ProductIndex_EN.aspxhttp://www.chemspider.com/http://www.drugbank.ca/http://www.qsar.org/http://www.chemweb.com/http://www.cas.org/http://www.prous.com/http://www.derwent.com/

10

estimate the persistence of organic chemicals based on computer analysis using

QSAR (Cronin and Livingstone, 2004).

Using computer models, it is possible to obtain data to some extent

without chemical testing or even without synthesizing chemicals.

The concern about animal tests (in vitro) for some decades can be replaced

by testing in silico.

Governments in the EU and North America have enacted legislations

mandating the prediction of toxicity of organic chemicals using computer

analysis. For example, the US EPA uses QSARs as a screening tool for

pre-manufactory notifications (PMNs) of new chemicals, whether toxic

substances are suspected to exist or not.

Computer analysis can save both cost and time to estimate the persistence

and toxicities of organic chemicals as compared to conventional

alternatives.

QSAR models are helpful for understanding the physicochemical and

biochemical characteristics of harmful compounds in the environment.

EPI Suite

EPI Suite was developed by the US EPA and is based on the BIOWIN and

AOPWIN modules to estimate the fates of organic chemicals in the environment.

11

In order to use EPI Suite, the chemicals Simplified Molecular Input Line Entry

System (SMILES) notation is required as input (Weininger, 1988). EPI Suite (ver.

4.0) consists of 13 discrete models as listed below:

AOPWIN - estimates atmospheric oxidation rates

BCFBAF - estimates bioconcentration factor (BCF) and biotransformation

rate (kM)

BioHCwin - estimates biodegradation of hydrocarbons

BIOWIN - estimates biodegradation probability

ECOSAR - estimates aquatic toxicity (LD50, LC50)

HENRYWIN - estimates Henrys law constant

HYDROWIN - estimates aqueous hydrolysis rates (acid-, base-catalyzed)

KOAWIN - estimates octanol-air partition coefficient

KOCWIN - estimates soil sorption coefficient (Koc)

KOWWIN - estimates octanol-water partition coefficient

MPBPVP - estimates melting point, boiling point, and vapor pressure (also

referred to as MPBPWIN)

WSKOWWIN - estimates water solubility (from log Kow)

WATERNT - estimates water solubility (using atom-fragment

methodology)

12

It is not necessary that the 13 models be used to estimate the persistence of a

chemical. Thus, the user can choose the models used to provide results from 13

discrete EPI Suite models.

BIOWIN

The biodegradation probability program for Microsoft Windows

(BIOWIN) models have been developed by the Syracuse Research Corp. (SRC)

on behalf of the US EPA since the 1980s (Boethling et al., 1994, 2003, 2004;

Boethling and Sabijic, 1989; Howard et al., 1986, 1987, 1992; Meylan et al.,

2007; Tunkel et al., 2000). Most BIOWIN models estimate the probability of

rapid aerobic biodegradability. BIOWIN7 is the exception as it estimates the

probability of rapid anaerobic biodegradability in the presence of heterogeneous

microorganisms. Although BIOWIN consists of seven models, EPI Suite uses

BIOWIN3 to estimate the fate of a chemical by default. The description of each

BIOWIN model is as follows:

BIOWIN1: linear probability model

BIOWIN2: nonlinear probability model

13

The BIOWIN1 and BIOWIN2 models were reviewed in an article by

Howard et al. (1992). The linear model (BIOWIN1) and the non-linear

model (BIOWIN2) were developed using 264 chemicals.

BIOWIN3: expert survey ultimate biodegradation model

BIOWIN4: expert survey primary biodegradation model

BIOWIN3 and BIOWIN4 models were introduced in an article by

Boethling et al. (1994). These models are survey models because an

expert panel was asked for their response to predicted rates for primary

degradation (loss of parent chemical identity) and ultimate degradation

(conversion to CO2 and H2O) under aerobic conditions.

BIOWIN5: MITI linear model

BIOWIN6: MITI nonlinear model

The linear model (BIOWIN5) and non-linear model (BIOWIN6) for

assessing organic compounds is based on the Japanese Ministry of

International Trade and Industry (MITI) ready biodegradation test

BIOWIN7: anaerobic biodegradation model

BIOWIN7 was developed using 169 chemicals and consists of a linear

model and a non-linear model to assess the probability of rapid anaerobic

biodegradation

14

There exist so many screening test protocols including the MITI test,

Association Franaise de Normalisation (AFNOR), and the OECD screening test

(e.g., BOD tests, activated sludge die-away tests, CO2 evolution tests, and etc.).

Thus, Howard et al. (1987) reviewed previous structure/biodegradation test data,

collected and evaluated available biodegradation data, and designed and

documented the procedures. Using available biodegradation results, Howard et al.

(1987) assumed that as the number of consistent test results or test results for

which apparent inconsistencies are resolvable increases, the greater is the

likelihood that the indication of biodegradability is a property of the chemical

rather than of the test system. Evaluation of biodegradability was done with

screening tests, biological treatment simulations, grab sample tests, and field

studies. The test results were evaluated using the BIODEG model. In order to

obtain unacclimated as well as acclimated results, BIODEG used a test time

period of 28 days which is the same as used for the OECD-recommended

screening tests.

The US EPA reviews thousands of pre-manufacture notices (PMNs;

substances not yet in commerce) for potential ecological and human health effects.

Under the Toxic Substances Control Act (TSCA), the US EPA regulates

chemicals except for pesticides, drugs, and food additives (Boethling et al., 2003).

Boethling and Sabijic (1989) surveyed 50 PMN chemicals by an expert panel.

Experts were asked to evaluate predicted removal rates in typical wastewater

15

treatment systems (REM) and aerobic ultimate degradation rates in receiving

waters (AERUD) for 50 PMN chemicals. Also, AERUD distinguished

compounds with high biodegradability rates from compounds with low

biodegradability rates using an expert model. Howard et al. (1992) developed a

linear model and a non-linear model using 264 chemicals. Under aerobic

conditions, the accuracy of the results for rapidly degraded compounds was

greater than that for the results of slowly degraded compounds because the

majority of selected target chemicals were rapidly biodegradable under aerobic

conditions. Boethling et al. (1994) developed four models. A linear BIODEG

model and a non-linear BIODEG model were developed using 295 chemicals

where 186 of the chemicals were designated rapidly biodegradable and 109

chemicals were designated does not rapidly biodegradable. These models of

Howard and Boethling were on the basis for BIOWIN1 and BIOWIN2.

BIOWIN3 and BIOWIN4 (two survey models) were developed for primary

degradation and ultimate degradation under aerobic conditions using 200

chemicals. A primary model for the loss of parent chemical identity and an

ultimate model for the conversion of CO2 and H2O were evaluated by 17 experts.

The accuracy of each BIODEG model was approximately 90%, while that of

survey models was approximately 83%. Tunkel et al. (2000) compared the test

results from BIOWIN (ver. 3.63) with those resulted from the MITI-I test using

884 discrete organic chemicals. In the BIOWIN model evaluation, two thirds of

16

the 884 chemicals were used as an evaluation data set, while the remaining one

third was used for validation. The compounds used for validation were randomly

selected. In the MITI-I test, the OECD pass criterion (inoculum = 30 mg sludge

solids/L, test period = 28 days, ThOD 60% for pass) was used to discriminate a

pass/fail for biodegradation. When using MITI fragments (BIOWIN: training set

= 884 chemicals, no validation set; MITI-I test = training set: 589 chemicals,

validation set = 295 chemicals), the accuracies of both a linear model and a non-

linear model in MITI-I tests were greater than those of BIOWIN. When, however,

BIOWIN used BIOWIN fragments (training set = 589 chemicals, validation set =

295 chemicals), the accuracies of both a linear model and a non-linear model in

MITI-I tests were comparable to those of BIOWIN.

Boethling et al. (2003) compared the accuracy of different BIOWIN

models (from BIOWIN1 to BIOWIN6) using a total of 944 PMN chemicals. The

US EPA used 305 PMN chemicals, 439 chemicals from the MITI databases, and

compared 200 PMN chemicals in the expert survey on which the BIOWIN3 and

other BIOWIN models were evaluated. The greatest accuracy among BIOWIN

models was demonstrated by BIOWIN3 (87%), whereas BIOWIN1 and

BIOWIN2 models were deemed not suitable for estimating PMNs. Meylan et al.

(2007) developed the BIOWIN7 model using data from serum bottle tests

(incubation period: at least 56 days) using 169 chemicals in the training set and 58

chemicals in two validation sets. The accuracy of the training set was

17

approximately 90% and those of the validation sets were 77% and 91%,

respectively.

AOPWIN

The Atmospheric Oxidation Program for Microsoft Windows (AOPWIN)

estimates the rate constant between photochemically produced hydroxyl radicals

and organic chemicals in air under environmental conditions. AOPWIN also

estimates the rate constant between ozone and olefinic compounds. The estimated

constants are used to calculate the half-lives of organic compounds in the

atmosphere. Since the 1980s, several methods have been developed to estimate

hydroxyl radical concentrations in the atmosphere (e.g., utilizing molecular

properties of chemicals or ionization energy) (Gsten et al., 1984; Grosjean and

Williams, 1992). However, most of these methods were not based on database

molecular properties. Atkinson (1986, 1987) developed estimation methods using

SARs, however the estimations for reactions with fluoresters, ethers, haloalkanes,

and haloalkanes containing CF3 groups were not accurate. Kwok and Atkinson

(1995) updated the AOPWIN model and reported that the accuracy of the

hydroxyl radical reaction rate constant was approximately 90% at 25oC. The

hydroxyl radical reaction rate constant with organic compounds in AOPWIN

model can be calculated as follows:

18

(4)

Shortcomings of EPI Suite

To ideally estimate the persistence of a chemical in the aqueous

environment, a half life of the target chemical under aqueous conditions (water or

sediment) is required. However, it takes a long time to measure half-lives of

chemicals in the aqueous environments because of long-transport times and the

complexity of the ecosystem. Thus, extrapolation from databases based on

aqueous conditions is usually used (Pavan and Worth, 2008). In order to obtain

data under aqueous conditions, several test methods such as MITI, AFNOR, and

the OECD screening test guidelines for ready biodegradation have been

developed and extensively used. Most test methods provide pass/fail results for

each test criteria. In general, most test methods commonly used can be classified

as follows:

Ready biodegradability

A ready biodegradability test is used for unfavorable (stringent)

conditions. If a chemical passes the test criteria, the chemical is readily

19

biodegradable. In contrast, if the chemical fails, the chemical does not

necessarily follow that it will not degrade in the environment.

Inherent biodegradability

An inherent biodegradability test is used for favorable (non-stringent)

conditions. If a chemical fails the test criteria, the chemical is non-readily

biodegradable. In contrast, if the chemical passes, the chemical does not

necessarily degrade in the environment.

The screening tests are highly affected by several test parameters such as pH,

dissolved oxygen, redox potential, temperature, medium concentration, inoculum,

enzymes of microorganisms, and nutrients. Cornelissen and Sijm (1996) reported

that biodegradation of a chemical is severely affected by environmental

conditions such as temperature, oxygen concentration, redox potential, pH,

salinity, and nutrients. Thus, the test results can be different for each test method

(Howard and Banerjee, 1984). Therefore, EPI Suite might not be used to compare

the biodegradability results of BIOWIN which is based on the OECD screening

test guidelines with results from other screening test methodologies.

EPI Suite qualitatively separates non-persistent chemicals from persistent

chemicals. The model results had large discrepancies when comparing the results

of EPI Suite with results published in references (handbooks). Specifically, there

was a difference of 2-3 orders of magnitude in the half-lives of persistent

20

chemicals (half-lives greater than 40 days) in water, soil, and sediment (Gouin et

al., 2004). Boethling et al. (2003) compared the accuracy between BIOWIN

models using PMNs, with molecular weights < 800 g/mol, which is the upper

limitation of SMILES notation input into EPI Suite. However, organic chemicals

with a molecular weight up to 2000 g/mol can be put into the most recent version

of EPI Suite (ver. 4.0). The previous articles regarding PMNs analyzed by EPI

Suite were limited to compounds with a molecular weight less than 800 g/mol.

Although the BIOWIN3 model provides the best fit for estimating the

biodegradability of organic chemicals among the BIOWIN models, the accuracy

was 87%. In addition, both BIOWIN3 and BIOWIN4 (survey models) were never

validated by experiments (Boethling et al., 2003). In spite of the lack of

confidence for estimating the fates of PMN chemicals, BIOWIN3 is still used to

predict the persistence of organic chemicals because there is no better alternative

in EPI Suite. The BIOWIN3 model was based on the OECD pass criterion (test

period = 28 days, ThOD 60% for pass) so that the accuracy of each BIOWIN

model is focused on estimating the ready biodegradation of an organic chemical

under aerobic condition. Nevertheless, the greatest shortcoming of each BIOWIN

model is the accuracy for biodegradation fast; BF chemicals which is lower

than that for biodegradation slow; BS chemicals in all BIOWIN models

(Boethling et al., 2003, 2004; Meylan et al., 2007; Rorije et al., 1999). This is

because BIOWIN models consider only degradation of the parent chemicals and

21

do not consider metabolites of them (Pavan and Worth, 2008), while the OECD

pass criterion requires 60% removal of ThOD which implies mineralization.

In addition, EPI Suite is set at 25oC to estimate the fate of an organic

chemical because models such as BIOWIN and AOPWIN were developed for this

condition. However, the real air, aquatic, soil, and sediment environments have

temporal and spatial variations in temperature that cannot be included in a QSAR.

Loftin et al. (2008) reported that pH and temperature significantly affected the

biodegradation of selected antibiotics. EPI Suite cannot also estimate the PBT of

inorganic chemicals and organometallic chemicals. Only the PBT of organic

chemicals be estimated in EPI Suite, and the predicted result should not be used in

the place of experimental data. Green and Bergman (2005) stated that although

persistence models have been improved they still lack accurate compartmental

degradation rates in their models.

Unlike bioaccumulation, the persistence of a chemical cannot be directly

measured and deduced from mass balances (Mackay and Webster, 2006). Thus,

EPI Suite is based on MacKays fugacity model (Level III fugacity multimedia

model). The Level III fugacity multimedia model is mass balance model between

air, water, soil, and sediment, but the migration of a contaminant to groundwater

is not considered. Thus, micropollutant contamination in aquifers or groundwater

cannot be accurately estimated with EPI Suite. The fugacity models are

introduced in the next section.

22

Fugacity Models

Fugacity is defined as the chemical activity of a gas, and expressed as an

escaping tendency from a compartment. The assumed temperature is 25oC for all

fugacity models and the fugacity is linearly related to the concentration at low

concentrations (Mackay et al., 1992). Conditions used in the fugacity models in

EPI Suite are shown in Table 1.

Table 1. Conditions Used in Fugacity Models

Parameter Value

Atmospheric height (m) 1000

Water surface area (km2) 10000

Water depth (m) 20

Water volume (m3) 210

11

Soil depth (cm) 10

Soil organic carbon (%) 2

Soil volume (m3) 910

9

Sediment depth (cm) 1

Sediment organic carbon (%) 4

Sediment volume (m3) 10

8

23

Level I Model

The Level I model shows the equilibrium partitioning for a given amount

(an arbitrary amount) of a chemical between media (Mackay et al., 1992). The

fugacity (f) can be shown as follows:

(5)

where:

f: fugacity (Pa)

M: total amount of a chemical (mol)

Vi: the medium volume (m3)

Zi: the corresponding fugacity capacity for a chemical in that medium

(mol/m3Pa)

The parameter and fugacity capacity are shown in Table 2.

24

Table 2. Equations for Phase Z Values Used in Level I and II and the Bulk Phase

Used in Level III

Parameter Fugacity capacity

Air

Water

Soil

Sediment

Suspended sediment

Fish

Aerosol

where,

R: gas constant (8.31 J/molK)

T: absolute temperature (K)

CS: water solubility (mol/m

3)

PS: vapor pressure (Pa)

H: Henrys law constant (Pam3/mol)

PS

L: liquid vapor pressure (Pa)

Kow: octanol-water partitioning coefficient

Koc: organic-carbon partitioning coefficient (= 0.41Kow)

i: density of phase i (kg/m3)

i: mass fraction organic-carbon of phase i (g/g)

L: liquid content for fish

25

Level II Model

In the Level II model, a steady state, continuous input and output, and

equilibrium conditions for different media are considered (Mackay et al., 1992).

As the Level II model is a steady-state continuous equilibrium model, the rates of

losses by reaction and advection for each medium can be included for estimation.

The advection is given for select flow rates in each medium. The reaction rate (k)

in each media is usually expressed as a first-order reaction rate constant. In

addition, the reactions can occur in a one-compartment system or multi-

compartment systems.

One-Compartment System

If a half life of an organic chemical is subjected to a first-order reaction,

the rate can be expressed as follows (Mackay and Webster, 2006):

(6)

where,

Ct: the concentration of a chemical at time t [M/L3]

Co: the initial concentration of a chemical [M/L3]

k: the first-order reaction rate constant [1/T]

1/k: average residence time of a chemical [T]

26

If mass is introduced in a constant volume and the first-order reaction is

considered, the emission rate (E) can be expressed as follows:

(7)

where,

E: emission rate of a chemical [M/T]

V: volume [L3]

C: the concentration (the first order) [M/L3]

m: mass of a chemical [M]

The average residence time suggests the preferred metric of persistence (P) of a

chemical. The total mass of a chemical (m) is equal to PE, where P is proportional

to mass in the environment.

Multi-Compartment Systems

In multi-compartment systems, the emission rate of a chemical is as

follows:

(8)

27

where,

mA and mS: mass in air and in soil, respectively [M]

kA and kS: reaction rate constant in air and in soil, respectively [1/T]

Thus, persistence (P) can be expressed by equation 9 or 10.

(9)

(10)

where,

Fi: the fraction of the mass of a chemical in each compartment

Level III Model

Currently, Level I and Level II models are no longer used extensively

since the Level III model was introduced. The Level III model can be defined as

set up for each compartment including expressions for rates of intermedia

transport (Mackay and Webster, 2006). The Level III model is not an

equilibrium mass balance model and it is a simple algebraic model consisted of

seven intermedia transport values. The Level III model was updated to show

28

intermedia transport by maintaining values of the existing fugacity models (Level

I and Level II models) and increasing the total volume of the phases.

EPI Suite is based on a fugacity model (exactly Level III multimedia mass

balance model of Mackey; Mackey et al., 1992). The Level III fugacity model is

not an equilibrium and steady-state multimedia fate model. It provides

information about environmental partitioning and intermedia transport. The

fugacity models estimate the partitioning of a chemical in the environment. Thus,

in order to use the Level III multimedia mass balance model in EPI Suite,

physicochemical properties and the half life information of a chemical in each

medium (air, water, soil, and sediment) must be estimated. In addition, the

emission rate of each medium and the advection times applied in air, water, and

sediment are required.

Physicochemical Properties of a Chemical

The required physicochemical properties are as follows:

Henrys constant

Vapor pressure

Melting point

log Kow

Koc

29

Molecular weight

Half Life

EPI Suite uses BIOWIN and AOPWIN models to estimate the half life of

a chemical by default.

air

An air half life is calculated directly from gas-phase hydroxyl radical and ozone

reaction rate constants. The half life is calculated from the rate constant an

average atmospheric concentration of these oxidants based on a 24 hour day

(Atkinson and Carter, 1984; Prinn et al., 1992).

(11)

(12)

water, soil, and sediment

An aqueous half life is estimated by the BIOWIN3 model (Boethling et al., 1994).

As the BIOWIN3 model is just a screening model, it is necessary to convert the

results from a BIOWIN model to half-lives. This conversion in EPI Suite follows

the guidelines of the PBT profiler (www.PBTprofiler.net). The relationship

http://www.pbtprofiler.net/

30

between the BIOWIN3 model and the converted half life is shown in Table 3.

Even though there are many recalcitrant chemicals half-lives of which are greater

than 180 days, the maximum half life BIOWIN3 can present is 180 days. In

addition, soil and sediment half-lives are calculated using a conversion factor

developed by the US EPAs pollution prevention (P2) framework (US EPA,

2005). Under the P2 framework, it is assumed that soil is under aerobic conditions,

and a soil half life is two times a water half life. Sediments are assumed to be

under anaerobic conditions, and a sediment half life is nine times a water half life.

Table 3. Relationship between BIOWIN3 Model and Converted Half Life

BIOWIN3 output Converted half life (day)

Hours 0.17

Hours to days 1.25

Days 2.33

Days to weeks 8.67

Weeks 15

Weeks to months 37.5

Months 60

Recalcitrant 180

31

Emission Rate and Advection Time

In EPI Suite, the default environmental emission rates are 1000 kg/hr to

air, water, and soil. The sediment emission rate has a value of zero. The advection

times apply to air, water, and sediment. The advection lifetimes of a chemical in

the air, water, and sediment compartments are set to the default values of 100,

1000, and 50000 hrs, respectively. These lifetimes are used to determine the

advective flow rate (m3/hr) calculated by dividing the volume of the compartment

by the advection time.

Pharmaceutical Drugs

Daughton and Ternes (1999) defined that Pharmaceutical drugs are

chemicals used for diagnosis, treatment (cure/mitigation), alteration, or prevention

of disease, health condition, or structure/function of the human body. In addition,

this definition is extended to veterinary pharmaceuticals and illicit (recreational)

drugs. Drugs used to treat humans and vertebrates are bioactive chemicals as they

act as receptors, enzymes, and so on. Jrgensen and Halling-Srensen (2000)

assumed that when these compounds occurred at low concentrations, they would

harm the environment, irrespective of their bioactivity. The persistence of each

drug, however, is determined by steric characteristics, chemical composition, and

32

other parameters including biodegradability, cometabolism, sorption, or chemical

oxidation which result from the interactions between the target drugs and the

environment.

Development History of Pharmaceutical Drugs

The development history of pharmaceutical drugs is well reviewed in an

article by Daemmrich and Bowden (2005). The origin of the modern

pharmaceutical industry traces back to around the 1880s when apothecaries

moved into wholesale production of drugs and dyes such as morphine, quinine,

and strychnine. According to the increment of the requests for the development of

dyes, immune antibodies, and other physiologically active agents, pharmaceutical

firms started cooperating with academic labs in the 1930s. In the middle of the

20th

century, many new vaccines, synthetic vitamins, antibiotics, and synthetic

hormones were developed. Deaths in infancy were sharply decreased and illnesses

such as tuberculosis, diphtheria, and pneumonia could be treated for the first time.

During the World Wars, some drugs such as antimalarials, penicillin were also

developed to support troops. The market of pharmaceutical drugs is focused on

treating cancers, acting on the central nervous system, and treating viral and

retroviral infections including therapies for HIV/AIDS. Most researchers in

biology, genetics, and genomics pay attention to developing new drugs and most

related investigations are also involved in new drug development.

33

Usages

The pharmaceutical drugs we usually use and dispose of can be classified

into seven categories (Monteiro and Boxall, 2010).

Analgesics and anti-inflammatory

Analgesics are drugs used to relieve pain. Analgesics such as codeine and

morphine include non-steroidal anti-inflammatory drugs (NSAIDs) and

acetaminophen. NSAIDs relieve pain and suppress inflammation, but

acetaminophen cannot treat inflammation. NSAIDs are acidic and have

varying degrees of hydrophobicity. Acetaminophen, acetylsalicylic acid

(aspirin), ibuprofen, and naproxen are the most commonly used analgesics.

Antibiotics

Antibiotics target specific microorganisms such as pathogenic bacteria.

These drugs are toxic to the microorganisms against which they are active.

The most common antibiotics are penicillin, sulfamethoxazole, and

erythromycin.

In addition, macrolides are widely used to prevent bacteria from

multiplying. Macrolides have no capacity for killing bacteria, but they

prevent their reproduction thereby preventing disease.

34

Beta-Blockers

Beta-Blockers are drugs that act on blood vessels. These drugs are used

for reducing the speed and force of heart contractions. There are two types

of Beta-Blockers. One type affects the 1 receptors which are located in

the heart muscle, while the other affects 2 receptors which are in blood

vessels. Beta-Blockers vary in hydrophobicity.

Hormones and steroids

Most hormones contain proteins, peptides, steroids, and derivatives of the

amino acid tyrosine. Protein and peptide hormones are produced by the

thyroid gland, the pancreas, the parathyroids, and the pituitary gland.

Steroid hormones are generated by ovaries or testis. Hormones and

synthetic hormones are hydrophobic compounds.

Lipid regulators

Lipid regulators are used to lower levels of triglycerides and low-density

lipoproteins (LDL) and increase the levels of high-density lipoproteins

(HDL) in the blood. There are three kinds of regulators. Fibrates such as

bezafibrozil and gemfibrozil are used to treat high level of triglycerides in

the blood. Statins such as atorvastatin and simvastatin decreases levels of

LDL. Niacin reduces the levels of triglycerides and increases the levels of

HDL. Lipid regulators tend to be hydrophobic organic compounds.

35

Selective serotonin reuptake inhibitors (SSRIs)

The selective serotonin reuptake inhibitors (SSRIs) treat clinical

depression by blocking the reuptake of the neurotransmitter serotonin by

the nerves in the brain. SSRIs are hydrophobic. Fluoxetine is one of the

most widely used SSRIs.

Other pharmaceuticals

- Antiepileptics

Antiepileptics such as carbamazepine are used for treatment of

epilepsy (periodic or unpredictable seizures).

- 2-Sympathomimetics

2-Sympathomimetics are used to treat asthma. They stimulate 1-

receptors.

- Iodinated X-ray contrast media

These drugs are used for intensifying the contrast of structure in the

human body for imaging and they are primarily used at imaging

facilities. Iopromide, iomeprol, diatrizoate, and iopadimol are all used

for this purpose. These compounds are iodated organic compounds

with unique properties.

36

Drug Metabolism in the Body

The main mechanism of pharmaceutical drug uptake in the body is simple

diffusion (Woolf, 1999). Simple diffusion is affected by physicochemical

properties of drugs such as molecular weight, shape, ionization of drug, and

solubility. Hydrophobic drugs are easily and rapidly absorbed in the body and

excreted after metabolism. Some drugs are excreted as non-metabolized, while

others are metabolized in the liver by the P450 microsomal oxidase system. The

greater the extent of drug metabolism in the body, the more hydrophilic the

metabolites become. Table 4 shows a sample reaction and some examples for

pharmaceutical metabolism (Monteiro and Boxall, 2010). The metabolism of

pharmaceutical drugs consists of two successive pathways as follows (Daughton

and Ternes, 1999; Halling-Srensen et al., 1998; Monteiro and Boxall, 2010):

Phase I

- Oxidative reactions, hydrolysis reactions

In the Phase I, monooxygenases, reductases, or hydrolases are used to add

reactive functional groups to the molecules. The products may be more

toxic than the parent drugs in this Phase.

Phase II

- Conjugation reactions

37

During Phase II reactions, covalent conjugation (glucuronidation) is used

to make the molecule hydrophilic and more excretable. Drugs can become

inactive by conjugation reactions, but metabolites can be reactivated via

enzymatic or chemical hydrolysis in the environment (Baronti et al., 2000;

Daughton and Ternes, 1999; Desbrow et al., 1998).

38

Table 4. Reaction and Examples for Pharmaceutical Metabolism

Occurrence and the Fates of Pharmaceutical Drugs in the Environment

Thousands of different drugs have been developed and registered around

the world (Daughton and Ternes, 1999; Halling-Srensen et al., 1998; Monteiro

and Boxall, 2010). The disposal of unused medication and the excretion via

human body result in the release of pharmaceutical drugs into the aquatic

39

environment and these drugs are removed in STPs to a large extent (Heberer,

2002a; Ternes, 1998). In addition, the increased flow rate caused by rainfall

events can affect the removal performance of pharmaceutical drugs in STPs as it

reduces the activity of microorganisms and sorption by activated sludge in an

aeration tank due to the decreased concentrations (Ternes, 1998). Many

investigators reported the occurrence and fate of pharmaceutical drugs in the

aquatic, soil, and sediment environments. A summary of the occurrence and the

fates of pharmaceutical drugs in the environment is shown in Table 5.

Table 5. Occurrence and the Fates of Pharmaceutical Drugs in the Environment

Drug (Drug group) Fate (Concentration) Environment Reference

Macrolide antibiotics sub ng/L River Abuin et al. (2006)

PPCPs ~g/L River Loraine and

Pettigrove (2006)

Neutral drugs ~g/L STP, river Ternes et al. (2001)

PPCPs ~g/L,

~g/kg

STP,

Biosolid

Xia et al. (2005)

10 selected drugs non-persistent

(half life < 20 d)

medium persistent

(half life 15-54 d)

water/sediment Lffler et al. (2005)

6 selected drugs ~g/L STP, river Tixer et al. (2003)

10 selected drugs ~ng/L Surface water Standley et al. (2008)

10 selected drugs higher removal

efficiency than

activated sludge

process

nitrifier culture Tran et al. (2009)

40

8 selected drugs half life 1.5-82d Surface water Lam et al. (2004)

PPCPs ~ng/L - ~g/L River Yoon et al. (2010)

PPCPs sub g/L River Ellis (2006)

PPCPs sub g/L,

~ng/L - ~g/L,

sub g/L

run-off,

STP,

River

Pedersen et al. (2005)

13 selected drugs sub ng/L STP, river Stumpf et al. (1999)

EDCs and drugs sub g/L,

sub ng/L

STP,

Aquifer

Snyder et al. (2004)

Drugs sub g/L,

sub ng/L

STP,

Aquifer

Drewes et al. (2003)

131 compounds

(most drugs)

sub g/L Soil column Cordy et al. (2004)

6 selected drugs residues decreased

with reaction period

Soil Kreuzig et al. (2003)

Drugs ~ng/L STP, river, wetland Gross et al. (2004)

95 compounds ~ng/L - ~g/L River Kolpin et al. (2002)

5 selected drugs removal efficiency:

77-99%

Drinking water facility Ternes et al. (2002)

98 compounds 16 compounds:

partially oxidized (32-

92%)

22 compounds:

completely oxidized

Water distribution

system

Gibs et al. (2007)

Diazepam n.a. STP van der Ven et al.

(2004)

60 selected drugs ~ng/L Groundwater Sacher et al. (2001)

18 antibiotics sub g/L STP, river Hirsch et al. (1999)

PPCPs ~ng/L - ~g/L River Moldovan (2006)

106 compounds sub g/L Drinking water facility Stackelberg et al.

(2004)

110 compounds ~g/L STP, river Glassmeyer et al.

(2005)

41

psychoactive drugs sub g/L STP, river Hummel et al. (2006)

Drugs sub g/L STP, river Ternes (1998)

Drugs sub ng/L STP, river, aquifer Heberer (2002a)

Ibuprofen ~ng/L STP, river Buser et al. (1999)

FDAs Drug Review Process

It takes years for pharmaceuticals developed in a lab to be delivered to

consumers. Rigorous tests such as animal tests, tests for side-effects, and properly

designed clinical trials are required for approval of new pharmaceutical drugs by

the US FDA. The drug review process for new pharmaceutical drugs by the US

FDA is very complex and major steps are as follows (US FDA):

Preclinical (animal) testing

An investigational new drug application (IND) outlines what the sponsor

of a new drug proposes for human testing in clinical trials

The pharmaceutical industries must submit to the FDA the results of

preclinical testing they have done in the laboratory using animals and what

they propose to do for human testing. At this stage, the FDA decides

whether it is reasonably safe for the company to move forward with

testing the drug in humans.

Phase I studies (typically involve 20 to 80 people)

42

Usually this study is conducted in healthy volunteers. The goal of Phase I

is to determine what the drugs most frequent side effects are and how the

drug is metabolized and excreted.

Phase II studies (typically involve a few dozen to about 300 people)

Phase II begins when unacceptable toxicity is not revealed in Phase I. The

goal of Phase II is on the effectiveness of each drug. In this Phase, a

placebo test is done in order to compare the effectiveness of the target

drug to a control.

Phase III studies (typically involve several hundred to about 3000 people).

Phase III beings when the evidence for effectiveness is demonstrated in

Phase II. The goal of Phase III is to gather much more information about

the target drug by testing different dosages and populations.

The pre-new drug application (NDA) period, just before a (NDA) is

submitted. This is a common time for the FDA and drug sponsors to meet.

Submission of an NDA is the formal step asking the FDA to consider a

drug for marketing approval.

When an NDA comes in, the FDA has 60 days to decide whether to file it

so that it can be reviewed. The FDA can refuse to file an application that is

incomplete.

The FDA reviews information that goes on a drugs professional labeling

(information on how to use the drug).

43

The FDA inspects the facilities where the drug will be manufactured as

part of the approval process.

FDA reviewers will approve the application or issue a complete response

letter.

Key Parameters Affecting the Fates of Pharmaceutical Drugs and Other Organic

Chemicals

Biodegradation

Biodegradation of a chemical in the environment can be defined as the

biotransformation of the target chemical by enzymes. The enzymatic

transformation by microorganisms is the most general way for organic chemicals

to be mineralized. Biodegradation is a very important mechanism to remove

chemicals in the soil or sediment environment (Pavan and Worth, 2008). It may

happen in the aquatic environment by extracellular enzymes or by intracellular

enzymes of organisms. Biodegradation by intracellular enzymes occurs after

passive diffusion of non-polar substances such as alcohols and phenolic

compounds. Polar and dissociated compounds like aliphatic and aromatic acids

require a transport system to pass into the cell. The polarity of a compound is very

important for biodegradation because it influences the toxicity of the target

chemical and regulates the internal concentration of a chemical in an organism

44

(Wittich, 1996). In order to assess biodegradation, it is necessary to consider the

toxicity of a chemical to organisms. Organic chemicals can also be divided into

non-polar narcosis (Narcosis I) and polar narcosis (Narcosis II). Narcosis I

compounds such as 1-octanol, 1,2-dichlorobenzene, and 2-butanone are

considered to have the least toxic mode of action and are often called baseline

toxicity. Most organic chemicals show characteristics of Narcosis I, and if the

measured toxicity is below the baseline toxicity, readily biodegradation is

possible. Narcosis II compounds such as phenol, 4-acetylpyridine, and 2-

chloroaniline have more toxic properties than Narcosis I compounds (Knemann,

1981; Sixt et al., 1995; Veith et al., 1983). The toxicity of Narcosis I compounds

is dependent upon log Kow, while that of Narcosis II increases as hydrogen

bonding groups on molecule increase (Lin et al., 2004). Thus, the log Kow is an

important factor to estimate and understand the biodegradability and the toxicity

in the environment. Benzene derivatives have been used for solvents, propellants,

additives, cooling agents, insecticides, and herbicides for many years (Sixt et al.,

1995). The toxicity of each benzene derivatives was measured by QSAR analysis

and the acute lethal toxicity for amphibians was linearly proportional to the log

Kow of each compound (Huang et al., 2003). Veith et al. (1983) reported that the

toxicities of several benzene derivatives, alcohols, and alkenes were proportional

to log Kow. However, this linearity deviated around a log Kow of 6. Smith et al.

(1994) measured the toxicity of fish and marine invertebrates which took up

45

phenol and 18 chlorophenols (up to pentachlorophenol and their isomers).

According to their report, as the number of chlorine atoms of each chlorophenol

increased, log Kow and the relative pKa value increased as well as the toxicity

caused by each chlorophenol also increased. For the case of phenol, when one

chlorine atom is attached to the aromatic ring the toxicity increased up to 11 fold.

Specifically, when meta-positions were substituted with chlorines the toxicity

markedly increased as compared to ortho- or para-positions at higher log Kows

(Smith et al., 1994). In addition, the toxicity is significantly affected by the pKa of

a chemical. Knemann and Musch (1981) reported that the toxicity was a function

of the hydrophobicity of a chemical, pKa, and pH. The toxicity of phenol

derivatives increased as log Kow and pH increased and pKa was decreased. Smith

et al. (1994) presented a regression curve consisted of log Kow and pKa of

chlorophenol derivatives. The coefficient of log Kow was greater than that of pKa.

It suggests that log Kow is a more important factor than pKa for biodegradation.

Pavan and Worth (2008) reported that degradation can be defined as follows:

Primary degradation

Production of organic derivatives

Mineralization

Complete degradation of an organic chemical to stable inorganic species

46

Degradation is also divided into abiotic degradation and biotic degradation as

follows (Pavan and Worth, 2008):

Abiotic degradation

Transformation by chemical reactions like oxidation, reduction, hydrolysis,

and photodegradation

Biotic degradation

Transformation by enzymatic reactions

Several investigators reported the effects of functional groups attached to

aliphatic carbon or aromatic carbon on biodegradation. The positive or negative

effects on biodegradation are shown in Table 6.

47

Table 6. Several Functional Groups Attached to Aliphatic Carbon or Aromatic

Carbon Affecting Biodegradation

Functional group Positive

effect

Negative

effect Reference

Highly branched compounds

Ether x

Howard et al.

(1987)

High Molecular mass

Alkyl branching

Degree of chlorination

Sites of unsaturation

Low solubility of molecules

Halogenation

Heterocyclic N-aromatics

x Boethling and

Sabijic (1989)

Hydrolyzable groups (ester, amide, anhydride)

Hydroxyl group

Carboxylic acid group

x Boethling and

Sabijic (1989)

Ester functional groups

Alcohols

Acids

Amide functional groups

x Howard et al.

(1992)

Chlorine atom attached to carbon

Halogen substitution

Quaternary carbons

x Howard et al.

(1992)

Long linear alkyl chains x Boethling et al.

(1994)

Heterocyclic N-aromatics x Boethling et al.

(1994)

Hydroxyl and carboxyl functional aromatics x Alexander and

Lustigman (1966)

Nitro and sulfonate functional groups x Alexander and

Lustigman (1966)

Hydrophobic chemicals x Alexander (1973)

48

Ester

Presence of CO bonds

Acyclic structure

Acid

Long non-branched alkyl chains

Hydroxyl groups attached to a chain structure

Carbonyl, ester, or acid groups attached to a chain

or a ring

x Pavan and Worth

(2008)

Aromatic rings

Halogen substituent on a chain or a ring

Quaternary carbons

Tertiary and aromatic amines

x Pavan and Worth

(2008)

Biodegradation of Xenobiotic Compounds

Many xenobiotic compounds are aromatic compounds. Representative

aromatic xenobiotics are bisphenol A, dichlorodiphenyltrichloroethane (DDT),

polychlorinated biphenyls (PCBs), and steroid hormones. Specifically, most

polychlorinated aromatic carbons including xenobiotic compounds released into

the environment have accumulated in the aquatic or soil environment due to their

low degradability (Smith et al., 1994). In general, the persistence of xenobiotic

compounds is predominantly determined by their substitutions on aromatic rings.

For instance, the replacement by sulpho, nitro groups, and halogens can severely

affect the degradation of xenobiotic compounds. However, Wittich (1996)

reported that despite restricting degradation by sequencing enzymes, xenobiotic

compounds could be still degraded by cometabolism. Consequently,

approximately 70% of xenobiotic compounds degraded were transformed into

49

carbon dioxide and water, while the approximately remaining 30% was converted

into biomass (Wittich, 1996).

Biodegradation of Pharmaceutical Drugs in the Environment

Many investigators reported the fates of pharmaceutical drugs in the

environment, especially STPs and receiving rivers (Heberer, 2002a; Heidler and

Halden, 2008; Kolpin et al., 2002; Ternes, 1998; Yoon et al., 2010). The main

removal mechanism of hydrophobic drugs and metabolites which do not have

polar properties is to be sorbed by activated sludge in STPs and solids in rivers.

Hydrophilic drugs and most metabolites are usually removed by biodegradation

(metabolism and/or cometabolism) in the aquatic or soil environment.

In general, the sources of pharmaceutical drugs can mainly be divided into

two groups. One is discharged from STPs together with the urine and feces. The

other comes from veterinary animal treatments and rainfall run-off (Pedersen et

al., 2005). Although the biodegradation pathways of pharmaceutical drugs are

severely affected by their exposure routes (from sources to endpoints) in the

environment, the fates of drugs are determined by their physical, chemical, and

biochemical properties. Specifically, the anticipated exposures of drugs can be

significantly affected by interactions between soil and water. This interaction is

dependent upon the hydrophobicity of drugs such as Kow or Koc because the

50

sorption effects of drugs determine the magnitude of the interaction between soil

and water (Jrgensen and Halling-Srensen, 2000).

In the aquatic environment, the occurrence and fates of pharmaceutical

drugs such as analgesics, lipid regulators, beta blockers, and steroid hormones has

been considered as one of the emerging issues (Heberer, 2002b; Kmmerer,

2001a; Roger et al., 1986; Shore et al., 1993). These compounds are partially or

completely mineralized by biochemical, chemical, and physical degradation, and

also transported other environments. In addition, xenobiotic drugs used or unused

can be directly or indirectly released into the environment. Heberer (2002a)

showed that most pharmaceutical drugs in the aquatic environment can easily

percolate into soil, groundwater, and affect even drinking water. Jrgensen and

Halling-Srensen (2000) stated that the possible fates of xenobiotic drugs in a

STP as follows:

Parent drugs and metabolites of them are decomposed to carbon dioxide

and water by heterogeneous fungi and bacteria.

Parent drugs and metabolites of them may be persistent and persistence

are subordinate to hydrophobicity or ionic bonding. The hydrophobicity

of chemicals determines the time in a STP.

Parent drugs and metabolites of them very polar cannot be easily degraded

in a STP and finally reach the aquatic environment.

51

Sorption

Sorption also plays a very important role in removing micropollutants

from the aquatic or soil environment because it determines the amount of

contaminants and their expected concentrations in the media. Sorption influences

the partitioning such as micropollutents captured in soil, volatilization,

solubilization, and biodegradation by microorganisms. Nendza (1998) reported

that sorption results from the following several physicochemical interactions

between adsorbates and adsorbents.

van der Waals interaction

hydrophobic bonding

hydrogen bonding

charge-transfer interactions

ligand exchange and ion bonding

direct and ion-dipole and dipole-dipole interactions

covelent binding

When the concentration of a chemical in n-octanol/water is at equilibrium,

the Kow is defined as the ratio of the concentration of a chemical in n-octanol to

that in water. The Kow is commonly used to evaluate the hydrophobicity or

hydrophilicity of a chemical (Ghasemi and Saaidpour, 2007; Nendza, 1998; Yang

52

et al., 2009). As the Kow presents the partitioning of an organic chemical this

ratio has been used as one of key parameters in QSAR analysis. Specifically, log

Kow was used to estimate the possibility of absorption/adsorption and

hydrophobicity of a chemical (Cronin and Schultz, 1996).

However, the Kow has some problems to discriminate whether a chemical

is hydrophobic or hydrophilic. Baker et al. (2000) reported that most correlations

between the Kow and the Koc were not suitable for estimating the fates of organic

chemicals because they have been developed by aqueous toxicity or food chain

models, not by persistent organic pollutants (POPs). Holten Ltzhft et al. (2000)

showed this important fact with humic acids. Even though a chemical is

considered as a non-dissociated form at a low Kow, humic acids can affect the

polarity of the chemical since humic acids contain several polar groups which can

interact with the polar groups of the chemical. Thus, the Koc of a chemical can

present more accurate results due to the electrostatic interactions (Holten Ltzhft

et al., 2000). The persistence of heterocyclic aromatics decreases as the Koc

(organic carbon partitioning coefficient) and temperature increased, and

persistence was proportional to the initial concentration of the chemical in soil

(Celis et al., 2006). Grathwohl (1990) reported that the amount of sorbed

chlorinated aliphatic hydrocarbons increased as the Koc increased. In addition,

log Kow is closely correlated to other parameters such as molecular weight, water

solubility, and log Koc. In general, log Kow is inversely proportional to log water

53

solubility. However, as log Kow increased greater than approximately 5, the

relationship between log water solubility and log Kow becomes unreliable due to

the deviation of data (Nendza, 1989). Baker et al. (1997, 2000) reported that log

Kows greater than 6 were not well correlated to log Koc. Correlations between

log Koc and log Kow are shown in Table 7.

Table 7. Correlations between log Koc and log Kow for Several Chemical Groups

Chemical group Model Reference

Pesticides log Koc = 0.52 log Kow + 1.12 Briggs (1981)

Pesticides log Koc = 0.54 log Kow + 1.38 Kenaga and Goring (1980)

Aromatics, PAHs log Koc = 0.83 log Kow + 0.29 Hodson and Williams (1988)

Aromatic herbicides log Koc = 0.94 log Kow - 0.01 Brown and Flagg (1981)

Aromatic hydrocarbons log Koc = 0.99 log Kow - 0.35 Karickhoff (1981)

Phenols, benzonitriles log Koc = 0.57 log Kow + 1.08 Sabljic et al. (1995)

Esters log Koc = 0.49 log Kow + 1.05 Sabljic et al. (1995)

Cometabolism

Since Leadbetter and Forster (1958) reported cometabolic methanotrophy,

many cometabolism papers have been published. There are several key definitions

for cometabolism as follows:

54

The transformation of an organic compound by a microorganism that is

unable to use the substrate as a source of energy or of one of its

constituent elements (Alexander, 1967)

Any oxidation of substances without utilization of the energy derived from

the oxidation to support microbial growth and does not infer the presence

or absence of growth substrate during the oxidation (Horvath, 1972)

The transformation of a non-growth substrate in the obligate presence of a

growth substrate or another transformable compound (Alexander, 1981)

The process whereby a substrate is modified but not utilized by an

organism growing on another substrate (Dalton and Stirling, 1982)

The abilities of microorganisms to transfer non-growth-supporting

substrates, typically in the presence of a growth supporting substrate (Arp

et al., 2001)

The common theme of the above definitions is that the degraded non-

growth substrate cannot be used for cell synthesis in the presence of the