Comparison of Selected in Vitro Assays for Assessing the Toxicity of Chemicals and Their Mixtures

Home , In vitro toxicology, Toxicity category rating

COMPARISON OF SELECTED IN VITRO ASSAYS FOR ASSESSING THE TOXICITY OF CHEMICALS AND THEIR MIXTURES

Rola Azzi

A thesis submitted for the degree of Doctor of Philosophy Chemical Safety and Applied Toxicology Laboratories School of Safety Science Faculty of Science The University of New South Wales

June 2006 Certificate of Originality

Certificate of Originality

I hereby declare that this submission is my own work and that, to the best of my knowledge it contains no materials previously published or written by another person, nor material which to a substantial extent has been accepted for the award of any degree at UNSW or any other education institution, except where due acknowledgment is made in the thesis. Any contributions made to the research by others, which whom I have worked, is explicitly acknowledged in the thesis.

I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project’s design and conception or in style, presentation and linguistic expression is acknowledged.

Rola Azzi

June 2006

i Acknowledgments

Acknowledgments

I would like to express my sincerest gratitude to my supervisor Dr Amanda Hayes for her assistance, constant encouragement and expertise. Without her support and guidance this project would not have been possible. I am also grateful to Associate Professor Chris Winder, for his constant support, advice, and scientific expertise throughout my work on this project.

I would also like to express my sincerest gratitude and appreciation to my uncle, Associate Professor Rachad Saliba, who was a constant support during the writing of this thesis, and was devoted to helping me grasp the science of statistics, and its ability to transform numbers into a story.

To my colleagues at the School of Safety Science: Dr Christian Khalil, Mrs Fatma Lestari, Mrs Shahnaz Bakand and Mrs Aparna Manglik. I thank you for your friendship and support.

To the special friends who shared this journey with me, you have always been there for me, even from across the seas. Thank you for the constant support and friendship, it made life much more meaningful.

Special thanks go to my family who have always been there to support and encourage me in all my endeavours. A special thank you to my grandparents for believing in me so much. Last and most importantly to my parents and sister: you are my backbone in life. Thank you for the considerable sacrifices and supporting me in following my path in life even when it meant living in another country. I could not have made it through this thesis without your unconditional love and unlimited support in every possible way.

This research is dedicated to my brother Tony. You are and will always be my inner strength, and my hope. I miss you.

ii ‘Indulge your passion for science but let your science be human, and such as may have a direct reference to action and society’

(David Hume 1711-1776) (Enquiry Concerning Human Understanding: Introduction, 9)

iii Abstract

Abstract

From a regulatory point of view, the main objective of acute toxicity testing is to classify chemicals according to their intrinsic toxicity. This is conventionally conducted on the basis of the animal LD50 methods however, this test has been widely criticised. Moreover, conventional toxicity testing focuses on single chemicals and often human exposures are to more than one chemical. This research study had two main objectives. The first was to investigate the accuracy of selected in vitro tests for predicting the acute toxic action of chemicals in rodents and humans, and in predicting the Globally Harmonised System of Classification and Labelling of Chemicals (GHS) categories. The second was to explore the relevance of in vitro tests in determining the nature of toxicological interactions (i.e. additive, antagonistic and synergistic) among binary and ternary chemical mixtures. A battery of cytotoxicity tests (MTS, NRU, LDH and ATP) assays were used to determine the toxicity of 21 chemicals spread across the GHS categories where mercuric chloride (GHS category 1) was the most toxic chemical and glycerol (GHS unclassified) the least toxic. Significant differences for the majority of test chemicals were found among all assays, highlighting the need for a battery of in vitro tests measuring different endpoints. The NRU assay was found to be a more sensitive measure of toxicity for most chemicals and predicted all of the GHS categories. In general, in vitro IC50 values correlated well with in vivo rodent (LD50); human acute toxicity (LDL0 and LC) data and published in vitro data. In addition, in vitro NOEC values correlated well with published TLV. Selected assays (MTS, NRU) were applied to binary (15) and ternary (5) chemical mixtures. Mixtures were prepared at mixture ratios proportional to the potency of individual components. Experimental data was used to assess the predictive capabilities of two approaches (mathematical model and concentration addition) commonly used by regulatory agencies which assume additive effect. However, all three interactions (antagonism, synergism and additivity) where observed in this study. These results suggest that these interactions cannot be excluded from toxicological risk assessments. The methods developed and information obtained from this study provide a comprehensive comparison between selected in vitro assays for assessing the toxicity of chemicals and their mixtures.

iv Table of Contents

Table of Contents

Certificate0H of Originality………………………………………………………………..i286H

Acknowledgments………………………………………………………………………1H ii287H

Table2H of Contents……………………………………………………………………… v28H

List3H of Tables …………………………………………………………………………..ix289H

Listof4H Figures…………………………………………………………………………..xi290H

Listof5H Abbreviations………………………………………………………………….xiii291H

Publicationsand6H Awards…………………………………………………………….xvii29H

Chapter7H 1. Introduction...... 1293H

Chapter8H 2. Literature Review...... 7294H

2.19H Toxicological Risk Assessment...... 7295H

2.1.110H Hazard identification...... 10296H

2.1.21H Dose-response assessment ...... 11297H

2.1.2.112H Hormesis ...... 13298H

2.1.313H Exposure assessment...... 1529H

2.1.414H Risk characterisation ...... 1730H

2.215H Methods of Toxicity Testing...... 18301H

2.2.116H Animal test methods...... 19302H

2.2.1.117H Criticism for the use of animal methods ...... 2330H

2.2.218H Non-animal test methods ...... 29304H

2.2.2.119H The Alternative movement...... 29305H

2.2.2.220H In vitro toxicity endpoints ...... 32306H

2.2.2.321H Mechanism of cellular toxicity...... 34307H

2.2.2.3.12H Basal cytotoxicity...... 35308H

2.323H Validating In Vitro Tests and Regulatory Acceptance ...... 37309H

2.3.124H The MEIC/EDIT approach...... 41310H

2.3.225H The ZEBET approach ...... 43H

26H2.3.3 ICCVAM...... 44312H

2.3.427H ECVAM ...... 4431H

2.428H Using InVitro Cytotoxicity Tests in Risk Assessments...... 45314H

2.529H Selected in vitro cytotoxicity assays ...... 49315H

2.5.130H The MTS cytotoxicity assay ...... 49316H

2.5.231H The NRU cytotoxicity assay ...... 51317H

2.5.332H The LDH cytotoxicity assay...... 52318H

v Table of Contents

2.5.43H The ATP cytotoxicity assay ...... 53319H

2.634H Risk Assessment of Chemical Mixtures ...... 54320H

2.6.135H Basic concepts of chemical mixture toxicology...... 55321H

2.6.1.136H Simple similar action (Concentration Addition)...... 5532H

2.6.1.237H Simple dissimilar action (Response Addition)...... 5632H

2.6.1.338H Interactions...... 57324H

2.6.239H Application to risk assessment...... 58325H

2.6.340H Role of in vitro in toxicity assessments of chemical mixtures...... 62326H

2.741H Regulatory Bodies: Hazardous Classification of Chemicals ...... 64327H

2.7.142H The USA ...... 65328H

2.7.243H The EU ...... 68329H

2.7.34H Australia ...... 7030H

2.7.445H International or multinational bodies ...... 7631H

2.7.546H The Globally Harmonised System (GHS) ...... 7832H

2.7.5.147H Classification criteria ...... 803H

2.7.5.248H Classification of mixtures ...... 8534H

Chapter49H 3. Aims and Objectives ...... 8735H

Chapter50H 4. Materials and Methods...... 9036H

4.151H Test Materials...... 9037H

4.1.152H Chemical selection ...... 9038H

4.1.253H Chemical information and mechanisms of toxicity ...... 9139H

4.254H In vitro Cytotoxicity Analysis ...... 103340H

4.2.15H Human cell culture ...... 103341H

4.2.256H Culturing method ...... 104342H

4.2.357H Preparation of test chemicals ...... 10434H

4.2.458H Dilution protocol ...... 10534H

4.2.4.159H Dilution protocol (i) ...... 105345H

4.2.4.260H Dilution protocol (ii) ...... 105346H

4.2.561H Controls...... 108347H

4.2.662H In vitro cytotoxicity protocols...... 108348H

4.2.6.163H The MTS assay...... 108349H

4.2.6.264H The NRU assay ...... 111350H

4.2.6.365H The LDH assay...... 112351H

4.2.6.46H The ATP assay ...... 117352H

4.2.767H Statistical analysis used for the calculation of cytotoxicity values...... 11835H

4.2.7.168H Calculation of standard error for MTS, NRU and ATP dose response curves

...... 118354H

4.2.7.269H Calculation of standard error for LDH dose response curve...... 11935H

4.2.7.370H Calculation of IC50 values (MTS, NRU, ATP and LDH) ...... 120356H

4.2.7.471H Calculation of NOEC ...... 121357H

4.2.7.572H Calculation of TLC values ...... 122358H

vi Table of Contents

4.373H Study of the Cytotoxicity of Chemical Mixtures ...... 123359H

4.3.174H Test materials ...... 124360H

4.3.1.175H Chemical selection ...... 124361H

4.3.276H In vitro cytotoxicity analysis for chemical mixtures...... 125362H

4.3.2.17H Human cell culture ...... 12536H

4.3.2.278H Preparation of test chemicals ...... 125364H

4.3.2.379H Protocol for MTS and NRU Assays...... 129365H

Chapter80H 5. Results and Discussions: Cytotoxicity of Test Chemicals...... 13036H

5.181H Introduction...... 130367H

5.282H Experimental Design...... 131368H

5.2.183H Human cell culture ...... 131369H

5.2.284H Test chemicals...... 131370H

5.2.385H Assays ...... 131371H

5.2.486H Dose response curve...... 132372H

5.2.587H Optimisation of cell numbers...... 13237H

5.2.68H Optimisation of serum concentration...... 136374H

5.2.789H Optimisation of exposure time...... 137375H

5.2.890H Inactivation of LDH activity with certain chemicals ...... 138376H

5.391H Cytotoxicity Results ...... 14137H

5.3.192H Ranking of test chemicals using the in vitro assays...... 141378H

5.3.293H Correlation between assays...... 145379H

5.3.2.194H Primary PCA modelling...... 152380H

5.3.395H Reproducibility of the in vitro assays ...... 153381H

5.496H Comparison of In vitro – In vivo Toxicity ...... 154382H

5.4.197H Correlation of in vitro – in vivo rodent data...... 15538H

5.4.298H Correlation of in vitro – in vivo human toxicity (LDL0)...... 161384H

5.4.39H Correlation of in vitro – in vivo human plasma (LC)...... 167385H

5.4.410H Correlation of in vitro - and TLV...... 169386H

5.510H Comparison between In vitro – In vitro Toxicity...... 171387H

5.5.1102H Correlation of in vitro – in vitro MEIC chemicals...... 17138H

5.5.2103H Correlation of in vitro – in vitro ZEBET (RC) ...... 174389H

5.6104H Cluster Analysis of In Vitro Data...... 174390H

5.6.1105H Comparison in vitro – GHS categories ...... 177391H

5.7106H Summary of Findings...... 178392H

Chapter107H 6. Results and Discussions: Cytotoxicity of Chemical Mixtures...... 18139H

6.1108H Introduction...... 181394H

6.2109H Experimental Design...... 182395H

6.2.110H Cell line ...... 182396H

vii Table of Contents

6.2.21H Test chemicals...... 182397H

6.2.312H Assays ...... 183398H

6.2.413H Dose response curve...... 18439H

6.314H Cytotoxicity Results ...... 18440H

6.415H Comparison between Assays ...... 186401H

6.4.116H Correlation in vitro – in vitro ...... 186402H

6.4.217H Comparison of in vitro data with GHS categories ...... 189403H

6.4.318H Reproducibility of the in vitro assays ...... 19140H

6.519H Data Analysis...... 194405H

6.5.1120H Prediction model ...... 194406H

6.5.212H Concentration Addition (CA) model...... 200407H

6.5.312H Graphical representation for binary mixtures ...... 204408H

6.5.4123H Graphical representation for ternary mixtures ...... 208409H

6.6124H General Discussion...... 213410H

6.7125H Summary of Findings...... 21641H

Chapter126H 7. Conclusion...... 218412H

Bibliography127H ...... 224413H

Appendix128H A ...... 25141H

Appendix129H B ...... 256415H

Appendix130H C ...... 259416H

Appendix13H D ...... 267417H

viii List of Tables

List of Tables

Table126H3 2-1 EPA estimations (1995) of chemical number in use ...... 8405H1

Table127H3 2-2 Types of animal toxicity tests used and relevant OECD guidelines...... 21406H19

Table128H34 2-3 MEIC battery of in vitro tests ...... 42407H2

Table129H35 2-4 List of alternative toxicity tests and their current status...... 46408H21

Table130H6 2-5 Major US chemical control laws and agencies...... 65409H2

Table13H7 2-6 CPSC classification of the toxicity of materials...... 67410H23

Table132H8 2-7 Major Australian chemical control laws and agencies ...... 7141H2

Table13H9 2-8 GHS, criteria for single dose toxicity ...... 82412H5

Table134H0 2-9 GHS criteria, skin corrosion/irritation ...... 82413H26

Table135H4 2-10 GHS criteria, serious eye damage/ eye irritation...... 8214H27

136H136H142HTable 2-11 GHS criteria, respiratory or skin sensitisation...... 83415H28

Table137H4 2-12 GHS criteria, single or repeated target organ systemic toxicity (TOST) .....84416H29

Table138H4 2-13 GHS criteria, germ cell mutagenicity ...... 84417H30

Table139H45 2-14 GHS criteria, reproductive and developmental effects...... 85418H3

Table140H6 2-15 GHS criteria, carcinogenicity ...... 85419H32

Table14H7 4-1 GHS categories for acute oral classification...... 90420H3

Table142H8 4-2 Selected chemicals and GHS categories for acute oral toxicity ...... 91421H3

Table143H9 4-3 Evaluation scheme of carcinogenicity to humans taken from IARC (2005)..9242H35

Table14H50 4-4 Test chemicals ...... 102423H6

Table145H 4-5 Test chemical concentrations for the study of 21 GHS reference chemicals10642H37 Table 4-6 Chemical mixtures and the GHS categories………………………………..123 Table 4-7 Selected ternary mixture combinations…………………………………….126 Table 4-8 Selected binary mixture combinations……………………………………..127 Table 4-9 Parameters for the study of the cytotoxicity of chemical mixtures………...128

Table146H52 5-1 Cytotoxicity assays endpoints...... 132425H38

Table147H53 5-2 Extracellular and intracellular LDH absorbance values for cells treated with

CuSO4, HgCl2 and CoCl2...... 139426H39

ix List of Tables

Table148H5 5-3 Experimental cytoxicity data of test chemicals as determined by assays.....142427H0

Table149H5 5-4 Ranking of chemicals based on their respective cytotoxicity assays ...... 143428H1

Table150H6 5-5 Correlation coefficient (r) of ranked chemicals...... 143429H

Table15H7 5-6 One way ANOVA and comparison of means...... 146430H

Table152H8 5-7 Correlation matrix between mean IC50 values of chemicals (21)...... 148431H

Table153H9 5-8 Calculated Coefficient of variation (CoV) for the assays...... 154432H5

Table154H60 5-9 Ratio of in vitro to in vivo rodent data and adjustment factors implement...15743H6

Table15H6 5-10 Correlation matrix in vitro – in vivo ...... 15843H7

Table156H2 5-11 Published in vivo human Lowest Lethal Dose (LDL0)...... 163435H8

Table157H63 5-12 Ratio of in vitro to in vivo human toxicity and adjustment factors ...... 164436H9

Table158H64 5-13 Correlation Table between IC50, LD50 values and LDL0 ...... 164437H50

Table159H6 5-14 Correlation Table between in vitro – in vivo human toxicity ...... 165438H51

Table160H 5-15 List of IC50, LD50 values, HETC and Human Plasma (LC) values ...... 168439H52

Table16H7 5-16 Correlation Table between IC50, HETC values and human plasma (LC)..16940H53

Table162H8 5-17 Cytotoxicity values (NOEC) and threshold limit values (TLV)...... 17041H5

Table163H9 5-18 correlation and regression analysis between IC50 and TLV ...... 17142H5

Table164H70 5-19 List of experimental in vitro and published in vitro data ...... 17243H56

Table165H7 5-20 Correlation coefficients of the in vitro assays ...... 1734H57

Table16H72 5-21 Grouping of chemicals as measured with NRU assay...... 17745H8

Table167H3 6-1 List of experimental IC50 values of tested chemical mixtures ...... 18546H59

Table168H74 6-2 Correlation coefficients of in vitro assays for chemical mixtures ...... 18847H60

Table169H75 6-3 Correlation and regression analysis of IC50 predicted and experimental .....19948H61

Table170H6 6-4 Concentration addition ratio and interaction effect of chemical mixtures. ..20349H62

Table17H 6-5 Parameters for graphical representation of binary mixture (NaF: Phenol)...205450H63

Table172H8 6-6 Parameters for the calculation of predicted IC50 value and 95% CI...... 205451H6

Table173H9 6-7 Parameters for ternary mixture (CdCl2: LiSO4: EtOH) ...... 209452H6

Table174H80 6-8 Parameters for the calculation of predicted IC50 value for ternary mixture .210453H6

x List of Figures

List of Figures

Figure24H30 2-1 Elements of Risk Assessment...... 9503H16

Figure25H31 2-2 Traditional toxicological dose-response curve ...... 11504H17

Figure26H3 2-3 Dose-response curve with hormesis...... 1450H18

Figure27H3 2-4 Areas of toxic mechanisms based on levels of organisation...... 35506H19

Figure28H34 2-5 Schematic figure representing the request for alternatives ...... 38507H2

Figure29H35 2-6 Schematic diagram for the conversion of MTS to Formazan ...... 51508H21

Figure230H6 2-7 The Luciferase reaction in the ATP assay...... 53509H2

Figure231H7 2-8 Modified overview of the US EPA guidelines for mixture risk assessment.59510H23

Figure23H8 4-1 Microtitre plate design for MTS, NRU and ATP assays...... 10751H24

Figure23H9 4-2 MTS dose-response curve (Absorbance vs. concentration) ...... 110512H

Figure234H0 4-3 MTS dose-response curve (% cell viability vs concentration)...... 110513H26

Figure235H41 4-4 Standard cytotoxicity plate for the LDH assay ...... 113514H27

Figure236H4 4-5 Microtitre plate design for LDH measurement ...... 11551H28

Figure237H4 4-6 Dose-response curve (Irgasan) % LDH released ...... 116516H29

Figure238H4 4-7 Dose-response curve (Irgasan) for extracellular LDH released ...... 116517H30

Figure239H45 4-8 Dose-response curve (Irgasan) for intracellular LDH (cell lysate) ...... 116518H3

Figure240H6 4-9 Dose-response curve for copper sulfate (MTS assay)...... 122519H32

Figure241H7 5-1 Skin fibroblast cell number against absorbance (MTS assay) ...... 132520H

Figure24H8 5-2 Skin fibroblast cell number against absorbance (NRU assay) ...... 133521H

Figure243H9 5-3 Skin fibroblast cell number against absorbance (ATP assay) ...... 13552H3

Figure24H50 5-4 Skin fibroblast cell number against absorbance (LDH assay)…………….135

Figure245H1 5-5 Effect of 4h and 24h exposure for KCl on %LDH leakage...... 138523H4 Figure 5-6 LDH assay: Enzymatic Reaction………………………………………….141

Figure246H5 5-7 Covalent bond between NADH and selected metals ...... 140524H3

Figure247H53 5-8 Correlation of chemicals based on ranking: NRU and LDH against MTS. 14452H36

Figure248H5 5-9 Correlation of chemicals based on ranking: MTS, NRU against ATP...... 144526H37

Figure249H5 5-10 Correlation of chemicals based on ranking: NRU, ATP against LDH. ....145527H38

xi List of Figures

Figure250H6 5-11 Regression plot of NRU, LDH IC50 values against MTS IC50 values...... 148528H39

Figure251H7 5-12 Regression plot of MTS, NRU IC50 values against ATP IC50 values...... 149529H40

Figure25H8 5-13 Regression plot using NRU, ATP IC50 values against LDH IC50 values .149530H41

Figure253H9 5-14 Bar plot using mean IC50 values of tested chemicals ...... 151531H42

Figure254H60 5-15 PCA modelling of in vitro assays...... 152532H4

Figure25H61 5-16 Graphical illustration of the observations and variables...... 15353H4

Figure256H 5-17 Regression between in vitro and in vivo LD50 values...... 158534H

Figure257H63 5-18 Correlation of in vitro values against in vivo human LDL0 ...... 16553H46

Figure258H64 5-19 PCA modelling of experimental in vitro and published in vitro...... 173536H47

Figure259H6 5-20 Hierarchical clustering of NRU in vitro IC50 values ...... 176537H48

Figure260H 6-1 Bar plot of in vitro IC50 values for chemical mixtures...... 187538H49

Figure261H7 6-2 Regression plot for selected chemical mixtures...... 188539H0

Figure26H8 6-3 Classification of chemical mixtures measured with NRU assay...... 192540H1

Figure263H9 6-4 Classification of chemical mixtures measured with the MTS assay...... 193541H2

Figure264H70 6-5 Bar plot of experimental and predicted MTS IC50 for chemical mixtures ..196542H3

Figure265H71 6-6 Bar plot of experimental and predicted NRU IC50 for chemical mixtures.197543H

Figure26H7 6-7 Theoretical additive interaction of a binary mixture of NaF and phenol ...20654H

Figure267H3 6-8 Graphical representation of 95% CI of additive line for binary mixture ....20654H6

Figure268H74 6-9 Example of an antagonistic interaction (NaF: Phenol) ...... 207546H7

Figure269H75 6-10 Example of an additive interaction (CuSO4 : phenol) (MTS)...... 207547H8

Figure270H6 6-11 Example of a synergistic interaction (SLS : NaF) (MTS)...... 208548H9

Figure271H 6-12 Region of additive interaction in a ternary mixture ...... 211549H60

Figure27H8 6-13 Plot of interaction regions in a ternary mixture...... 21150H61

Figure273H9 6-14 Example of an additve interaction (CdCl2:LiSO4: EtOH) (MTS)...... 21251H62

Figure274H80 6-15 Example of a synergistic interaction (SLS: Irgasan: NaF) (MTS)...... 21252H63

Figure275H81 6-16 Example of an antagonistic interaction (HgCl2: KCN: CdCl2) (MTS).....21353H64

Figure276H8 6-17 Interaction nature (%) of tested chemical mixtures...... 21454H6

xii List of Abbreviations

List of Abbreviations

ABS Animal Bovine Serum ACCC Australian Competition and Consumer Agency ACGIH American Council of Government and Industrial Hygienists ADI Acceptable Daily Intake ANOVA Analysis of Variance ANZCCART Australian and New Zealand Council for the Care of Animals ATLA Alternatives to Laboratory Animals ATP Adenosine Triphosphate ATSDR Agency for Toxic Substances and Disease Registry BMD Benchmark Dose BUAV British Union for the Abolition of Vivisection CA Concentration Addition CAAT The John Hopkins Centre for Alternatives to Animal Testing CPSC Consumer Product Safety Commission DMEM Dulbecco's Modified Eagle's Medium DPBS Dulbecco's Phosphate Buffered Saline EC European Commission ECEAE European Coalition to End Animal Experiments ECOPA European Consensus Platform on Alternatives ECVAM European Centre for the Validation of Alternatives to Animal Testing EDIT Evaluation-Guided Development of In vitro Tests EU European Union FAO Food and Agriculture Organisation FCS Foetal Calf Serum FDA Food and Drug Administration FQPA US Food Quality and Protection Act FRAME Fund for the Replacement of Animals in Medical Experiments GHS Globally Harmonised System of Classification and Labelling of Chemicals

xiii List of Abbreviations

HBSS Hank's Balanced Salt Solution HI Hazard Index HPV High Production Volume HSDB Hazardous Substance Data Bank IA Independent Action IARC International Agency for Research on Cancer

IC50 Inhibitory Concentration 50% ICAPO International Council for Animal Protection ICCVAM Interagency Coordinating Committee on the Validation of Alternative Methods ICITTS Integrated Toxicity Testing Scheme

ICX Inhibitory Concentration ILO International Labour Organisation IOMC Interorganisational Programme for the Sound Management of Chemicals IPCS International Programme on Chemical Safety IRS Integrated Risk System JECFA Joint Expert Committee on Food Additives JMPR Joint Meeting on Pesticide Residues JSAAE Japanese Society for Alternatives to Animal Experiments LC Human Plasma Lethal Concentration

LD50 Lethal Dose 50% LDH Lactate Dehydrogenase

LDL0 Lethal Dose Lowest LOEL Lowest Observable Effect Level MDI Diphenylmethane-4,4-diisocyanate MEIC Multicentre Evaluation of In vitro Cytotoxicity MRL Minimal Risk Levels MSDS Material Safety and Data Sheets MTS 3-(4, 5-dimethylthiazol–2-yl)-5-(3-carboxymethoxyphenyl)-2-(4- sulfophenyl)-2H-tetrazolium MTT 3-(4, 5-dimethylthiazol–2-yl)-2,5-DIphenyltetrazolium bromide NADH Nicotine Adenine Dinucleotide (reduced form)

xiv List of Abbreviations

NCA Netherlands Centre for Alternatives to Animal Use NHEXAS National Human Exposure Assessment Survey NHK Normal Human Keratinocytes NICEATM National Toxicology Program’s Interagency Centre for the Evaluation of Alternative Toxicological Methods NICNAS National Industrial Chemicals Notification and Assessment Scheme NIOSH National Institute for Occupational Safety and Health NOAEL No Observable Adverse Effect Level NOEC No Observable Effect Concentration NRU Neutral Red Uptake Assay NTP National Toxicology Program OECD Organisation of Economic Cooperation and Development OSHA Occupational Safety and Health Administration PAH Polycyclic Aromatic Hydrocarbon PBB Polybrominated Biphenyl PBDE Polybrominated Diphenyl Ethers PBPK Physiologically Based Pharmacokinetic Models PBT Persistent Bioaccumulative and Toxic PBTK Physiologically Based Toxicokinetic Model PCA Principal Components Analysis PCB Polychlorinated Biphenyls PEL Permissible Exposure Limit PMS Phenazine Methosulfate QSAR Quantitative Structure Activity Relationship RA Response Addition REACH Registration, Evaluation, and Authorisation of Chemicals RfC Inhalation Reference Concentration RfD Oral Reference Dose RTECS Registry of Toxic Effects of Chemical Substances SAR Structure Activity Relationship SDS Safety Data Sheets

TDL0 Toxic Dose Lowest TEF Toxicity Equivalency Factor

xv List of Abbreviations

TGA Therapeutic Goods Administration TLC Total Lethal Concentration TLV Threshold Limit Value TWA Time-Weighted Average UDP Up and Down Procedure UN United Nations US EPA The United States Environmental Protection Agency WHO World Health Organisation WOE Weight of Evidence XTT 2,3-bis (2-methoxy-4-nitro-5-sulfophenyl)-5-[(phenylamino) carbonyl] -2H – tetrazolium hydroxide ZEBET National Centre for Documentation and Evaluation of Alternative Methods to Animal Experiments. Location is in Germany

xvi Publications and Awards

Publications and Awards

Publications arising from this thesis

Winder, C., Azzi, R. and Wagner, D. (2005). The Development of the Globally Harmonised System (GHS) of Classification and Labelling of Hazardous Chemicals. Journal of Hazardous Materials 125, (1-3): 29-44.

Papers in preparation

Azzi, R., Hayes, A.J., Winder, C. Comparative analysis of selected in vitro cytotoxicity assays using human cells. To be submitted to Toxicology In Vitro.

Azzi, R., Hayes, A.J., Winder, C. Comparative in vitro study for determining the toxicity and safety of chemical mixtures. To be submitted to Toxicology journal.

Conference presentations

Azzi, R., Hayes, A.J. and Winder, C. (2005). An in vitro cytotoxicity study of the interactive effect of 24 binary and ternary chemical mixtures from the GHS classification groups. Oral presentation. 5th World congress on alternatives and animal use in the life sciences (ATLA). August 21-25, Berlin, Germany.

Azzi, R., Hayes, A.J., Khalil, C., and Winder C (2004). Comparison of in vivo toxicity data for GHS classification groups with data from in vitro cytotoxicity assays. Poster presenation. Australian Health and Medical Research Congress. Sydney, Australia, November 2004.

xvii Publications and Awards

Azzi, R., Hayes, A.J., Khalil, C., and Winder C (2004). Comparative analysis of selected in vitro cytotoxicity assays for use in standard setting of industrial chemicals. Poster presentation. 10th International Congress of Toxicology (ICT-X). Tampere, Finland, July 2004. Toxicology and Applied Pharmacology (Abstract No 608).

Azzi, R., Hayes, A.J., Khalil, C., and Winder C (2003). Comparison between in vitro and published in vivo data for standard setting of industrial chemicals. Australasian Society of Clinical Experimental Pharmacologists (ASCEPT). Sydney, Australia, November 2003.

Grants and awards received

Australian Postgraduate Award (APA) for the duration of the thesis.

Travel grant to attend the 5th World Congress in Alternatives and Animal Use in the Life Sciences. Berlin, Germany, August 2005. Awarded by Procter and Gamble (P&G), UK and the Humane Society o the United States (HSUS).

Travel grant to attend the 10th International Congress of Toxicology provided by the International Union of Toxicologists through ASCEPT (2004).

Industry Prize Blackmore Ltd for In Vitro Research Excellence (2002).

xviii Introduction

Chapter 1. Introduction

A prominent newspaper in Australia published on the 26th of April, 2002 an article on a consumer who used handy-man filler to stop a chimney draught in his home, and sued its manufacturer for $750,000 in damages. The consumer claimed the product left him brain damaged, destroying his marriage and career (Crishton 2002). The handy-filler used was made of self-expanding polyurethane foam. The making of polyurethane is brought about by the reaction between two chemicals a polyol (propylene oxide) and a polyisocyanate (diphenylmethane-4.4-diisocyanate, MDI). The consumer claimed that the MDI, contained in the product and classified by the National Occupational Health and Safety Commission of Australia as “harmful; Xn”, caused his illness. But, according to the safety data provided by the manufacturing company the polyurethane product is free from hazards in use since once the basic polyurethane-forming reaction has taken place the polyurethane formed is inert. Yet, within 48 hours of the second application using the polyurethane product the consumer was in a coma, where doctors diagnosed seizures and toxic encephalopathy. Doctors and toxicologist who treated him believed the filler caused the brain damage. The case was later dismissed in court based “on the balance of probabilities” that did not persuade the presiding judge that MDI could have contributed to the encephalopathy the consumer suffered. According to the judge the product was not used in an enclosed space and the room was well ventilated.

However, many unfortunate human experiences in the past such as pharmaceutical agents like thalidomide and diethylstilbestrol or occupational and environmental contaminants like lead and polychlorinated biphenyls (PCBs) (Zurlo et al., 1994; Greenburg and Phillips 2003; Langley 2005) have raised many questions on the availability of sufficient and adequate information on the adverse effects of exposure to chemicals, and their mixtures, which is essential in order to protect human health and the environment. Not forgetting the persistent need to test toxic effects of man-made mixtures of Persistent, Bioaccumulative and Toxic (PBT) chemicals that are potentially harmful when they accumulate in the body or the environment. Currently the toxic potential of industrial chemicals and household products is often assessed by using

1 Introduction

standard animal models, comprising the basic tests for risk assessments, as described in the Organisation for Economic Cooperation and Development (OECD) Guidelines for Testing of Chemicals (OECD 1983; OECD 2004). Therefore, attempting to predict the toxic potential of a chemical to humans by extrapolating the results obtained from animal experiments remains problematic (Barile 1994a). This information is then used by regulators to classify each chemical according to internationally harmonised guidelines in the first step (e.g. harmful, toxic, and irritant), then according to countries’ risk (R) phrases, for example “R-42: May cause sensitisation by inhalation”. For ethical, scientific and economic reasons, over thirty years there has been continuing intensive debate, and much research as to how these animal tests can be reduced, replaced or refined, to scientifically improve and harmonise the safety regulations and standards set by industries, governments and international organisations (Gad 2000).

Today, thousands of relatively common chemicals have never been tested in any system, mainly due to the cost and time consumption of animal tests (Barile et al., 1994b). Many new substances are also added and produced each year which could be assessed using routine carcinogenicity and cytotoxicity tests. Moreover, the current approach for dealing with individual chemicals is time consuming and expensive (US $270,000-450,000 per chemical, not including a 2-year standard cancer bioassay, costing US $3 million) and requires large numbers of test animals (Langley 2005). These resource constraints, as well as increased societal unease about the use of animal testing, have led to the development of alternative approaches such as in vitro assays and computer based modellings using structure-activity relationship (SAR) methods, among others (OECD 1996).

It is time to broaden the conventional method of setting chemical exposure standards to include not just animal studies but also in vitro derived data from human cell lines for a more accurate risk assessment. Further, in vitro toxicology could be a reliable method as bioassays have been able to predict the toxicity of not just individual chemicals but their mixtures, detecting possible toxic interactions between the compounds (Marinovich et al., 1996; Malich et al., 1998; Kortenkamp 2004). Potential hazards to health arise from the exposure to the several hundred new synthetic chemicals introduced each year and

2 Introduction

the myriad of approximately four million mixtures, formulations and blends already in commercial use (Yang 1996). These risks may be present for the work force producing the chemicals, the consumers using the chemicals, the general public, ecosystems and local, regional or global environments.

A number of national and international organisations and agencies have developed guidelines on assessment of exposure and the various health end-points (EPA 1992; IPCS 1999; OECD 2001a; ATSDR 2004). The various human health effects covered are: acute lethal toxicity; dermal and ocular irritation and corrosion; skin and respiratory sensitisation; target organ and target system toxicity; genotoxicity; carcinogenicity; reproductive toxicity; and the biokinetic endpoints of absorption, distribution and metabolism (Worth and Balls 2002). The major organ systems most likely to be affected by single or short term systemic toxicity are: liver, central nervous system, kidney, heart, hematopoietic system, skin and the lung (ICCVAM 2000).

Toxicological test methods range from information derived from human data, animal studies, ecotoxicological studies, in vitro studies, and structure-activity relationships (Cairns and Pratt 1990). Currently toxicological evaluations, manufacturing, distribution, packaging, labelling, use and disposal of chemicals and other materials are regulated or monitored on the basis of single or repeated dose acute toxicity data in animals. These data form the basis of risk assessments by a diverse and frequently confusing number of agencies, guidelines and regulations (Gad 2000). Regulations tend to vary between countries and even agencies within the same country (IFCS 2003). This discrepancy among labelling and classification criteria for different countries has created difficulties in classifying, labelling, packaging, transporting, using and disposing of chemical materials. A joint international effort has developed recommendations to address these concerns and the need for a Globally Harmonised System of Classification and Labelling of Chemicals (GHS) has been identified. The GHS, which was adopted in December, 2002, now moves to the front line of major regulatory issues facing virtually all government agencies with responsibility for regulating chemicals, as well as industry and unions over the coming years. This new system, which is the collaborative effort of the World Health Organisation (WHO), the International Labour Organisation (ILO), OECD, and the United Nations (UN), has

3 Introduction

broad support from the chemical industry because of its promise to harmonize at international level the manner in which chemicals are classified. The GHS provided the infrastructure for a globalised and consistent approach for the classification of chemicals. In addition, it also provided a coherent and consistent approach to defining and classifying chemical hazards and communicating information on labels and safety data sheets (Taskforce 2000).

This research is a study of the role that in vitro testing can play in limiting discrepancies and more accurately setting regulatory standards to chemical exposure. It tries to embody the Three Rs concept proposed by Russell and Burch in the Principles of Humane Experimental Technique (Russell and Burch 1959), through the application of alternative methods in toxicity testing. In the context of animal use, alternatives include “all procedures which can completely replace the need for animal experiments, reduce the number of animal required, or diminish the amount of distress or pain suffered by animals in meeting the essential needs of man and other animals” (Smith 1978).

The focus of the study is on the application of selected in vitro toxicity methods to detect the toxicity of selected chemicals and their mixtures on a human cell culture. The study will demonstrate the ability of these tests, to detect the cytotoxicity of selected reference chemicals taken from across the categories of the GHS, and binary/ternary mixtures of individual chemicals from the different categories of the GHS. The aim of the study is the potential integration into risk assessments for setting safety standards for industrial/commercial chemicals that may be used as a basis for classifying and labelling new and existing chemicals and products. Such an approach should be accurate, ethical (alternative to using animals), cost effective, eliminate the need for inter/intra species extrapolation through the use of human cells, and rapid to keep up with the thousands of newly untested products which are marketed each year.

4 Introduction

Organisational aspects

This research is fully funded by the School of Safety Science, The University of New South Wales (UNSW), Sydney, Australia. An Australian Postgraduate Award (APA) has been awarded for the duration of the Ph.D. All experimental work was performed at the Chemical Safety and Applied Toxicology Laboratories, School of Safety Science, UNSW.

Structure of the thesis

A general literature review on the framework of the risk assessment process for chemicals and the developments of in vitro toxicology are presented in Chapter 2. This chapter provides the research knowledge framework for this thesis. The review focuses on the elements of the toxicological risk assessment and its application for characterising and quantifying the potential adverse effects of chemicals (Section 2.1). Methods for toxicity testing of chemicals are discussed with a specific focus on research and recent developments in in vitro toxicology (Section 2.2). A review of the basic concepts of chemical mixtures in toxicology, application to risk assessment and the role of in vitro toxicology in toxicity assessments of chemical mixtures are discussed (Section 2.3). Finally, an overview of the regulatory bodies: the legislative and administrative measures to deal with chemical hazards are given with a specific focus on the GHS (Winder et al., 2005).

The aims and objectives of the project are defined in Chapter 3. Chapter 4 describes general materials and methods used in the project for the cytotoxicity assessment of selected chemicals. These include: criteria for chemical selection for single chemicals and mixtures, chemical information and mechanisms of toxicity, in vitro cytotoxicity analysis and the development of statistical analysis for the calculation of the cytotoxicity values.

5 Introduction

Procedure refinement of a battery of four in vitro cytotoxicity tests including: MTS, NRU, LDH and ATP assays is presented in Chapter 5. The cytotoxicity assays were also evaluated in their ability to determine the cytotoxicity of selected substances (21 chemicals) spread across the GHS categories for classification and labelling of chemicals. Cytotoxicity values were compared with published in vitro, in vivo rodent, human toxicity data and exposure threshold limit values. Cluster analysis of the chemicals using the experimental in vitro cytotoxicity data was also performed and compared with the GHS method of classification.

Method development including experimental design and data analysis of selected ternary (6) and binary (15) chemical mixtures taken from the different categories of the GHS are presented in Chapter 6. The application of two in vitro cytotoxicity indicators on the study of the selected chemical mixtures was also investigated (Chapter 6).

This thesis investigates the usefulness and validity of using in vitro cytotoxicity assays as part of a battery of in vitro tests in conducting future risk assessments on new and existing chemicals. This research develops new ways to examine the complex area of chemical mixtures, and enhances our knowledge about the interactive effects between chemicals in a mixture. The feasibility of detecting the different types of interaction using in vitro cytotoxicity assays as an ethical and more scientifically sound alternative to conventional animal testing is discussed. The distribution and dissemination of this work will be made available to the scientific community in the field of toxicology, regulatory bodies responsible for chemical regulation, and to the general public. Significant areas for future research will be identified.

6 Chapter 2

Chapter 2. Literature Review

Toxicology continues to develop by among other things, the application of the risk assessment process as a systemic, scientific approach for characterising and quantifying potential adverse effects to chemicals (Faustman and Omenn, 2001). Toxicity data are usually applied to toxicity classification and labelling requirements for risk assessment of chemical substances (Gribaldo et al., 2005). Toxicological interactions are rarely disclosed on labels or MSDS’s (Cote et al., 1998). Considering the widespread occurrence of multiple chemical exposures in the workplace, the need to anticipate possible toxicological interactions is essential. For example, simultaneous exposure to toluene and acetylsalicylic acid results in auditory impairment significantly greater than the sum of the effect induced by each substance individually (synergism) (Cote et al., 1998).

2.1 Toxicological Risk Assessment

Today’s modern society is becoming constantly more dependent on the use of chemicals (e.g. industrial products, pharmaceuticals, cosmetics, pesticides and household products) (Blaauboer 2002b). Reliable quality and adequate information on the adverse effects of exposure to chemicals, physical and biological agents is essential in order to protect the human health and the environment. According to US EPA estimations (1995) about 2,000 new chemicals are produced each year and 80,000 chemicals are already found in commerce, see Table 2-1.

Toxicological risk assessment is the process by which chemicals are evaluated for their potential impact on human health (Frazier and Goldberg 1990). It consists of several well defined elements including hazard identification, exposure assessment and risk characterisation (Blaauboer 2001; Blaauboer 2002a; Blaauboer 2002b) (Figure 2-1). Toxicological evaluations, mainly on the basis of animal experiments, currently form

7 Chapter 2

the basis of the assessment of risks from existing and new chemicals (Frazier and Goldberg 1990; BUAV 2001; Blaauboer 2002; Worth and Balls 2002; Blaauboer 2002b).

Risk assessments are currently based on information mostly derived from tests conducted on laboratory animals, this basis has been for many years criticised both for ethical and scientific reasons (BUAV 2001; Blaauboer 2002b). A review of the current alternative methods to animal or in vivo testing is looked at and the possibility of integrating in vitro tests in future risk assessments.

Table 2-1 EPA estimations (1995) of the number of chemicals available (Koëter, 2002)

Types of chemicals Estimated numbers of chemicals

Number of chemicals 5,000,000

Chemicals in commerce 80,000

Industrial chemicals 72,000 (millions of products)

New chemicals 2,000/year (1,000 in US)

Pesticides 600 (21,000 products)

Food additives 8,700

Cosmetic ingredients 7,500 (40,000 products)

Human pharmaceuticals 3,300

Toxicological risk assessment aims to provide estimates of chemical exposure levels and health risks, while identifying sources of uncertainty in scientific data (NRC 1988). Results of these assessments identify those chemicals that pose a considerable health or environmental risk and those that do not, thereby aiding both non-regulatory and regulatory risk management (Faustman and Omenn, 2001). The elements and framework of a risk assessment are outlined in Figure 2.1.

8 Chapter 2

Sources of data Sources of uncertainty Exposures and effects Human toxicity data Potency Exposure data Toxicity tests in animals Species sensitivity Exposed populations In vitro tests Susceptibility Toxic effects Structure activity Extrapolation methods Research relationships Other uncertainties

Toxicity assessment Dose-response Exposure Hazard identification assessment assessment

Risk Characterisation Risk Assessment

Evaluation of public health, environmental Policy decisions health social and Agency actions political consequences Risk Management

Figure 2-1 Elements of Risk Assessment Modified from NRC 1983; NRC, 1994; Faustman and Omenn, 2001; Kleinjans, 2003.

Risk assessment consists of both qualitative information on strength of the evidence and nature of the outcome and quantitative assessment of the chemical hazard, exposure and magnitude of risk. A risk assessment process involves combining predictions of toxic hazard with evaluations of likely exposure under specified conditions (Balls and Fentem 1992). At the moment toxicological risk assessment mathematical models dealing with the dose-response relationships, pharmacokinetic factors (e.g. physiologically based pharmacokinetic models), and inter-species extrapolation are being employed (IPCS 1999). The various elements of the risk-assessment paradigm encompass a variety of experimental activities. Hazard identification and characterisation (sometimes referred to collectively as hazard assessment) often rely on the use of animal experiments. Whereas exposure assessment is generally the result of chemical analysis, it might also depend on biomonitoring in animals or humans, and on computer-based estimations of exposure levels (Worth and Balls 2002).

9 Chapter 2

The most prominent elements of an assessment of the risk of the use of a chemical are: x Hazard Identification: identification of the inherent ability of a chemical to cause an adverse effect, and a quantitative description of the nature of these effects (Blaauboer et al., 1998; Worth and Balls 2002). x Dose-Response Assessment: or in general terms hazard characterisation is the (semi)-quantitative evaluation of the nature of adverse effects following exposure to a chemical, and where possible, a dose-response assessment (Holme and Dybing 2002). x Exposure Assessment: it is the third step in the process of risk assessment. The (semi)-quantitative evaluation of the likelihood of exposure of man, organism, system or (sub) population to a chemical (Frazier and Goldberg 1990). x Risk Characterisation: the (semi)-quantitative estimate of the probability of an adverse effect, its severity and duration in a given population, under defined exposure conditions, based on the above elements (Blaauboer et al., 1998).

A brief overview of the elements of the risk-assessment paradigm was shown in Figure 2-1 and is given in the following pages.

2.1.1 Hazard identification

There is an increasing public demand for a greater assurance regarding the safety of chemicals in the environment and its potential for human exposure (IPCS 1999). Hazard identification involves toxicity testing and rigorous gathering of the results of available epidemiological, toxicological and structure-activity studies (Faustman and Omenn 2001). While identifying the hazard of a chemical the assessor should utilise all available and relevant data including those from studies carried out for research or scientific investigation (e.g. academic laboratories, industry labs). Toxicity testing is an

10 Chapter 2

essential part of the development of chemicals to ensure their safe use under anticipated conditions. The use of animal studies is still currently a key component of the hazard identification process. A more detailed overview of toxicity testing is given in Section 2.2.

2.1.2 Dose-response assessment

The dose-response assessment describes the quantitative relationships between exposure to an agent and the incidence of adverse health effects as shown in Figure 2 -2.

No measurable effect Sub-lethal range Total Lethal range

LD50 Response

Increasing Log [dose]

Figure 2-2 Traditional toxicological dose-response curve for most non-carcinogenic and non sensitising substances with no beneficial effects

For toxicological effects considered to be sensitizing, genotoxic, carcinogenic and developmental it is assumed on conservative grounds that there is a probability of harm at any level of exposure (IPCS 1999). The remainder of toxicological effects are

11 Chapter 2

considered to show a threshold, a dose below which there would be no detectable effect (IPCS 1999).

Animal bioassay data are generally used for dose-response assessment. A number of statistically derived estimates are used to characterise the dose-response curve and form a basis for risk assessment. The No Observable Adverse Effect Level (NOAEL) is a simple estimate of the highest dose in an animal study in which the incidence of toxic effect was not toxicologically significantly different from the untreated group (Worth and Balls 2002). In some cases the lowest group being treated still shows a significant observable effect as compared to the controls, therefore the study cannot define a NOAEL giving the estimate of the Lowest Observable Adverse Effect Level (LOAEL). NOAEL are mostly used as a basis for calculating risk assessment such as Reference Doses (RfD) and Acceptable Daily Intake values (ADI) (Kalberlah et al., 2003). RfDs and Reference Concentrations (RfCs) are estimated from a daily exposure to an agent that is assumed be without adverse health impact on the human populations and may be based on epidemiological data. Uncertainty factors are applied to the NOAEL and LOAEL to convert these values in animals into a safe level of human exposure to determine ADI, and RfD (Kleinjans 2003). The application of appropriate uncertainty factors require expert judgment based on relevant experience. That is why the derivation of ADIs and similar parameters is performed by expert groups such as the Food and Agriculture Organisation of the UN (FAO/WHO), Joint Expert Committee on Food Additives (JECFA) and the Joint Meeting on Pesticide Residues (JMPR) (IPCS 1999). In addition to those international committees, the ADI approach is widely used in national and international regulatory committees such as the Scientific Committee for Food in the European Commission.

Conversion of the NOAEL from the toxicity database to acceptable human intakes or exposures must allow for differences between animals and humans (inter-species differences) and differences within the human population (inter-individual differences). Traditionally, a 100-fold uncertainty factor is used to achieve this extrapolation (Blaauboer 2001; Lutz 2002; Meek et al., 2002). This is comprised of two 10-fold factors to allow for uncertainties in both inter and intra species differences and additional uncertainty factors may be used if the quality of the data is not sufficient.

12 Chapter 2

Even though most extrapolation systems still use NOAEL itself as a starting point for risk assessment, it has been criticised (Kalberlah et al., 2003). The main criticism is that using only one dose neglects the information on the dose-response relationship gathered in the experiment, and study designs with low capacity of detecting toxic effects (use of low animal numbers) will be rewarded with higher NOAELs or concentrations. Therefore NOAEL is not the same as the biological threshold and may either underestimate or overestimate the true no-effect level (IPCS, 1999). The uncertainty with the selection of a NOAEL cannot be described quantitatively and may vary from case to case (Kalberlah et al., 2003).

Many scientists favour the use of mathematical dose response modelling such as the benchmark approach or the benchmark dose (BMD) (Kalberlah et al., 2003; Goldstein 2005). The BMD is thought to have a greater deal of stability than the usual determination of the experimental NOAEL to which safety factors are added (Crump 1995). The BMD mathematically describes the dose-response curve throughout the experimental range and gives the upper and lower bounds of the dose for a certain effect. This procedure is used by the US EPA within the Integrated Risk Information System that holds data in support of human health risk assessment (Kalberlah et al., 2003).

2.1.2.1 Hormesis

At present there is a belief that the toxicological predictive models that regulatory toxicologists use to predict and extrapolate dose responses from some chemicals (especially chemicals for which there are beneficial, or potentially beneficial effect) are based on erroneous assumptions (Calabrese and Baldwin 2003). This phenomenon is called hormesis see Figure 2 3. The most fundamental shape of the dose response is said to be neither threshold (for non-carcinogens) nor linear (for carcinogens) but U-shaped or J-shaped.

13 Chapter 2

Toxicant with no beneficial effect Toxicant with beneficial effect(s) Responsee Adverse 0 Beneficial Log[D ose]

Figure 2-3 Dose-response curve with hormesis Modified from (Calabrese et al. 1999)

Under the hormetic response a modest stimulation of response occurs at low doses and an inhibition of response occurs at high doses. There are many chemicals in the literature that do not abide by the threshold rule such as saccharin, dioxins, cadmium, mercury, numerous insecticides/herbicides, and numerous pharmaceutical agents (Calabrese and Baldwin 2003). For example, a low or modest consumption of ethanol reduces total mortality in humans while increasing it at higher levels of consumption (Calabrese 2002). But they are considered as the exception and not the rule, especially since most toxicological experiments are designed to assess doses that are too high for the hormetic domain (Calabrese and Baldwin 2003). The dose response affects nearly all aspects of toxicological, pharmacological, epidemiological and clinical evaluation (Calabrese and Baldwin 2003). Current hazard assessments are not designed to assess explicitly the concept of hormesis (Calabrese and Baldwin 2003). Considerable more time and resources are needed such as the requirements for greater number of doses below the NOAEL, and the use of more subjects to enhance statistical power and replication.

Numerous papers have been published over the past several years which indicate that the hormetic dose–response model is common in the biomedical/toxicological literature,

14 Chapter 2

and is highly generalizable according to biological model tested, endpoint measured and chemical class/physical agent employed (Gaylor 1998; Sielken and Stenvenson 1998; Calabrese et al., 1999; Calabrese 2002).

The risk assessment implications of hormesis for both toxicologists and regulators have been explored for both non-carcinogens (Gaylor 1998; Calabrese et al., 1999; Calabrese 2005a) and carcinogens (Sielken and Stevenson 1998). Toxicologists are affected in how they select biological models, choose endpoints to measure, design studies, assess risks and can help explain why there are cases of stimulation followed by inhibition in dose responses. The incorporation of the concept of hormesis and its quantitative features into the risk assessment process represents a data-driven decision to make the process more toxicologically based, with a formal recognition of the role of low-dose adaptive responses as legitimate and expected components of the dose-response spectrum, something that is presently excluded by the US EPA (Calabrese 2005a). As traditional risk assessment looks only at the harmful effects of a chemical and discards any possible beneficial effect, hormesis gives traditional risk assessors a broader ranger of toxicologically based options (Calabrese 2005a). Yet hormesis is not easy to study, for it requires the use of very low doses and needs more subjects (Calabrese and Baldwin 2003). However, the general consistency of the hormetic dose-response curve can allow for the harmonisation of risk assessment procedures for both non-carcinogens and carcinogens (Calabrese 2005b). However, it might still be a more prudent and precautionary approach to give adverse effects a priority.

2.1.3 Exposure assessment

In the past decade, exposure assessment has moved more rapidly than perhaps any other aspect of the four-step risk paradigm (Paustenbach 2000). The primary objectives of exposure assessment is understanding the hazard through determining the source, type, magnitude and duration of contact with naturally (e.g. aflatoxins in food) and nonnaturally (e.g. benzene in ground water, food additives) occurring toxicants (Paustenbach 2000).

15 Chapter 2

Human exposure means contact with the chemical or agent (Hodgson et al., 1998). This could be either: (a) External contact: through visible exterior of the person (skin and openings into the body such as mouth and nostrils) or (b) Absorption: through the exchange boundaries where absorption takes place (skin, lung, gastrointestinal tract).

An exposure assessment is the quantitative or qualitative evaluation of that contact; it describes the intensity, frequency, and duration of contact, and often evaluates the rates at which the chemical crosses the boundary (chemical intake or uptake rates), the route by which it crosses the boundary (exposure route; e.g., dermal, oral, or respiratory), the resulting amount of the chemical that actually crosses the boundary (a dose) and the amount absorbed (internal dose) (EPA 1992; Hodgson et al., 1998; Paustenbach 2000). Depending on the purpose for which an exposure assessment will be used, the numerical output of an exposure assessment may be an estimate of either exposure or dose. If exposure assessments are being done as part of a risk assessment that uses a dose- response relationship, the output usually includes an estimate of dose (EPA 1992).

To determine the magnitude and severity of the effects of exposure to a chemical, the nature of the exposed individual or population must be considered: age, gender, body size, nutritional status, health status, life style factors (e.g. smoking, drinking, drug habits) and both past and current exposures to other agents or stresses that might interact with the agent of concern (Faustman et al., 2001). Assessment of sources, pathways, environmental and biotransformations, routes of entry, time course of exposure, total exposure from all sources and activities and extrapolation from ambient levels to target tissue effective dose is expected (Omenn 1995).

The US EPA has published documents which provide guidelines for determining relevant exposure pathways for the calculation of an overall exposure assessment (EPA 1992). These calculations provide an estimation of total exposures for a specified population as well as calculation of exposure for highly exposed individuals. Measurement studies of the microenvironments in which humans are exposed have replaced area measurements of the past (Goldstein 2005). Sophisticated theoretical

16 Chapter 2

approaches using data from direct exposure measurements such as the National Human Exposure Assessment Survey (NHEXAS) study are being used (Goldstein 2005). Exposure assessment needs to assess questions like multiple chemical exposures (Mumtaz and Sipes 1993; Mumtaz et al., 1994). Exposure assessments are still conducted to characterise the risk associated with exposure to a single toxic agent by one route. However, it is known that individuals could be exposed to several toxicants by multiple routes regularly. The risk associated with such an exposure may be cumulative (EPA 1999; Chen et al., 2001). In an aggregate assessment the dose in this case is considered to be the total dose of a single chemical absorbed by an individual via all possible pathways. In cumulative assessment we have the addition of multiple chemical exposures by multiple routes. In this case an estimate of human exposure (dose) to a group of chemicals that act with the same mechanism of toxicity is measured (IPCS 1999).

Among the many advances in exposure assessment has been the increasing amount of data to understand the extent and variability of human exposure and to test models (Goldstein 2005). Interindividual variability or heterogeneity consists of real differences among people in the risk from a given hazard. These differences could be in terms of susceptibility as exposure may involve sensitive sub-populations, neonates, children, aged, pregnant women, and malnourished or diseased individuals (IPCS, 1999). Advances in both the fields of genomics and proteomics, will have great relevance for risk analysis in terms of identifying susceptible populations and a better understanding of the individual variability observed in responses to environmental agents (Goldstein 2005)

2.1.4 Risk characterisation

Risk characterisation is the final step of the risk-assessment paradigm, and it overlaps with risk communication and management (Goldstein 2005). Risk characterisation aims to provide a synthesis of estimates of exposure levels and health risks; it also summarises sources of uncertainty in scientific data and provides the primary basis of making risk management decisions (IPCS 1999). In this step, data on the dose-response

17 Chapter 2

relationship of a chemical are integrated with estimates of the degree of exposure in a population to characterize the likelihood and severity of a risk (Williams and Paustenbach 2002). Decisions concerning management of risks are made on the basis of identified and quantified risk and the potential for impact on individual humans, groups, populations and the environment. Health-based scientific data are one part of a two-step process to establish regulatory exposure limits; socioeconomic, political, risk benefit and cost benefit factors play a role in the decision-making (IPCS 1999). Too often, characterisation fails to integrate uncertainties inherent in key exposure parameters, dose-response assessments or analytical limitations, when interpreting risk information (Williams and Paustenbach 2002). This has sometimes led to the misinterpretation of the findings of a risk assessment or having false impressions about the degree of accuracy in reported risk estimates.

Specifying the risk to vulnerable populations is an important aspect of risk characterisation. A measure of uncertainty always accompanies the quantitative risk- assessment process. Public perception of risks will clearly drive political decisions to the setting of priorities and the responsiveness of risk managers. There is usually a difference between the public perception of risks and the expert ranking of risks. The precautionary principle has emerged in recent years and has been used to supplement risk assessments, to an extent of overlapping the scientific assessment of the risk (Goldstein 2005). Issues of the application of the precautionary principle and risk perception have played a major role in outlining US-European differences. For example the EU has used the precautionary principle to ban importation of beef from hormone- treated animals (Goldstein 2005). Future decisions on genetically modified (GM) foods will definitely involve the combination of risk assessments, the precautionary principle and risk perception in how nations will assess their risks.

2.2 Methods of Toxicity Testing

Toxicity testing is one of the two major components of risk assessment, the process by which new substances are evaluated for their potential impact on human health and

18 Chapter 2

welfare. The other component is assessment of exposure (Zurlo et al., 1994). Toxicity testing is required for new chemicals introduced into the marketplace, old chemicals that are proposed for new mixtures of old and new chemicals. The LD50 test is a classic example of an exposure-response test. It is a measure of acute lethality (Frazier 1992; Eaton and Klaassen 2001).

Toxicity data can be obtained from different sources, including toxicological studies, epidemiological studies, quantitative structure-activity relationships (QSARs) and physiologically based toxicokinetics (PBTK) studies. Within the context of risk assessment, the main objectives of toxicity testing are: to determine which potential adverse effects are of concern for a given chemical (Hazard Identification) and to provide adequate data to estimate the quantitative exposure-response relationship in the animal of concern (Frazier and Goldberg 1990). Traditionally, these goals were attained through the use of whole-animal studies. In fact, when toxicity testing began in the 1920’s and 1930’s the use of whole animals was the most logical choice because few alternatives were available, and some of these such as testing in human volunteers are now considered unethical.

2.2.1 Animal test methods

The design, conduct and completeness of reporting of experimental findings in toxicological studies on mammalian species are of critical importance in determining the validity and relevance of results (IPCS 1999). In general, these studies have three major objectives: x To identify the major toxic effects of the substances in question by examining a multitude of potential target tissues; x To identify toxic doses either from single or repeated exposure; x To determine the level of intake that does not result in adverse effects (NOAEL). Mammals, predominantly rodents are often used as the experimental animal species (Huggett et al., 1996).

19 Chapter 2

Animal based studies also generally lead to observations of the clinical, histopathological and/or functional changes in the animals caused by given dose of the chemical under study (Worth and Balls 2002). Stringent guidelines for animal testing and the development of toxicological risk evaluation over the past four decades has led to a number of standard procedures for which guidelines have been formulated by agencies and international bodies for example in the form of the OECD guidelines. Examples of such guidelines are those for acute oral toxicity OECD Test Guidelines 420 or 423, and 402 and 403 for dermal and inhalation toxicities (OECD 1983; OECD 2001a).The regulatory process relies to a great extent on data derived from experiments with laboratory animals in order to identify hazards and the dose response relationship of chemicals. The day to day use of these procedures has in many cases led to the relative safe use of chemicals and cosmetics. Animal tests are currently used in several areas of toxicity testing outlined in Table 2-2, with the respective OECD guideline. The main toxicological end points are: acute toxicity, skin irritation, eye irritation, corrosion, dermal sensitization, respiratory sensitization, chronic toxicity, mutagenicity, teratogenicity/embryotoxicity and carcinogenicity (Cote et al., 1998).

20 Chapter 2

Table 2-2 Types of animal toxicity tests used and relevant OECD guidelines Toxicological Type of animal test endpoints Eye irritancy Animal Adult albino rabbit (3 animals)

Test Dose effect monitored (21 days). Irritancy is scored according to a standard.

OECD No 405: Acute eye irritation/Corrosion (updated 24 April 2002) Skin irritancy and Animal Adult albino rabbit (3 animals) corrosion Test Dose applied on a shaved area of the skin. Irritancy is scored by checking against control patch of skin.

OECD No 404: Acute Dermal Irritation/Corrosion (updated 24 April 2002) No 434: Acute Dermal Toxicity – Fixed Dose Procedure (updating guideline 404) Skin allergy Animal Albino guinea pigs (17-30 per chemical) Test Dose substance is delivered to the skin by surface application or injection. Multiple doses applied in order to cause any local reaction

OECD No 406: Skin Sensitisation (updated 27 July 1995) Acute toxicity Animal Usually rats (15-30 per chemical).

Test Oral dosing of animals and observation for 14 days. All animals are autopsied at the end of the test period. Sex-specific responses are noted.

OECD No 401: Acute Oral Toxicity (deleted 20 December 2002) No 420: Acute Oral Toxicity - Fixed Dose Method (adopted 20 December 2001) No 423: Acute Oral Toxicity - Acute Toxic Class Method (adopted 20 December 2001) No 425: Acute Oral Toxicity: Up -and Down -Procedure (updated 23 March 2006) Repeat dose toxicity Animal Usually rats (40-80 rats per chemical). Sometimes dogs are also used (32 dogs per chemical). Animal repeatedly dosed with a chemical for 28-90 days. Dose administered orally, dermally or inhaled. Test Animals are then killed and their tissues examined pathologically and biochemically

OECD No 407: Repeated Dose 28-day Oral Toxicity Study in Rodents (updated 27 July 1995) No 408: Repeated Dose 90-day Oral Toxicity Study in Rodents (updated 21 September 1998) No 409: Repeated Dose 90-day Oral Toxicity Study in Non- Rodents (updated 21 September 1998) No 412: Repeated Dose Inhalation Toxicity (original 12th May 1981).

21 Chapter 2

Carcinogenicity Animal Very young rats or mice (at least 400)

Test Doses are administered either orally, dermally or inhalation. Outcome of exposure is assessed by blood sampling, pathological appearance and tissue and organ examination to detect cancerous changes. Tests costs $2 million per chemical and take 5 years to complete.

OECD No 451: Carcinogenicity Studies (original 12th May 1981) Chronic toxicity Animal Rats (400) or dogs as second species (32).

Test Oral and inhalation routes of administration. Length of study from 12 months to 2 years.

OECD No 452: Chronic Toxicity Studies (original 12th May 1981) Teratogenicity Animal Pregnant rats (80) or rabbits (48)

Test Doses delivered by mouth and the embryos killed and examined for gross or more subtle changes.

OECD No 421: Reproduction/ Developmental Toxicity Screening Test (original 27th July 1995). Mutagenicity Animal Rats, mice or Chinese hamsters (minimum 40)

Test Single or multiple orally/injected doses are administered into the body cavity. Two control groups are used (one receives no chemical and another receives a chemical known to cause an effect on genes). Tissue sampling is performed up to 48 hrs after dosing. OECD No 478: Genetic Toxicology: Rodent Dominant Lethal Test (Updated guideline 4th April 1984). Reproductive toxicity Animal Rats (100 females (80 pregnant) and 40 males)

Test Doses administered during reproductive cycle. Assessment is made on fertility, pregnancy. Tissue, brain or secondary sexual systems are studied for effects.

OECD No 421: Reproduction/ Developmental Toxicity Screening Test (original 27th July 1995). Toxicokinetics Animal Rodents and sometimes dogs (8 of each species).

Test Doses are administered (oral, inhalation, via the skin). Animals are then killed and examined for accumulation of test substance in presumed target organs. Excretion and metabolism time courses are also followed.

OECD No 417: Toxicokinetics (updated 4th April 1984) Endocrine disruptors Currently no validated methods specific to endocrine disruptors. Both in vivo and in vitro are being developed.

(Table modified from BUAV report (2001) and OECD Guideline for Testing of Chemicals – Section 4)

22 Chapter 2

For many decades, chemical classification for acute toxicity relied on the much controversial LD50 test (Hayes and Markovic 2000; Worth and Balls 2002). The LD50 test itself was designed in 1927 for the purpose of standardising biological preparations, and then adapted for testing the acute toxicity of chemical substances. The principle of the LD50 test is to dose groups of animals with a single dose of a test substance at concentrations expected to cause death in at least a fraction of the animals dosed. Normally at least 5 animals are used for the test. Results of the test enable the calculation of the LD50 value, being the dose that would kill 50% of the animals within 14 days after a single exposure. This type of information has continuously provided an important basis for the safety assessment of chemical substances for humans and mammalian wildlife, and is crucial when considering accidental exposures at the workplace, of young children at home and (for environmental assessments) following transport accidents. For testing by the oral route, the LD50 method was deleted from OECD test guidelines in 2002 in favour of protocols using fewer animals. It is now replaced by The Fixed Dose Procedure (TG 420), the Acute Toxic Class Method (TG 423), and the Up and Down Procedure (TG 425) (OECD 2001a). Although these alternatives still require the use of animals, the numbers of animals are drastically reduced and would normally be below one-third of those under the conventional LD50 test. Moreover, Guideline 420 does not require death to animals as an endpoint, whereas for the other alternatives (Guidelines 423 and 425) the expected number of deaths is typically not more than 3. However, for dosing by the dermal (skin) route and by inhalation, the LD50/LC50 tests are still the norm (Langley 2005). Acute toxicity data are used mainly to identify lethal/toxic doses of chemicals for humans (primarily for the regulatory purposes of classification and labelling) and indicate the mode of toxicity in humans, including the susceptibility of key target organs (Gennari et al., 2004).

2.2.1.1 Criticism for the use of animal methods

For many years now it has been argued that animal test data should not be continued to be used in making chemical policies. The reasons include the suffering caused by animal testing; the unreliability of animal (usually rodent) data when extrapolating to humans; the way that animal data can obstruct regulatory decision-making; and the cost

23 Chapter 2

and delays that would be caused by seeking animal data (BUAV 2001; Langley 2004; Langley 2005). However, the price for chemical safety has been a large one especially in terms of causing animal distress. According to a report published by the British Union for the Abolition of Vivisection (BUAV) in 2001, upon conventional testing the BUAV report estimates that up to 2123 animals will be used to test each High Production Volume (HPV) chemical. It has been estimated that a minimum of 12.8 million animals would be required to test 30, 000 substances; this figure rises to over 50 million when offspring produced during reproductive studies are included (BUAV 2001; Worth and Balls 2002).

There is both an ethical objection and scientific motivation for criticizing the heavy reliance on animal data in toxicology (Blaauboer 2002b). The literature is full of unfortunate accidents where reliance on animal testing did not always correctly predict human toxicity (thalidomide, asbestos, paracetamol, fialuridine, Bisphenol A) (Thorne 2001; Greenburg and Phillips 2003; Langley 2004).

From the regulatory point of view, acute toxicity data are mainly required for classifying and labelling chemicals according to their intrinsic properties (Gribaldo et al., 2005). Acute toxicity tests on animals, whether using the LD50 or other protocols, have never undergone formal validation to modern standards to establish their relevance for humans (OECD 1996). The European Coalition to End Animal Experiments (ECEAE) and the British Union for the Abolition of Vivisection (BUAV), have published a series of influential reports on the use of animal toxicity tests. These included The Way Forward, a comprehensive non-animal testing strategy, in 2001 and in 2004, A Regulatory Smokescreen that presented an analysis of the systemic failing of animal toxicity testing in chemicals regulation. A list of criticisms for all the current toxicity tests conducted on animals is found.

In summary the main reasons for inaccurate prediction from animal toxicity studies leading to either false negative or false positive responses and therefore hindering the process of risk assessment are: x Species differences: Variations in sensitivities to toxic chemicals, differences in rates and routes of absorption, distribution and excretion. These variations result in

24 Chapter 2

dissimilarities in the concentration of the compound at target tissues, therefore resulting in different toxicity readings. Or the species differences can be in the mode of the toxic action of the compound. An example is the case with paracetamol, even

among closely related species such as rats and mice. The LD50 for paracetamol in

mice is 400 mg/kg and death is caused by liver damage, whereas, in rats LD50 is 1000 mg/kg with no apparent liver damage. Other examples with a variation in ratio

between rat LD50 and mice LD50 are: nicotine (16.7), ethanol (2.0), and mercury II chloride (0.17) (BUAV 2001). In addition, rats and mice are more closely genotypically and phenotypically related to each other than they are to humans. Differences in reactions to chemicals between closely related species such as rats and mice demonstrate why acute toxicity tests on rodents have even more dubious predictivity for humans. x Exposure site (ear, skin, dorsal exposure) and mode of administration of

compounds: LD50 for the same chemical can vary depending on the mode of compound administration, environmental conditions and the statistical method

employed to calculate the LD50 (Shrivastava et al., 1992). Problems with dosing routes: conflicting study of Bisphenol A the route of chemical administration can have dramatic effects on apparent toxicity. In this case without the presence of direct information on human absorption and ability to metabolize the chemical, the animal results shed little light on human risk assessment (Langley 2004). For example, an orally administered chemical that is highly absorbed from the gut into the bloodstream can potentially affect every organ body but can also be immediately exposed to the metabolic activity of the liver. This may neutralise the chemical or convert it to a more toxic form. Whereas a chemical injected in the abdomen or under the skin is not immediately exposed to the liver’s metabolic activity. Therefore, if it is toxic it will cause adverse effects at lower doses than the same chemical applied orally (Langley 2004). Tests where a chemical is administered to animals by different routes may therefore yield different dose/response curves and different measures of toxicity. x Dosimetry: Problems with using small animals and large doses: most of the animal tests are conducted on rodents which are small-bodied and short-lived animals,

25 Chapter 2

where as the results are tended to be applied to large-bodies, long-lived animals (humans). Traditionally, scaling up has been used from small animals to larger humans to adjust for metabolic differences in rates or sizes. However, available evidence shows that predicting the clearance of a chemical in humans on the basis of body weight ratios from animal tests does not give reliable results. An error rate of more than 30% is observed for drugs (Mahmood and Balian 1999). There is even less certainty with chemicals, because unlike drugs, there are seldom clinical trial results to confirm or contradict the animal test data (Langley 2004). Tests are also conducted at unrealistically high doses in an effort to maximise the sensitivity of the experiment, since most small animals have short life spans and therefore can never be exposed to the chemicals as long as humans may be. For example an exposure “over a substantial period of an animals’ lifetime” is two-years with rat, and is assumed to be equivalent (in relative terms) to 60 years in a human. This is especially relevant for studies on carcinogenicity x Failure to consider absence of pre-existing pathological condition: extrapolations are being made from a rather small but homogeneous group of laboratory animals to the very heterogeneous human general population. The test animals which are usually inbred, specific pathogen free and genetically homogeneous do not represent normal animals of their species or the human population of concern. x Extrapolation from in vivo to humans: Problems with extrapolation to different species, breeds, and genders in addition to extrapolation from controlled experimental conditions to variable human situations exist. Risk assessors have traditionally attempted to overcome these uncertainties by introducing safety factors in establishing safety standards for human exposures (Faustman and Omenn 2001; Eisenbrand et al., 2002). x Weak reproducibility: reproducibility of acute toxicity tests for rodents is found to be quite weak. Reasons could vary from the difference in strains and species, ages of animals used between the different laboratories, as well as the differences in weights and diet of the animal all could significantly affect the results (Langley 2005).

26 Chapter 2

x More costly and time consuming: the new Chemical’s Policy of the EU (REACH) has set a goal to implement the testing of 30, 000 chemicals within the next 10 years. Each chemical that goes through the multiple tests required for registration can use up to 5,000 animals or 12,000 if the chemical is a pesticide. The cost of doing this for the 30,000 unregistered chemicals so that they comply with REACH has been estimated at between €5 billion (AUS $8 billion) and €10 billion (Abbott 2005). Animal tests are often slower than non-animal equivalents. For example it will take about 5 years and 400 rats for a carcinogenicity test (BUAV 2001; Abbott 2005). x Ethical considerations: most toxicity tests cause many animals to experience distress and serious illness, even the modified OECD testing guideline such as the Fixed

Dose Procedure causes deaths. And lethality is still the endpoint in the LD50/LC50 tests (i.e. dermal and inhalation routes), the Acute Toxic Class method and the Up- and-Down Procedure (Barlow et al., 2002). x Difficulty of evaluating effects of exposure to mixtures of chemicals: The potential synergy between many chemicals to which humans are exposure to cannot be studied in animal tests without access to considerable animal resources (Balls and Fentem 1992; Langley 2004). x Problems with fixed toxicological endpoints: While an impressive number of toxic endpoints are recognised in standard toxicity testing and classification systems (OECD 2004; UN 2005), other toxic endpoints emerge, such as endocrine disruption, decreased biodiversity, for which standardised methods are not yet available. x Reliance on animal data has displaced the role of monitoring programmes: most animal tests are currently devised to address recognised types of toxicity. Recently, efforts have been made to devise valid animal tests for “new” endpoints such as immune system damage, endocrine disruptors and damage to the nervous system. These tests have proven very difficult, slow and costly especially in the case of endocrine disruption. Two solutions are available: the first being the use of in vitro

27 Chapter 2

tests for a wide range of toxic endpoints that include tests to study changes in gene activity (microarray methods) or the production of metabolites (metabonomics) indicating significant changes in cellular functions. Second solution: routine monitoring which identifies emerging chemicals of concern on the basis of exposure and cumulative effects in people and wildlife, noting that toxicity testing traditionally measures hazard and not risk. x Problems with multiple chemical forms: many chemicals exist in several related forms (e.g. asbestos) or are single members of a large series of related chemicals (e.g. Brominated Flame Retardants: PBDE). Routine regulatory animal tests are conducted on a single chemical at a time, and the results are often extremely difficult to interpret. Therefore regulatory decisions on one chemical can be delayed while further animal testing is called for on yet another related chemical. This is the case with PBDE that are known to structurally relate known toxins (PCBs) and (PBBs). But the animal tests so far have been inconclusive to the toxicity of the PBDE and risk assessors are still trying to fill the “data gaps” with more animal testing. Instead of using in vitro testing and QSARs applied to the chemicals in a series which have missing data.

The data obtained from these tests leaves some gaps in understanding the actual effect of many chemicals and mixtures on people using products currently marketed as ‘safe’. The current system of controlling chemicals places great weight on animal test results, but, far from providing definitive evidence of toxic hazards, this reliance has led to delays in classification and control and to misclassification of chemicals (Langley 2004). Over the past 60 years, testing strategies based on whole animal procedures have evolved to produce the experimental database for safety/hazard evaluation. Animals were selected because they provided at the time the satisfactory surrogates for humans. But today alternatives do exist (Worth and Balls 2002). However, in light of the biotechnological advances of the last 30 years, this position is currently being re- evaluated. Developments in biomedical sciences have yielded a wide range of new techniques. The advancement of cell and tissue culture and of molecular biology provides enormous potential (Galli et al., 1993; Blaauboer 1996; Paine 1996; Zucco et al., 2004).

28 Chapter 2

Lethality is still the endpoint in LD50/LC50 classification and labelling based on LD50 values have been in use for decades. Now, implementation of the GHS is well underway (Winder et al., 2005). This international system means that all participating countries will interpret test results in the same manner. In the area of acute toxicity, the GHS currently relies on lethality and toxicity data from animal tests but need not be test- specific. The system could be adapted to use non-animal data to inform classification and labelling decisions (in vitro and in silico tests).

2.2.2 Non-animal test methods

“Testing of animals cannot be eliminated at present but every effort should be made to discover, develop and validate alternative test systems”. (OECD, 1982)

2.2.2.1 The Alternative movement

The emergence of the alternative movement is acknowledged to have originated in 1959 with the work of two British Researchers, W.M.S. Russell and R.L. Burch. They derived the concept of the three R’s (Reduction, Refinement and Replacement) in their classic work, The Principles of Humane Experimental Technique (Russell and Burch 1959). The issue of the three R’s and their application to toxicological research and testing has continued to influence research and development of new methodologies and testing strategies. They can be summarized as the development of test procedures which reduce the number of animals necessary for an experiment; refine existing animal tests by minimizing animal pain and distress, with the ultimate goal being the replacement of existing animal models with non-animal alternatives (Smith 1978; Russell and Burch 1959; Hayes and Markovic 2000). From the late 1980’s, in vitro tests have been proposed as a pre-screen or as an alternative method for endpoints such as prenatal toxicity, eye irritation, dermal

29 Chapter 2

irritation, tumour promotion and target organ toxicity (Frazier and Goldberg 1990; Frazier 1992; Atterwill 1995; Purchase et al., 1998).

In vitro toxicity studies have traditionally been aimed at assessing the mechanisms of action of chemical substances at molecular and cellular level, with interest in the physiological and biochemical processes. Toxicity studies have been often applied in pharmacology and toxicology, but genetic toxicology was the first sub discipline of toxicology in which in vitro test systems were used for toxicity testing to identify the mutagenic effects of compounds such as the Ames Mutagenicity Assay with strains of Salmonella typhimurium (Ames et al., 1973).

The particular development of non-genotoxic in vitro toxicity studies for testing purposes did not begin to expand again until the mid 1980s. In vivo toxicity methods were facing increased criticism, ethically and scientifically, while in vitro studies were thought to be easier, less time consuming, more humane and cost efficient. In vitro tests are considered to be more humane than are earlier test forms, and are more reliable, reproducible and predictive of some potential hazards in humans (Gad 2000).

Ecological and toxicological problems associated with environmental contamination needed the development of test methods that could produce large amounts of toxicity data to characterise chemical substances.

Over the past 70 years the goals of toxicity testing were attained through the use of whole-animal procedures that have evolved to provide the experimental database for safety/hazard evaluation (Gad 2000). Toxicity testing itself began in the 1920s and 1930s and at that time the use of animals was the most appropriate method because few alternatives were present (Frazier and Goldberg 1990). Using animals to test chemicals is expensive, and is becoming an ethically and sociably unacceptable approach for safety scientists (Purchase et al., 1998). Fortunately, due to the biotechnological advances of the last 20 years, new options are becoming viable. In general, in vitro toxicity describes the use of tests that do not use intact vertebrates as model systems. Toxicity in vitro is the study of toxic effects as observed in a system outside the body of a whole organism. These tests could include anything from lower

30 Chapter 2

organisms (planaria and bacteria) to cultured cells and computer models. In vitro refers primarily to the handling of cells and tissues outside the body under conditions, which support their growth, differentiation, and stability (Castell and Gomez-Lechon 1997). Alternatives incorporate everything that reduces higher animal usage and suffering in the existing traditional test designs to total replacement of the animals. In vitro cytotoxicology refers to the use of cell culture techniques in toxicology investigations (Barile 1994a; Gad 2000). Cell culture techniques have been widely used in toxicology, biomedical, medicinal chemistry, pharmacology, reproductive biology, cosmetics, oncology, environmental remediation and industrial chemicals (Barile 1994a). Cultured cell models can either be primary cultures or immortalised cell lines. Many advantages exist for using cultured cell models such as single organism can generate multiple cultures for use; cultures are stable; can be useful for a protracted periods of time and effects can be studied precisely at both the cellular and molecular level (Gad 2000).

The advantages of using in vitro studies include: x The feasibility of using human cells and tissues thereby avoiding species differences; x Greater simplicity than animal models making it easier to get results; x Greater ease in the application of modern, biochemical, cell and molecular biology techniques in mechanistic studies. x Ability to study components of potential target organs and target systems separately and in combination; x Ability to conduct studies on chemical mixtures.

However, there are a number of potential problems that arise with the reliance on in vitro testing. Applications of in vitro methods in toxicology are being faced by a number of difficulties. x The simplicity of the cell culture system, they cannot represent the complexity of the entire organism (Zucco et al., 2004); x In vitro systems lack possible mitigating systems (e.g. hormones, immunity, nervous system) and cannot replicate the biodynamics of the whole human body.

31 Chapter 2

x Difficulty in the interpretation of results as well as in extrapolation of results to potential effects in humans. Extrapolation to the outcome in humans requires knowledge of the pharmacokinetics of the compound. In addition, the limiting factors of the in vitro system such as: time limits, cell-cell or organ to organ interactions, need to be taken into account (Gad 2000). x Most in vitro tests to date are limited to acute toxicity testing rather than short term or long term repeated dose toxicity (Eisenbrand et al., 2002). x Several technical problems, could also affect in vitro tests, and they include: o type of cells employed to study in vitro cytotoxicity; o solubility of the test chemicals studied (e.g. alcohols did not form a medium- solvent partition), o exposure period may not always be sufficient contact time between cells and test chemicals to evaluate any delayed toxic potential of compounds.

Some of these difficulties can be overcome with the development and application of predictive tools such as PBTK models, and QSAR, as well as with the improvement of genomics and proteomics (Frazier 1992). Basic toxicology knowledge of absorption, distribution, metabolism and elimination (ADME) of chemicals is needed.

2.2.2.2 In vitro toxicity endpoints

Cytotoxicity as perceived in vitro is ultimately dependent on the endpoints and methods used to define it in a given test system. Both qualitative and quantitative endpoints have been used to assess cytotoxicity. The ideal endpoint is one that is sensitive only to serious insults and relevant by any mode of insult (Stark et al., 1986; Eisenbrand et al., 2002). To assess the potential cytotoxicity of chemical substances several endpoints have been developed that measure different biological endpoints, these include: x Cell death: This is a non-specific endpoint providing no opportunity to establish mechanisms or the reversibility of the cell damage. Chemicals can produce cellular lethality by either direct damage to the structural components of the cell or indirect interference with the normal physiology and metabolism of the cell. This is through

32 Chapter 2

impaired protein synthesis, respiration, ion exchange, and DNA synthesis capabilities. x Cell viability: This is an endpoint measuring living cells. Measurements to assess cell viability are based on changes in membrane permeability through dye exclusion of viable cells, and membrane leakage of dead or damaged cells (Barile 1994a). Trypan blue and nigrosin are two types of vital stains that can measure membrane damaged cells that have lost their ability for cell exclusion. Neutral red dye is a supravital stain that can diffuse through the plasma membrane and concentrate in lysosomes of living cells (Borenfreund and Babich 1992). x Membrane leakage: This is an endpoint measuring cellular damage through the release of cellular constituents or products. An example is the leakage of soluble cellular cytosolic enzymes such as lactate dehydrogenase (LDH), into the cytoplasm that can be used to quantitate lethality (Allen and Rushton 1994). Other examples include the release of radiolabeled compound, proteins or DNA. Chromium-51 (51Cr) is a common radiolabel used for the study of cell damage (Eisenbrand et al., 2002). x Cell growth and reproduction: These endpoints are widely used for assessing the viability of cells in culture. Examples include determining cloning efficiency, measurement of reductions in DNA synthesis, mitogenicity using the mitotic index and cell-cycle kinetics (Wilson 2000). x Cell morphology: morphologic changes in cell membrane associated with toxicity such as changes in size, shape or integrity of the cell membrane and other cellular components (Barile 1994a). x Cell function: This is an endpoint used to measure thermodynamic and metabolic function. Adenosine Triphosphate (ATP) can be used as an indicator of cytotoxicity because it is the primary energy source at the cellular level (Barile 1997). Inhibition of metabolic cooperation, co-factor depletion (e.g. ATP) and impairment of mitochondrial function (e.g. MTT, MTS and XTT tetrazolium salt assays).

33 Chapter 2

x Cell adhesion: attachment to culture surface, detachment from culture surface, cell- cell adhesion (Wilson 2000).

Advances in cell culture techniques and the continuous knowledge accumulated about cellular and molecular functioning, gave greater understanding of toxicity mechanisms, hence expanding the development of new tests, such as the research on apoptosis (Zucco et al., 2004). Research on apoptosis has immensely changed the knowledge of mechanisms involved in cell death, leading to the development of mechanistically based end points (Eisenbrand et al., 2002). Many morphological and biological changes occur at the level of the nucleus, of the cellular membrane, of specific proteases and of DNA and the identification of these endpoints may enlighten specific mechanisms of action (Zucco et al., 2004). In addition, rapid progress in genomics technology (DNA sequencing) and proteomics (the study of proteins expressed by a genome, tissue or cell) will be powerful tools for the improvement of toxicological testing (Eisenbrand et al., 2002)

2.2.2.3 Mechanism of cellular toxicity

There are three types of mechanisms of toxicity. They are basal cytotoxicity, organ- specific cytoxicity and organizational toxicity (Barile 1994a).

34 Chapter 2

Cytotoxic Classification

Basal Cytotoxicity

Organ-specific Organizational Cytotoxity Cytotoxicity

Figure 2-4 Areas of toxic mechanisms based on levels of organisation (adapted from Ekwall et al., 1988; Barile, 1994).

2.2.2.3.1 Basal cytotoxicity

Over the last two decades there has been considerable interest in using basal cytotoxicity data to predict the acute effects of compounds in vivo. If a compound is acutely toxic, it is anticipated that in most cases this will reflect an insult to the intrinsic functions of cells. Numerous mammalian in vitro toxicity studies have been devised in the past three decades (Bondesson et al., 1989). Progress for health risk assessment was based on the rationale that inter-species extrapolation could be avoided by using cell lines of human origin. This approach is based on the concept of “Basal Cytotoxicity” formalized by Ekwall (Ekwall 1983). Basal cytotoxicity depends on the fact that the toxic action of chemicals will cause interference with basal cell functions, which in turn will influence specific functions. The basal cell toxicity concept assumes that in most cases, target organ toxicity is basal cell toxicity distributed to the target organ. Basal cell toxicity is defined as adverse effect of such structures and functions that are essential for cell survival and proliferation (Seibert et al., 1996) and these can be tested according to Ekwall with undifferentiated cells (Ekwall 1995). These effects may involve structures and functions common to all cells in the body, including the integrity of membranes and the cytoskeleton, cellular metabolism for example as a function of its mitochondrial activity, protein or DNA synthesis, the synthesis and degradation or release of cellular

35 Chapter 2

constituents or products, ion regulation, cell division or other cellular processes, ribosomes and lysosomes. Basal cell functions generally support organ-specific cell functions. Thus, chemicals capable of affecting basal cell activities are also likely to interfere with specialized functions. Studies aimed at understanding basal cytotoxic phenomena can be designed with either primary culture or continuous cell lines.

Based on this concept, in vitro tests can be used as screens and as potential replacements for in vivo toxicity testing, and in specific acute lethal toxicity (Blaauboer 2002b). An overview of the available test systems can be found in the reports organized by the European Centre for the Validation of Alternatives to Animal Testing (ECVAM) published in Alternatives for Laboratory Animals. Results of an international workshop organized by the Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM) and the National Toxicology Program (NTP) Interagency Centre for the Evaluation of Alternative Toxicological Methods (NICEATM) published in August, 2001 concluded that presently in vitro data could be used for determining the starting dose in stepwise animal procedures to determine acute toxicity value such as the “Up and Down Procedure” (ICCVAM 2001b).

The ECVAM workshop conducted in 1996, divides cytotoxicity into three general types. Chemicals can have three main toxic effects at the cellular level these are: Basal Level where toxicity occurs at all levels, Metabolic Level (target cells) and Functional Level (whole organism). All three types of effect can result in acute toxicity in vivo hence any non-animal test must take all these possibilities into account. Therefore, there may be inherent benefits and weaknesses of some tests. One potential criticism with basal cytotoxicity is that it may not be sensitive enough to detect function level effects especially if these occur at concentrations lower than the in vitro basal cytotoxicity level (ICCVAM 2001a). However, even animal toxicity tests have these problems when it comes to animal/human extrapolations.

Through the literature it is apparent that a large number of tests for basal cytotoxicity has been developed, but that insufficient attention has been paid to how the data they provide can be applied, particularly as a means of making regulatory decisions that help in labelling and ranking of chemicals for the purpose of manufacturers, workers and

36 Chapter 2

consumers health (Seibert et al., 1996). There are also procedural limitations in converting a concentration of a toxicant in cell culture media with exposure to humans. There are certain disadvantages to basal cytotoxicity testing in the face that many continuous cell lines are not capable of metabolizing chemicals. As a result, chemicals which require biotransformation for their toxicity will not show in vitro toxicity (Barile 1994a). At the moment the use of organ-specific primary cultures that can retain a high level of enzymatic activities or the incorporation of PBPK modelling and specific information on biokinetics of the chemical might help overcome this. Whereas basal cytotoxicity can be studied with undifferentiated finite or continuous cell lines, organ- specific cytotoxicity must be studied with primary cultures with well-differentiated cells from different organs, and organizational toxicity maybe studied indirectly in cell cultures by examining the products of cell metabolism.

2.3 Validating In Vitro Tests and Regulatory Acceptance

Validation of in vitro assays is necessary to demonstrate the relevance, reliability and predictability of new methodology, before gaining acceptance and usage as replacement for traditional in vivo methods (Balls and Karcher 1995). In order to accurately assess the specificity and sensitivity of an assay, a wide variety of compounds with different mechanisms of action must be tested using a battery of in vitro tests with different endpoints. General points to be considered in a validation are the reproducibility and predictive ability of the method (Balls and Fentem 1992). The reproducibility of test results (reliability) of the test method, between laboratories and over time, is established. The purpose of an alternative method needs to be identified, whether for the intended application for a specific toxicological endpoint, for e.g. acute toxicity (Barile et al., 1994b; Balls and Fentem 1999).

One of the challenges in determining the feasibility of using in vitro methods is the difficulty in comparing results of reductionist, mechanism based assays with those from apical (whole) in vivo tests. The solution could be in establishing a battery of in vitro

37 Chapter 2

tests but it will take some time in validating each test (Daston and McNamee 2005). There is a need for an increase in research, validation and acceptance of alternatives is on an increase especially with a recent legislation adopted in Europe calling for a ban of most animal testing in cosmetic products by 2009 and a ban on all animal testing by 2013. The testing and marketing (sale) of cosmetics has been regulated by European legislation since 1976. The legislation (Council Directive 76/768/EEC) was amended for the 7th time in 2003 and Directive 2003/15/EC was adopted (EU Directive, 2003).

The request for the development and validation of alternatives to animals depends very much on the actual products being tested. The public demand for alternatives depends on their perception of the risk, the perceived personal impact and their degree of involvement as seen in Figure 2-5 (Garthoff 2004). Re q uest for alternatives

Pharmaceuticals

Chemicals/agrochemicals

Cosmetics

Figure 2-5 Schematic figure representing the request for alternatives

Validation studies have been set up in various ways by groups of companies, by industry associations, by animal welfare organisation, by national government agencies,

38 Chapter 2

and by international agencies such as ECVAM, ICCVAM, and the OECD (Balls and Fentem 1999). Several organisations around the world have begun to pursue, form, and validate in vitro toxicity systems. They are distributed around three main continents:

In the USA: x CAAT: John Hopkins Centre for Alternatives to Animal Testing, and it was supported by the Cosmetic, Toiletry, and Fragrance Association, 1981. They were established to find new methods to replace the use of laboratory animals in experiments, reduce the number of animals tested, and refine necessary tests to eliminate pain and distress. They also continue to manage an international online clearinghouse of alternatives resources (Altweb: http://altweb.jhsph.edu). x University of California Alternatives Centre was established at UC-Davis in 1990. x ICCVAM: the US government established the Interagency Coordinating Committee on the Validation of Alternative Methods, 1994 (Schechtman 2002). x NICEATM: The National Toxicology Program Interagency Centre for the Evaluation of Alternative Toxicological Methods was established to provide support to ICCVAM, 1998.

In Europe: x FRAME: The fund for the replacement of animals in medical experiments was established in 1982. Its role is to develop and validate alternative in vitro test systems. It also manages an alternative test validation scheme, in which sets of compounds are coded and supplied to research groups to validate new test methods blindly at a number of laboratories throughout the world. x MEIC: Multicentre Evaluation of In vitro Cytotoxicity. The Scandinavian Society for Cell Toxicology organizes it; 1989. MEIC is a 5-year program to validate in vitro tests for general toxicity. x EDIT: the Evaluation-guided Development of New In vitro Test Batteries programme was initiated in 1998 as a 6-year effort by international cytotoxicology laboratories. The aim was to develop and validate new in vitro tests relevant to

39 Chapter 2

toxicity and toxicokinetics, for testing acute and chronic systemic toxicity (Clemedson et al., 2002) x ECVAM: European Centre for Validation of Alternative Methods, 1993. x NCA: The Netherlands Centre for Alternatives to Animal Use is established as a national information centre on alternatives in 1994. x ECOPA- Organisation: European Consensus Platform on Alternatives was founded in 2002 in Brussels by national platforms from 10 European states (i.e. Austria, Belgium, Czech Rep., Finland, Germany, Italy, Netherlands, Spain, Switzerland, and UK). It represents the different parties interested in fostering research, development and implementation of alternatives to animal experiments. (www.ecopa.vub.ac.be). x ZEBET: the centre for the Documentation and Evaluation of Alternative Methods to Animal Experiments was established in Germany, 1989. x R Research Foundation Switzerland: was established to support alternatives research, 1997. x ECEAE: European Coalition to End Animal Experiments was founded in 1993 to campaign to end the use of all animal experiments. It works with EU institutions to ensure that laboratory animals are high on the political agenda. ECEAE is a member of the International Council for Animal Protection (ICAPO) (www.eceae.org).

In Asia: x JSAAE: Japanese Society for Alternatives to Animal Experiments was founded in 1984. JSAAE is the centre in Japan for scientific research in the application of alternatives to laboratory animal testing. (wwwsoc.nii.ac.jp/jsaae/indexa-e.html).

In Australia: x ANZCCART: Australian and New Zealand Council for the Care of Animals in Research and Teaching Ltd. The council was established in 1987. ANZCCART aims to provide consensus on ethical, social and scientific issues relating to the use of animals in research and teaching (http://www.adelaide.edu.au/ANZCCART).

40 Chapter 2

x NSW Agriculture Animal Welfare Unit: The Animal Welfare Unit monitors and ensures the full compliance with the Prevention of Cruelty to Animals Act 1979, the Animal Research Act 1985 and the Exhibited Animals Protection Act 1986. Other state authorities also exist.

2.3.1 The MEIC/EDIT approach

The MEIC program was established by the Scandinavian Society for Cell Toxicology in 1989. The intention of the program was to investigate the relevance of in vitro test results for predicting the acute toxic action of chemicals in humans directly rather than animals (ICCVAM 2001b). The MEIC study concentrated on the predictivity of in vitro results compared with in vivo response (Gad 2000). 50 reference chemicals were selected to represent the different classes of chemicals, based on the availability of good data on acute toxicity such as lethal blood/serum concentrations (LC) in humans or oral

LD50 values in rats and mice.

MEIC was a voluntary effort involving 96 international laboratories that evaluated the relevance and reliability of in vitro cytotoxicity tests originally developed as alternatives to or supplements for animal tests for acute systemic toxicity, chronic systemic toxicity, organ toxicity, skin irritancy or other forms of general toxicity. All in vitro toxicity test results collected during MEIC are now found on a searchable database in the Cytotoxicology Laboratory, Uppsala (CTLU) website (www.ctlu.se) (NTP and NICEATM 2000). The majority of the assays were based on measurement of effects on cell viability and/or cell growth. The test results submitted were analysed statistically and the predictability of in vivo acute data toxicity from the in vitro IC50 (50% inhibitory concentration) data was assessed against lethal blood concentrations in humans (Ekwall 1998). By 1996, 39 laboratories had tested the first 30 reference chemicals in 82 in vitro assays, while the last 20 chemicals were tested in 67 in vitro assays. The MEIC concluded that the battery in Table 2-3 can be used directly as surrogate of the LD50 test. But it only predicts lethal blood concentrations, not lethal doses, therefore at the moment it is not a counterpart of the animal LD50 test. Thus, it must be supplemented with data on gut absorption as well as volume of distribution

41 Chapter 2

(Vd) of chemicals. Currently a second voluntary Multicentre program has been initiated by the CTLU. The Evaluation-Guided Development of In vitro Tests (EDIT) program aims to provide a full replacement of the animal acute toxicity tests. To provide data on all the missing tests still needed to add to the current battery. Such as, assays on the accumulation of chemicals in cells (test of Vd), passage across the intestinal blood-brain barriers, and biotransformation to more toxic metabolites (Ekwall et al., 2000).

Table 2-3 MEIC battery of in vitro tests Exposure time Cell line Protein Content 24 hours Hep G2 ATP Content 24 hours HL-60 Inhibition of elongation of cells 24 hours Chang Liver cells (a subset of HeLa cells) pH change 168 hours Chang Liver cells

The cell battery identified by the MEIC team is not yet valid to be used for regulatory purposes. An improvement in the predictive capability of this proposed battery is needed through taking into account non-basal cytotoxicity as 20% of chemicals assayed by Ekwall using HeLa cells did not fit into the basal cytotoxicity concept (Ekwall 1999a).

It is important to note that US regulatory agencies have confirmed that if there was a validated in vitro cytotoxicity test, which could accurately predict the approximate rodent LD50 value in vivo, its implementation would result in a significant reduction in animal use. Thus the primary focus for future research in the in vitro field is to identify and evaluate candidate in vitro cytotoxicity tests that could possibly serve as reduction and replacement alternatives for current rodent acute oral toxicity tests for determining

LD50 values (ICCVAM 2001a).

42 Chapter 2

2.3.2 The ZEBET approach

The Halle and Gores Registry of Cytotoxicity (RC) contains 1912 individual IC50 values for 347 chemicals. The RC is a database of acute oral LD50 data from rats and mice as well as IC50 values for chemicals and drugs from in vitro cytotoxicity assays. It is run by ZEBET, a German based centre. The word ZEBET is the acronym for Zentralstelle zur Erfassung und Bewertung von Erstaz- und Ergaenzungsmethoden zum Tierversuch (National Centre for Documentation and Evaluation of Alternative Methods to Animal Experiments). Scientists at the ZEBET centre have proposed a strategy to reduce the number of animals required for acute oral toxicity testing. The strategy describes how basal cytotoxicity data could be used to predict a starting dose for an in vivo lethality assay. Report findings of an initial study conducted to assess the feasibility of applying the standard regression between mean IC50 values and acute oral LD50 data taken from the Register of Cytotoxicity (RC) database, which can then be used to determine the in vivo starting dose. A prediction model was developed that can consequently be used as a mathematical model for prediction of rodent oral LD50 values from basal cytotoxicity.

The IC50 values were obtained from multiple reports in the literature, averaged for each chemical and then paired with acute oral LD50 values for the rat and with intraperitoneal

(i.p) LD50 values for the mouse, obtained from the National Institute for Occupational Safety and Health (NIOSH) Registry of Toxic Effects of Chemical Substances (RTECS) (ICCVAM 2001b; Worth and Balls 2002).

The accumulated results of many cytotoxicity studies and the ZEBET/MEIC initiatives suggest that, it may be possible to obtain reasonable estimates of LD50 values for regulatory decisions. One or more reasonably predictive assay of the LD50 can be applied to test the considerable number of chemicals on which such risk assessment data are needed (e.g. for high production volume (HPV) chemicals), to make a significant reduction/replacement of animal usage. Certain agencies have been established for a formal process in validation and regulatory acceptance of toxicological test methods such as ICCVAM in the US and EVCAM in Europe.

43 Chapter 2

2.3.3 ICCVAM

ICCVAM has developed a rigorous, objective, and peer reviewed process to determine whether proposed new assays are suitable alternatives to existing ones. The review process is administered through NICEATM (Daston and McNamee 2005). An ICCVAM/NICEATM International Workshop on In vitro Methods for Assessing Acute Systemic Toxicity was held in October 2000. Workgroups evaluated the current validation status of in vitro cytotoxicity methods for assessing various toxicity endpoints (acute toxicity, metabolism, toxicokinetics and assessing organ specific toxicity). The workgroups reviewed the status of in vitro methods for assessing various endpoints, and recommended future directions (ICCVAM 2001a; ICCVAM 2001b). ICCVAM and NICETAM have conducted a validation review activity on the Murine Lymph Node Assay (LLNA) used to assess allergic contact dermatitis of chemicals. The method was later accepted as a validated method by the OECD member countries and OECD Test Guideline 429 - Skin Sensitisation: LLNA was adopted on the 24th of April 2002. CORROSITEX® an in vitro method for assessing the dermal corrosivity potential of chemicals has been reviewed and accepted by regulatory agencies in March 2000. FETAX is proposed as a screening method to evaluate the developmental toxicity potential of chemicals. Further research and development is still being undertaken for this method. A list of alternative toxicity tests is found in Table 2-4.

2.3.4 ECVAM

ECVAM plays a major role in validating alternative methods that reduce, refine or replace animal experiments. It also provides a forum of scientific discussion through its workshops, at which experts from academia, industry and other communities discuss the state of the art in a particular field and make recommendations for further progress (Worth and Balls 2002). ECVAM was created by the European Parliament in October 1991 to address a requirement in the protection of Laboratory Animals Directive (86/609/EEIC) on the protection of animals used for experimental and other scientific purposes. ECVAM is required to actively support the development, validation and

44 Chapter 2

acceptance of alternative methods to animal testing (Daston and McNamee 2005). This is accomplished through research and test development. ECVAM’s criteria for development and validation have been defined in a number of documents (Balls and Karcher 1995; Balls and Fentem 1999). There are no major differences in the criteria for validation established by ECVAM as to those of the OECD (OECD 1996) and ICCVAM (NIH 1999).

Replacement tests are now available and accepted for assessing skin corrosion, phototoxicity, and skin permeability; refinement or reduction methods have been developed for predicting acute oral toxicity and skin sensitisation; promising methods for assessing skin and eye irritation and developmental toxicity are now being validated or further developed (ICCVAM 2001b; OECD 2004a; OECD 2004b; OECD 2004c; OECD 2004d). Examples of successfully validated alternatives (see Table 2-4) are the 3T3 NRU phototoxicity test and the Rat Skin Transcutaneous Electrical Resistance (TER) and EPISKINTM tests (Balls and Fentem 1999; Spielmann and Liebsch 2001).

2.4 Using InVitro Cytotoxicity Tests in Risk Assessments

A prerequisite for the successful application of in vitro approaches is the availability of appropriate validation test systems (OECD 1996; NIH 1999; Eisenbrand et al., 2002). Validation studies are conducted principally to provide objective information on new tests, to confirm that they are robust and transferable between laboratories and to show that the data generated can be relied on for decision-making processes (Eisenbrand et al., 2002).

45 Chapter 2

Table 2-4 List of alternative toxicity tests and their current status

Toxicity endpoints Alternative test methods Current status of tests

EU OECD US Acute systemic toxicity Neutral Red cytotoxicity test - - - MEIC/EDIT - - - Modelled cell cultures - - - Skin corrosion CORROSITEXTM R - A Transcutaneous Electrical Resistance (TER) assay A V V EpiskinTM A U V EpiDermTM A U V Skin irritancy EpiskinTM U - V EpiDermTM U - V Skin integrity function test (SIFT) N/A - N/A Skin sensitivity DEREK computer model U - - Protein binding assays U - - Dendritic cell line U - - Ocular toxicity EYETEX R N/A N/A EpiOcular assay - N/A U Neutral red release assay U - - Isolated rabbit eye test U - - Isolated chicken eye test U - - HET-CAM test U - - BCOP test U - - Genotoxicity/carcinogenicity In vitro Syrian hamster embryo (SHE) cell R - + Transformation assay U - N/A In vitro micronucleus assay U - N/A Transformed human cell line U - N/A

Repeated dose (28&90 day) In vitro cell lines - - - studies Reproductive/development FETAX R - A screening/teratogenicity Whole embryo culture V - - Embryonic stem cells V - - Endocrine disruptors In vitro MCF-7 Focus assay using human U - - breast cancer cells

Adapted from Botham et al., 2001; BUAV, 2001 and Langley et al., 2005.

All alternative tests described are still under ongoing research and development activity. Key: - no ongoing activity; R rejected after validation; A accepted after validation; V validated; U under review; N/A: No available information.

46 Chapter 2

As part of a battery of tests, in vitro tests can be applied as screening protocols in the risk assessment process. Traditionally, in vitro studies have most frequently been applied in the hazard identification stage. However, over the last 20 years there has been an increased recognition that in vitro studies can also be of value at the hazard characterisation stage (see Figure 2-1) (Holme and Dybing 2002). The use of in vitro tests in the different parts of the risk-assessment process has increased. Data generated by in vitro tests can be extrapolated to toxicological effects, using prediction models (PMs), which account for differences in absorption, distribution, metabolism and excretion (ADME).

While acute toxicity testing was traditionally conducted using animals, many studies published in recent years have shown strong correlations between cytotoxicity in vitro and in vivo acute toxicity (Ekwall et al., 1989; Ekwall et al., 1991; Shrivastava et al., 1992; Garle et al., 1994; Barile et al., 1994b; Ekwall 1998; Clemedson and Ekwall 1999; Ekwall 1999a; Dierickx 2000; NTP and NICEATM 2000; Dierickx 2003).

Basal cytotoxicity data at present is still not considered part of the regulatory agenda needed to screen toxic substances before acceptance by industrial, scientific, agricultural and manufacturing groups. But results from current initiatives and major international validation programs, have indicated the usefulness of basal cytotoxicity data in future risk assessments. Many studies such as those conducted by MEIC and ZEBET have indicated the usefulness of in vitro basal cytotoxicity data for the prediction of human toxicity. The MEIC programme found that for 50 reference chemicals rat and mouse

LD50 values predicted human acute lethal doses rather poorly (R2 = 0.61 and R2 = 0.65 respectively). To investigate the relevance of in vitro findings, IC50 values from in vitro assays were correlated with peak human lethal blood concentrations (LCs) (R2 = 0.79), when information on blood brain barrier (BBB) penetration was incorporated into the battery of tests it helped increase predictability (R2 = 0.83). Information on the passage across the BBB can be predicted from the chemical formula and/or physico-chemical properties from appropriate in vitro tests. MEIC also determined that additional in vitro tests need to be included to improve prediction of human acute systemic toxicity to determine key kinetic events (e.g. biotransformation, absorption in the gut and passage through biological barriers, distribution volume, protein binding) as well as any organ

47 Chapter 2

specific toxicity (Clemedson et al., 2002). Several studies have indicated that IC50’s using cell lines are similar regardless of the origin of the cultures but human cell lines have been seen as more suitable for detecting cytotoxicity than cells from animal origin (Ekwall et al., 1989; Ekwall 1998; Clemedson and Ekwall 1999; Ekwall 1999a; Clemedson et al., 2002).

Basal cytotoxicity data is more suited for risk assessment than most of the toxicity data generated in conventional protocols. For example, by using human cells, issues like species differences can be overcome and a direct comparison can be made with known human toxic blood and tissue concentrations. The results of the MEIC study demonstrated that human basal cytotoxicity tests are relevant for predicting human acute toxicity of chemicals (Gribaldo et al., 2005). In addition, the results showed that specific mechanisms exist, which could only be measured by using organotypic in vitro toxicity tests. In addition, the results also showed that the inclusion of various biokinetic events (such as biotransformation and passage through biological barriers) could improve the modelling of human toxicity. The EDIT program (Section 2.2.3) was initiated as a follow up of the MEIC study in 1998, with the aim of developing and validating new in vitro tests relevant to toxicity and toxicokinetics, for the incorporation into a battery of in vitro tests of acute and chronic systemic toxicity (Ekwall et al., 1999b; Clemedson et al., 2002; Clemedson et al., 2003).

In vitro and computational methods can be combined in stepwise or decision trees strategies tailored to each type of toxicity. Stepwise testing has already been accepted and used within the USA and the European Union. Stepwise strategy using non animal methods, testing strategies from quick and simple screening methods through tests specific to toxic mechanisms, to more sophisticated in vitro assays, where needed. In vitro metabolism studies in combination with computer modelling of chemical absorption, distribution and excretion permit extrapolation of in vitro results to the whole body situation (BUAV 2001). A decision point about a chemical can be made at any stage. The OECD, which is responsible for the development and the regulatory acceptance of test guidelines, approved four new in vitro test guidelines in 2002. They included tests on percutaneous absorption, phototoxicity, and two test methods for skin corrosion (Louekari 2004). Currently a collaborative effort by ICCVAM and ECVAM

48 Chapter 2

is underway for the validation study on the use of murine 3T3 cells and human keratinocytes with the neutral red assay for predicting acute systemic toxicity (NTP and NICEATM 2003; NTP and NICEATM 2003b)

2.5 Selected in vitro cytotoxicity assays

A large number of in vitro test methods exist for determining the basal cytotoxicity of chemicals. These methods use various types of mammalian cells that are exposed to a concentration range of the test compound for certain periods of time. They are used to determine a critical concentration of a test compound at a specific target site and specific toxic endpoint by direct exposure of the cells to the test compound. Many in vitro models have been developed which are based on either a colorimetric or bioluminescence reaction. In vitro assays are currently available to measure a variety of different markers that can indicate the number of dead cells (cytotoxicity assay), the number of live cells (viability assay), the total number of cells or the mechanism of cell death (e.g., apoptosis). There are a variety of cytotoxicity assays that measure different cytotoxicity endpoints including the neutral red uptake assay which measures lysosomal activity through the uptake a neutral red (NR) dye; Kenacid Blue assay which measures total cell protein though the uptake of Kenacid dye; MTS and MTT assay which measure cell viability through mitochondrial oxidation.

2.5.1 The MTS cytotoxicity assay

The MTS (3-(4, 5-dimethylthiazol–2-yl)-5-(3-carboxymethoxyphenyl)-2-(4- sulfophenyl)-2H-tetrazolium) assay measures the metabolic activity of mitochondrial dehydrogenases, which are produced by living cells (Promega 2001a). Tetrazolium salts have been used for many years to distinguish living cells from dead ones. They are reduced to formazan by the cytochrome system of viable cells, and the colour developed is a direct measure of the viability of the culture (Malich et al., 1997). The reduction of

49 Chapter 2

the MTS tetrazolium salt to the formazan product is as a result of the action of dehydrogenase enzymes generating reducing equivalents such as NADH or NADPH. NADH can then transfer its electrons to an electron transfer reagent such Phenazine methosulfate (PMS) resulting in reduction of these compounds. The reduced ETRs, in turn, can directly interact with and reduce the MTS tetrazolium compound producing the deeply coloured, aqueous soluble formazan product that can be assayed colorimetrically as seen in Figure 2-6.

The amount of formazan product was time-dependent and proportional to the number of viable cells. One of the advantages of this formazan product is its solubility in culture medium. In brief the advantages of the MTS assay: x Non-radioactive: requires no scintillation cocktail or radioactive waste disposal. x Fast: perform the assay in a 96 well plate with no washing or cell harvesting. Also eliminates the solubilisation steps because the MTS formazan product is soluble in tissue culture medium unlike its predecessor MTT (Riss et al., 2003). x Safe: requires no volatile organic solvent to solubilise the formazan product. x Convenient: supplied as ready-to-use stable, frozen sterile solutions. x Flexible: plates can be read and returned to the incubator for further colour development, unlike MTT (Promega 2001a).

50 Chapter 2

dehydrogenase Substrate Product

NAD NADH

ETR reduced ETR O- O O- S O O S OH O OH

O O N N O S N NH CH O N 3 S CH N N 3 H N N N CH3 CH3 MTS Formazan

Figure 2-6 Schematic diagram showing cellular metabolism resulting in the conversion of MTS to Formazan

The efficiency of MTS has been demonstrated by a number of researchers (Malich et al., 1997; Da Costa et al., 1999; Hayes and Markovic 2000; Zarei and Markovic 2000; Promega 2001a; Eirheim et al., 2004).

2.5.2 The NRU cytotoxicity assay

The neutral red (NR) cytotoxicity assay is a cell survival/viability chemosensitivity assay, based on the ability of viable cells to incorporate and bind neutral red, a supravital dye (Basic Red 5, Toluylene red). NR readily penetrates cell membranes of viable cells by non-ionic diffusion, accumulating intracellularly in the lysosomes. Alterations of the cell surface or the sensitive lysosomal membrane due to the action of chemicals result in a decreased uptake and binding of NR. Damage to the cell surface or lysosomal membranes leads to lysosomal fragility and ultimately decreased uptake and binding of neutral red. It is important to note that chemicals which have a direct effect on lysosomes (e.g. chlorine sulfate) may exhibit greater toxicity with NR (Barile

51 Chapter 2

1994a). Neutral red uptake measured by extraction and spectrophotometric absorption, has been used as a reliable, reproducible and inexpensive in vitro assay for cell viability (Borenfreund and Puerner 1985).

The neutral red uptake assay has an extended use among researchers, and it has been shown to be a reliable, and sensitive assay (Spielmann et al., 1999; Dierickx 2000; ICCVAM 2001a; Dierickx and Scheers 2002; Putnam et al., 2002; Jirova et al., 2003; Eirheim et al., 2004). Currently two proposed in vitro neutral red uptake (NRU) assays using mouse fibroblast (BALB/c) 3T3 cells and normal human keratinocytes (NHK) are under validation (NTP and NICEATM 2003)

2.5.3 The LDH cytotoxicity assay

Lactate dehydrogenase (LDH) is a stable cytosolic enzyme present within all mammalian cells. The normal plasma membrane is impermeable to LDH, but damage to the cell membrane results in a change in membrane permeability and subsequently the leakage of LDH into the extracellular fluid (Allen and Rushton 1994). Release of LDH in culture supernatants is measured with a 30-minute coupled enzymatic assay, which results in the conversion of a tetrazolium salt (INT) into a red formazan product. The resulting formazan product absorbs maximally at 492nm and can be measured quantitatively using a micro-plate reader. The release of LDH into culture supernatant correlated with the amount of cell death and membrane damage proving an accurate measure of the cellular toxicity induced by the test substance (Promega 2004). The enzymatic reactions associated with the assay are found in Section 5.2.8 (Figure 5-6)

Advantage of the LDH endpoint is that it quantitates enzyme leakage from cells that have been lysed, in addition to those that are damaged with leaky membranes (Gad 2000). The amount of LDH that the dead or damaged cells release in the culture medium can be assayed using sodium pyruvate as substrate and nicotine adenine dinucleotide (NADH) as a cofactor. Lysing the remaining viable cells and measuring the levels of enzymes subsequently released can also assess the LDH content of the

52 Chapter 2

surviving cells. Using this technique, LDH leakage can be presented as the percentage released relative to the total amount of enzyme in the culture.

The LDH assay has proven to be a popular and reliable enzymatic assay and has an extensive use among researchers (Allen and Rushton 1994; Yang and Acosta 1994; Niederau et al., 1995; McKarns et al., 1997; Da Costa et al., 1999; Putnam et al., 2002; Eirheim et al., 2004; Issa et al., 2004; Promega 2004; Baba et al., 2005; Zhu et al., 2005).

2.5.4 The ATP cytotoxicity assay

The Adenosine Triphosphate (ATP) cytotoxicity assay is a luminescent assay that determines the number of viable cells in culture based on the quantitation of the ATP present, which signals the presence of metabolically active cells (Promega 2001b). It generates a “glow type” luminescent signal, produced by a luciferase reaction. During the luciferase reaction a mono-oxygenation of luciferin is catalysed by luciferase in the presence of Mg2+, ATP and molecular oxygen (Figure 2-7). Within minutes after a loss of membrane integrity, cells lose the ability to synthesise ATP and endogenous ATPases destroy any remaining ATP; thus the level of ATP falls precipitously (Riss et al., 2003)

O Recombinant Firefly - - OH S N OH Luciferase O S N O + ATP + O2 + AMP + PPi + CO2 + Light N S Mg2+ N S

Beetle Luciferin Oxyluciferin

Figure 2-7 The Luciferase reaction in the ATP assay

ATP can be used as an indicator of cytotoxicity because it is the primary energy source at the cellular level. In order for the cell to function optimally, it must maintain an intricate balance between energy production and consumption.

53 Chapter 2

One of the major advantages of this assay is its sensitivity, as it requires only a small number of cells, below the detection limits of most colorimetric and fluorometric assays with a high signal to background ratio. Researchers have extensively used the ATP assay (Petty et al., 1995; Eirheim et al., 2004; Mueller et al., 2004; O'Neill et al., 2004)

2.6 Risk Assessment of Chemical Mixtures

Occupational exposure limit standards cover situations where there is exposure to a single chemical. This rarely occurs in most exposures. In the majority of cases, people are exposed to more than one chemical at work and in outside interests (such as hobbies). Even lifestyle activities (alcohol, smoking) can produce exposure to a range of chemicals. Toxic effects can arise from exposure to combinations of chemicals. Interaction between chemicals following exposure to mixtures is still a new area in toxicology and filled with uncertainty (Malich et al., 1998). Approximately 95% of the resources of toxicology are still devoted to studies of single chemicals (Yang 1994b; Cassee et al., 1998).

54 Chapter 2

2.6.1 Basic concepts of chemical mixture toxicology

Three basic concepts for the description of toxicological action between constituents in a mixture have been defined by Loewe and Muischnek (1926) and Bliss (1939), and are still valid till today. An extensive compilation of how these concepts have been developed and used in different applications and case studies is contained in books edited by Yang (1994) and Greco et al., (1995).

2.6.1.1 Simple similar action (Concentration Addition)

Concentration Addition (dose addition) was first introduced by Loewe and Muischnek in 1926 (Rajapakse et al., 2001). Concentration addition (CA) assumes that chemicals interact with the same target contributing to a joint effect in proportion to their dose. This is a non interactive process, which means that the chemicals in the mixture do not affect the toxicity of one another (Mumtaz et al., 1994; Cassee et al., 1998). From a strict mechanistic point of view it is applicable to these toxicants that competitively interact with an identical molecular binding site (Altenburger et al., 2000; Faust et al., 2003). But from a broader phenomenological point of view it can be seen as different substances are able to cause a common toxicological response (e.g. death, inhibition of reproduction), which may apply to almost all chemicals (Berenbaum 1985; Berenbaum 1989). This has led concentration addition to be suggested as the “general solution” to the problem of calculating an expected quantitative effect for any combination of chemicals, without reference to their mechanism of action (Berenbaum 1985; Backhaus et al., 2000a). From a regulatory perspective, concentration addition is commonly used as it is seen as a pragmatic and precautionary default assumption (Faust et al., 2003)

There is a consensus that concentration addition is suitable and valid concept for the prediction of mixture effects of similarly acting agents (EPA 1986; ATSDR 2004). The concentration addition concept rests on the assumption that all chemicals are expected to act in the same way, by the same mechanism and only differ in their potencies. Therefore chemicals behave as concentrations or dilutions of one another (Mumtaz et

55 Chapter 2

al., 1994). Therefore, one chemical can be replaced by an equal fraction of an equally effective concentration of another, without diminishing the overall mixture effect (Silva et al., 2002). These assumptions have been used as a basis for the calculation of the toxicity of a mixture by a simple algebraic process called the “hazard index” (EPA 1986). The concentration addition concept also serves as a basis for the use of “toxic equivalency factors” , and is used to describe the combined toxicity of isomers or structural analogues, such as the US EPA’s approach for dioxins and PCB’s (Cassee et al., 1998; Safe 1998; EPA 2000). The additivity effect is described mathematically using the summation of the doses of the individual compounds in a mixture after adjustment for the different potencies for a single exposure (Cassee et al., 1998). Supporting evidence and extensive use of this concept have come from a number of mixture studies with different groups of toxicants, and on different types of cell lines and mainly aquatic organisms (Altenburger et al., 2000; Backhaus et al., 2000b; Faust et al., 2001; Cleuvers 2004).

2.6.1.2 Simple dissimilar action (Response Addition)

Response addition also known as independent action (IA) was first developed by Bliss in 1939 (Cassee et al., 1998). IA assumes independence of toxic action across the components. The chemicals work in diverse mechanisms and the additivity expectation is an effect multiplication (Faust et al., 2003); the fractional effects of individual mixture components (e.g. 50% response) are expected to be independent from each other in a probabilistic manner. By definition, IA is determined by summing the response of the organisms or cell lines to each toxicant in a mixture (Cassee et al., 1998). This though has led to more complicated calculations than the concentration addition concept, because of the difference in the modes of action of chemicals, therefore the susceptibility of an organism to one of the components in a mixture may or may not be correlated with its susceptibility to the other (Mumtaz et al., 1994). Independent action is generally acknowledged as being theoretically important, but many scientists are skeptical about its practical relevance, at least on the level of entire organisms or populations (Faust et al., 2003). IA has more commonly been used in human pharmacology in assessing the joint effects of dissimilarly acting drugs and toxic

56 Chapter 2

agents (Posh 1993; Payne et al., 2000). However, a study conducted by Silva et al., (2002) found that for 8 weak estrogenic chemicals applying the concept of independent action led to a clear underestimation of the experimentally observed results.

2.6.1.3 Interactions

The third concept “interaction” describes the combined effect between two chemicals resulting in a stronger effect (synergism, potentiation) or weaker effect (antagonism, inhibition). The term “interaction” is used here as an empirical description to characterize departure from additivity (Groten 2000). An interaction may occur in the toxicokinetic phase (processes of uptake, distribution, biotransformation and excretion) or in the toxicodynamic phase (effects of chemicals on the receptor, cellular target or organ) (Cassee et al., 1998; Groten 2000).

Antagonistic interactions: the situation in which the toxicity of two or more compounds administered together or sequentially is less than expected from their toxicities when administered alone (Mileson et al., 1998). Metabolism of methanol produces formaldehyde, a dangerous toxicant. Co-exposure of methanol and ethanol will reduce the production of formaldehyde from methanol, as ethanol will compete for the same biotransformation enzyme (alcohol dehydrogenase) (Preston et al., 2000).

Potentiative interactions: the situation when exposure to one compound does not produce an adverse toxic effect on its own. However, the toxicity of the compound is greatly increased when combined with another toxic chemical. An example of chemicals causing potentiation is the potential of carbon tetrachloride to cause liver damage is markedly increased if drugs such as isopropanol are taken at the same time (Eaton and Klaassen 2001).

Synergistic interactions: this situation occurs when the combined effect of two chemicals is much greater than the sum of the effects of each agent alone. An example of synergism is the co-exposure of n-hexane and a ketone (for example, methyl ethyl ketone) aggravates the development of delayed neuropathy (Eaton and Klaassen 2001).

57 Chapter 2

All three basic principles of joint action are theoretical. In reality, however, one will most likely have to deal with these concepts at the same time, especially when mixtures consist of more than two compounds and when the targets (individuals rather than cells) are more complex (Cassee et al., 1998; Groten et al., 2001). CA and IA do not always yield identical predictions, and the presumed modes of action do not always provide guidance as to which model to chose. The choice of concept is usually based by experimental validation (Kortenkamp 2004).

2.6.2 Application to risk assessment

It is common prudent practice in occupational hygiene to assume that the interaction of multiple exposures to chemicals at work can be estimated additively by summing exposures as a fraction of their exposure standard. But this is problematic in workplace multiple chemical exposures, because most standards are established just at the point the dose-response relationship increases above the NOAEL (Borgert et al., 2001). In these cases the margin of safety is so small such that the possible contribution from other exposures could be sufficient to render the protection offered by this standard very weak (Borgert et al., 2001). Another limitation of toxicological experiments whether on whole mixtures or mixture components is that usually the tests are not done at levels that reflect those seen in environmental samples (Teuschler et al., 2002).

Most of the federal agencies and international organisations such as ATSDR, US EPA, NIOSH, use a default assumption of dose or response additivity for the assessment of aggregate toxicity of multiple chemicals to which the human population can be exposed to (EPA 1986; EPA 2000; ATSDR 2004). Most regulatory actions and industrial practices are based on use of the default assumption that individual chemicals act independently of other chemicals in inducing an effect when multiple chemicals are taken into the body. Thus, in assessing the risks of exposures to mixtures of chemicals, the risks are treated in an additive manner. These assumptions are the basis for the assessment of toxicity of mixtures by simple algebraic process called the “Hazard Index” (HI). This also serves as the basis for “Toxic Equivalency” method which is the

58 Chapter 2

US EPA’s approach for chlorinated and brominated dioxins and furans (EPA 1986). The HI, which is a mathematical combination of risk assessments made on individual chemicals does not account for the number of individuals who might be affected by exposure or the severity of the effects. The EPA (1986), recommends the use of the hazard index initially to screen risk assessments on chemical mixtures. The use of the hazard index may under or overestimate risk if the chemicals act by different modes of action in that case independent joint action may be used. A very detailed attempt on the appropriate assessment of health impact of chemical mixtures is made in the Guidelines for the Health Risk Assessment of Chemical Mixtures (EPA 1986). A schematic of the recommendations in the US EPA guidelines is presented in Figure 2-8.

Exposure and toxicity data assessment

Mixture of Similar Additivity Decline risk concern mixture assumptions assessment

Risk assessment on MIXTURE Risk assessment on COMPONENTS

Use of Comparison of available interactions methods data

Recommended risk assessment

Figure 2-8 Modified overview of the US EPA guidelines for mixture risk assessment (EPA, 1986)

Mumtaz (1995) proposed a weight-of-evidence (WOE) scheme that uses interaction data in component-based risk assessment of mixtures. This scheme is intended as an extension of the EPA guidelines for the risk assessment of mixtures. The WOE takes

59 Chapter 2

into account interaction data and looks at all possible binary combinations of chemicals in the mixture, taking into account information about the toxicity and pharmacokinetics of the individual chemicals, and interactions data on related chemicals. One should first determine if the mixture is greater than additive, additive or less than additive. Then a study of the mechanism of the interaction of the chemicals or related compounds should be undertaken. The third step in the WOE scheme is the classification of the toxicological significance, and how clear or unclear it is and if it has been demonstrated directly or can be inferred. In vivo, in vitro data, anticipated route and sequence and duration of exposure are also included. But there are many limitations to the WOE approach in that it relies on data only from binary interactions, it fails to consider the magnitude of the interaction, and it fails to take into account inter and intra species variability (Mumtaz 1995).

ATSDR utilises Chemical Specific Minimal Risk Levels (MRLs) to assist in evaluating risks associated with exposure to hazardous substances (Pohl et al., 1997). MRLs for chemical mixtures are either based on Toxicity Equivalency Factors (TEFs), similar to the US EPA, or in some cases the most toxic chemical is assumed to drive the health assessment for the whole mixture (ATSDR 2004). In the TEF approach the dose of each component in the mixture is normalized generally against the dose of the most potent compound; the relative potencies are then summed to estimate the toxic potency of the whole mixture (Carpy et al., 2000). There are some limitations with the TEF approach in that the mixture of chemicals should consist of a well-defined group of chemicals (e.g. dioxins); it should have a broad database of information, consistency across endpoints and additivity of effects (EPA 2000). Most of the current approaches to generate combined risk assessments rely of chemical mixtures having similar mechanism of actions (Wilkinson et al., 2000). Similar approaches are now being attempted for Polycyclic Aromatic Hydrocarbons (PAH), endocrine disrupting chemicals such as estrogenic compounds and tricresyl phosphates (EPA 1984; Safe 1998; Segner et al., 2003)

Decisions the outcomes of which have multibillion dollar impacts and potentially affect millions of people, are being made on the basis of very meagre information about the toxicology of chemical mixtures (Yang 1994b), whereas a more scientifically sound

60 Chapter 2

basis for decisions on regulating exposure to chemical mixtures is needed now (Teuschler et al., 2002). Historically individual chemicals have been the focus of criteria derivation procedures. Usually, the target chemical or group of chemicals is identified by a government agency, international organisation, or some advisory body (Mumtaz et al., 1994). Sometimes, however, “real-world” exposures involve complex mixtures of chemicals, in that case, the available information on the mixture is reviewed and a criterion for the mixture is derived (Yang 1994b). The rationale for concern regarding risk assessment of mixtures is simple:

1. People are rarely ever exposed to only one chemical; background or residual exposures occur as well as multiple exposures. Individuals expose themselves to a variety of pharmacologically active chemicals in food, medications, and recreational substances such as ethanol and tobacco and environmental background exposure and, 2. Experimental evidence shows that various chemicals interact toxicologically. Hence exposure to one chemical may affect the toxicological potency of properties of another chemical.

Through the years different governmental organisations have tried to set standards and improve risk assessments for mixture toxicity. Some of these include: x The US EPA also recommended the use of TEF for dioxins, furans, PAHs, endocrine disruptors, and tricresyl phosphates (EPA 1984; EPA 1986). It currently requires an assessment of cumulative exposure for the approval of new pesticides (EPA 1999). x The US Food Quality and Protection Act (FQPA) of 1996: emphasized chemical mixtures by requiring assessments of cumulative risks (EPA 1999). HI and TEF, have commonly been conducted to estimate the cumulative risk (Chen et al., 2001). x ATSDR has pursued for many years a “Mixtures Research and Assessment Program” that consists of three components: trend analysis, joint toxicity assessment, and experimental testing. In the year 2001, the Agency for Toxic

61 Chapter 2

Substances and Disease Registry (ATSDR), proposed studies on chemical mixtures as one of six priority research areas (De Rosa et al., 2001). x OECD published in 2001, the “Harmonised Hazard Classification Criteria for Mixtures” (OECD 2001e). The GHS classification of chemical mixtures is dependent on a tiered approach that is dependent on the amount of information available for the mixture itself and for its ingredients (Section 2.7.5.2).

2.6.3 Role of in vitro in toxicity assessments of chemical mixtures

Conventional animal toxicology testing methods are inadequate for the evaluation of chemical mixtures because of the complexity and high demand on resources (Yang 1998). A list of the limitations of animal studies can be found in Section 2.2.1.1. Other limitations specific to the study of chemical mixtures apply to conventional animal testing such as: x Single time point determinations at terminal sacrifice; x The mechanistic base for the interaction of the chemicals typically remains unknown as they are not generally designed to obtain pharmacokinetic information on the chemical in the body and its pharmacodynamics (Teuschler et al., 2002).

A review of the literature showed the following in vitro studies being carried out on mixtures: x A study undertaken in our laboratory (Malich et al., 1998) used MTS in vitro cytotoxicity assay to detect the cytotoxic effects of 34 binary and ternary chemical mixtures of structurally different chemicals on human HeLa cells. They tried to establish a quick and reliable method that accounts for smaller than additive, additive and greater than additive effects of the chemicals. The mixture cytotoxicity data derived using the MTS assay was compared to data on independent effects of

62 Chapter 2

each component, data calculated with a predictive mathematical algorithm and then cytotoxicity on blind samples were carried out to simulate testing of unidentified samples. The results of this study showed that a rapid and accurate in vitro toxicological test method such as the MTS assay can be used as ideal screening tool for qualitatively identifying and characterising health hazards being described as major risk-related activity inherent in the regulatory decision making process. x An investigation conducted on three related herbicide formulations and their individual components in which an in vitro assay using sub-mitochondrial particles (SMPs) (Oakes and Pollak 2000) provided further evidence that it is essential to establish the toxicities of the complete formulations rather than just evaluating the toxicities of the active components.

The US EPA has long known and acknowledged the potential for adverse effects from the use of pesticide products containing what is termed as “inert” ingredients (US EPA 1987) and subsequently classified a number of inert ingredients as being of “toxicological concern” (US EPA 1989). In 1997, a directive was issued encouraging those responsible for regulations of pesticides in particular, to replace the term “inert” ingredients with “other” ingredients since inert wrongly implies that the ingredient has no intrinsic toxicity (US EPA 1997). These directives should be expanded to classify most chemical products containing ‘inert’ ingredients. It is also well known that components such as surfactants may interfere with the ordered structure of the phospholipid bilayer of biological membranes (Argese et al., 1994).

Diverse in vitro assays mainly detecting cellular growth and vitality have been used on chemical mixtures such as; MTT, MTS, LDH leakage, NRU, ATP and Coomassie blue assay. Chemical mixtures included; triclosan and sodium lauryl sulfate, combined toxicity of aldehydes, nickel and cobalt chloride, benomyl and organophosphorous insecticides, and mixtures of estrogenic chemicals (Cassee et al., 1996; Marinovich et al., 1996; Babich and Babich 1997; Sauvant et al., 1997; Cassee et al., 1998; Feron et al., 1998; Payne et al., 2000; Bae et al., 2001; Cross et al., 2001; Payne et al., 2001; Rajapakse et al., 2001; Dalzell et al., 2002; Silva et al., 2002). Chemicals were mostly

63 Chapter 2

tested using human cell lines (HeLa cells, HL-60 leukaemia, fibroblasts, neuroblastoma and gingival S-G epithelial cell lines) and aquatic organisms.

2.7 Regulatory Bodies: Hazardous Classification of Chemicals

In many countries, legislative and administrative measures have been introduced to deal with chemical hazards. There has been intense activity in many nations over the past 30- 40 years to identify and deal with problems arising out of the use of chemicals resulting in the continuous development of legislation in response to local as well as international developments (for example, thalidomide, asbestos, persistent bioaccumulative toxic chemicals, hazardous wastes, ozone depleting chemicals, greenhouse gases and so on) (Brickman et al., 1985).

Chemical control regulation is a highly complex area (Brickman et al., 1985; Winder and Barter 2004), in which scientific and legal issues are brought together. This is further complicated by historical precedents or jurisdictional subtleties. Chemical control legislation operates at different levels, with different jurisdictional demarcations and different administrative arrangements (Winder and Barter 2004). It may also deal with chemicals risk in different ways; for example, contrast the difference between the hazards and risks of 1,000 mls of Xylene (a hazardous substance) will have a hazard of harmful vapours in use, as compared to 10,000 L of Xylene (a Dangerous Goods) which has a hazard of flammability in storage.

Sectoral responsibilities within regulatory agencies will often take different approaches in the development of legislation, standards and administrative structures. Further, this complexity of legislation is not limited to one state or one nation. A summary review of the chemicals classification and control legislation in two nations (USA and Australia) and the European Union follows.

64 Chapter 2

2.7.1 The USA

In the USA, four federal agencies are primarily responsible for regulating exposures to chemicals. These agencies administer over two dozen statutes, which have been enacted over time, and all have protection of health as their main goal. Table 2 5 summarises the stepwise development of the major chemical control laws at the Federal level in the USA.

Table 2-5 Major US chemical control laws and agencies Act or Law Responsible Federal Agency Food and Drugs Act 1906 FDA Food, Drug and Cosmetic Act 1938, amended 1951, 1962 FDA Food Additives Amendment 1958 FDA Pesticide Residue Amendment 1954 EPA Consumer Product Safety Act 1972 Now CPSC Medical Devices Amendment 1976 FDA Federal Hazardous Substances Act 1960 CPSC Poison Prevention Packaging Act 1979 CPSC Labelling of Hazardous Materials Amendment 1988 Occupational Health and Safety Act 1970 OSHA Consumer Products Safety Act 1972 CPSC Federal Water Pollution Control Act (Clean Water Act) 1948 EPA Amendments 1972,1983,1992,1996 Safe Drinking Water Act 1974 EPA Amended in 1986, 1996 Clean Air Act 1970 EPA Amended 1977, 1990 Toxic Substances Control Act 1976 EPA Amended 1981, 1984, 1986 Resource Conservation and Recovery Act 1976 EPA Amended 1980, 1984, 1986 Federal Insecticide, Fungicide and Rodenticide Act 1947 EPA Amended 1972, 1988, 1996 Comprehensive Environmental Response, Compensation and Liability Act 1980 EPA Hazard Communication Standard 1983 OSHA Amended 1988 Superfund Amendments and Reauthorisation Act 1986 EPA Abbreviations: FDA: Food and Drug Administration EPA: Environmental Protection Agency CPSC: Consumer Product Safety Commission OSHA: Occupational Safety and Health Administration

65 Chapter 2

Other regulatory arrangements relating to consumer products include: x Department of Transportation (DOT): Materials transported on US roadways, railways or airways must be shipped in appropriately labelled vessels. They are ranked into class A poisons and class B poisons. Class “A” poisons are considered extremely dangerous poisons, represent inhalation hazards and are defined as “poisonous gases or liquids of such nature that a very small amount of the gas or vapour of the liquid, mixed with air, is dangerous to life.” These would, for example, include materials such as: phosgenes or cyanide producing materials. Class “B” poisons are materials that will produce death within 48 hours in half or more than half of a group of 10 or more white laboratory rats weighing 200 to 300 grams at single dose of 50 milligram or less per kilogram of body weight when administered orally; or if administered by continuous contact with bare skin for 24 hours or less it has to produce death to half or more than half of a group of ten or more rabbits at a dosage of 200 milligrams or less per kilogram of body weight. x Consumer Product Safety Commission (CPSC): This government commission was created in 1972 and is responsible for assuring that the consumer is not exposed to any unduly hazardous products and that any potentially hazardous products are properly labelled. The CPSC plays the least important role of the US Federal agencies controlling hazardous chemicals (Howell 1998). The Commission is empowered to promulgate safety standards that will prevent or reduce an unreasonable risk of injury related to the consumer product. The Commission has the right to ban a product (CPSC, Section 8), if no feasible standard could adequately protect the public from “unreasonable risk of injury”. In assessing this need for a standard or ban of a product, the agency needs to balance the likelihood that a product will cause harm, and the severity of harm it will likely cause, against the effects of reducing the risk on the product’s utility, cost and availability to consumers(Howell 1998). The CPSC administers the Federal Hazardous Substances Act (FHSA), 1960. This act defines the severity of toxicity of substances based on certain criteria. A substance is considered “toxic” if it has the “capacity to produce injury or illness to human through ingestion, inhalation, or absorption through any body surface”. The rate of toxicity by ingestion or dermal absorption is

66 Chapter 2

based on acute toxicity tests conducted mostly on animals and concentration cut off values exist for toxic and highly toxic classifications Table 2 6. CPSC has prescribed labelling for products containing substances that are acutely toxic such as “DANGER” label for highly toxic and “WARNING” or “CAUTION” for other hazardous substances. The Labelling of Hazardous and Materials Act of 1988 (LHAMA) required the CPSC to also provide labelling for material that has the potential of producing chronic adverse health effects (Merrill 2001).

Table 2-6 CPSC classification of the toxicity of materials based on acute oral or dermal toxicity tests

Type of test Toxic Highly (mg/kg) (mg/kg)

Oral (death after oral administration to half or more of Above 50 to 2000 Below 50 a group of laboratory rats within 14 days)

Dermal (death after continuous contact with skin for Above 200 to 2000 Below 200 24 hours or less administration to half or more of a group of laboratory rabbits within 14 days)

x The US EPA: has responsibility for registration and labelling of pesticides under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), 1996 and for regulation of chemicals and other potentially hazardous materials under the Toxic Substances Control Act (TSCA), 1976. The US EPA is responsible for establishing labelling and packaging standards for pesticides. It has also developed a set of testing guidelines, including acute testing guidelines, for use in the testing of pesticides and toxic substances and developing test data for submission to the agency for review. In contrast to most regulations that provide minimal specifications (species weight, dose, and duration of exposure and observation periods), the above-cited guidelines provide several pages of detailed information and testing procedures (Merrill 2001).

67 Chapter 2

x Occupational Health and Safety Administration (OSHA): Has the mission to assure the safety and health of America’s workers by setting and enforcing standards; providing training, outreach, and education; establishing partnerships; and encouraging continual improvement in workplace safety and health, including, among other things, the workplace control of chemicals. OSHA has also been working with the European Commission on development and implementation of the GHS. x Department of Health, and Human Services, which has responsibility for: Food and Drug Administration (FDA): the agency that regulates pharmaceuticals, as well as through its Bureau of Foods, it exercises premarketing approval authority for the safety of direct and indirect food and colour additives used in food; ATSDR: charged under Superfund legislation to assess the presence and nature of health hazards at specific Superfund sites, and to reduce or prevent illnesses that result from such exposures.

2.7.2 The EU

While there were national systems for classification and labelling of chemicals in Europe (such as the UK Health and Safety Executive), the then European Economic Community (EEC) became the regulatory agency for chemicals in Europe in the 1960s. Regulatory instruments of the now EU include regulations, directives, decisions, recommendations and opinions. Of these regulations and directives are binding on member states with specification of dates of compliance. Directives differ from regulations in that they specify objectives, but not the methods for compliance. The EU introduced the first Directive on dangerous substances or chemicals harmful to people or the environment in 1967 (Council Directive 67/548/EEC introduced the administrative structures to harmonise the laws of Member States governing the testing, classification, packaging and labelling of dangerous substances). It has been amended and updated many times since 1967, and additional directives have broadened the scope of EU Chemicals policy. The more important of these directives and regulations have been:

68 Chapter 2

x Council Directive 79/831/EEC was the sixth amendment, which introduced a pre- market testing and notification system for new chemical substances being marketed in the European Union. This included introduction of European Inventory of Existing Commercial Chemical Substances (EINECS) under Council Directive 81/437/EEC. x Council Directive 82/501/EEC was the Seveso directive, introduced after the 1976 accident in the Italian town of Seveso and concerned with controlling the risks of exceptional or major accidents such as fire, explosions or major emissions and requires various measures to be taken to prevent and contain such accidents and their consequences. x Council Regulation 428/89/EEC was introduced for the control of the export of chemicals used in the development or production of chemical weapons. x Commission Directive 91/155/EEC, which defined and outlined arrangements for the system of specific information relating to chemical products (dangerous preparations). x Commission Directive 91/322/EEC on the establishment of indicative limit values for exposure to chemical, physical and biological agents in the workplace. x Council Regulation 793/93/EEC dealt with the evaluation and control of substances (for example, existing chemicals) not covered under Directive 79/831/EEC. The Directive requires that a data notification procedure be undertaken for evaluating the risks posed by existing substances, including all those listed in EINECS (then containing 100,195 substances). x Commission Regulation 1488/94/EEC laying down the principles for the assessment of risks to humans and the environment. x Commission Directive 2001/58/EC on the establishment of indicative limit values for exposure to chemical, physical and biological agents in the workplace.

69 Chapter 2

Many of these directives have now been amended by later directives.

REACH: The European Commission (EC) set up a proposal, in October 2003, for the complete and radical review of the EU’s existing chemical substances policy. The proposal sets up a single regulatory framework for the Registration, Evaluation, and Authorisation of Chemicals (REACH). It still needs to be approved by the Council and the European Parliament. An estimated 30,000 substances; existing (placed on market before 1981) and new are to be assessed through the REACH process. This system aims at shifting the burden of testing and assessment of chemicals from authorities to industry. It follows a top-down approach to toxicity testing, which is dependent on the amount of a chemical produced (tonnage) based on the traditional assumption that the higher the level of a substance, the greater the potential level of overall human exposure.

2.7.3 Australia

There are four national chemical assessment and registration schemes which cover food, industrial chemicals, pharmaceuticals and agricultural and veterinary chemicals. The schemes operate in a complementary manner to ensure there is no duplication or any unnecessary regulatory burden on industry.

The scope of each of the four chemicals assessment and/or registration schemes is defined by legislation. Legislation also specifies what chemical/chemical products are to be covered by each of the schemes, as well as the requirements for anyone involved in chemicals manufacture and/or importation.

In Australia, responsibility for chemical regulation in the health sector is shared by a number of Australian Government Commonwealth bodies and in some cases in conjunction with New Zealand (Taskforce 2000). Specific legislation empowers the operation of various Commonwealth public health and Safety regulatory functions. While jurisdictional coverage and most government functions occur at the State/Territory level in Australia, responsibility for chemicals notification and

70 Chapter 2

assessment occurs at the Federal government levels. The main regulatory agencies are summarised in Table 2 7 with their specific legislation.

Table 2-7 Major Australian chemical control laws and agencies

Act, Law, Statutory Standard Responsible Federal Agency

Therapeutic Goods Act 1989 TGA/DoHA Standard for the Uniform Scheduling of Drugs and Poisons

Agricultural and Veterinary Chemicals (Administration) Act 1992; Agriculture and APVMA/DAFF Veterinary Chemicals Code Act 1994.

Food Standards Australia New Zealand Act (1991) FSANZ Food Standards Code

Australian Dangerous Goods Code FORS/DoT Industrial Chemicals (Notification and Assessment) Act 1989 (National Industrial DoHA/DEH/ Chemical Notification Assessment Scheme or NICNAS) NOHSC

Trade Practices Act 1974 ACCC Prices Surveillance Act 1983

Abbreviations: ACCC : Australian Competition and Consumer Agency APVMA: Australian Pesticides and Veterinary Medicines Authority FSANZ: Food Standards Australia New Zealand CPSC: Consumer Product Safety Commission DAFF: Australian Department of Agriculture, Fisheries and Forestry DoHA: Commonwealth Department of Health and Ageing FORS: Federal Office of Road Safety, Commonwealth Department of Transport DEH: Department of Environment and Heritage NOHSC: National Occupational Health and Safety Commission TGA: Therapeutic Goods Administration

Other regulatory arrangements relating to consumer products are discussed below: x Therapeutic Goods Administration (TGA): TGA is a unit of the Australian Government Department of Health and Ageing is a member of the TGA Group of Regulators and was established in 1991. The TGA is responsible for administering the provisions of the Therapeutic Goods Act 1989, and its Regulations and Orders. The objective of the Act, which came into effect on 15 February 1991, is to provide

71 Chapter 2

a national framework for the regulation of therapeutic goods in Australia and ensure their quality, safety and efficacy. The TGA is responsible for regulating the supply in Australia of therapeutic goods including prescription, non-prescription and complementary medicines (herbal products, vitamins, minerals and homoeopathic products) and therapeutic devices. Legislative amendments were made in 1999 to the Therapeutic Goods Act 1989, to cover processes for establishing national standards for drugs and poisons. This assists in providing national uniformity in the levels of control of drugs and poisons, including in the area of product labelling. x The Office of Chemical Safety, TGA Group of Regulators, within the Australian Government Department of Health and Ageing, comprises: o The National Industrial Chemicals Notification and Assessment Scheme (NICNAS), which has the objective of aiding in the protection of people at work, the public and the environment from the harmful effects of industrial chemicals. Currently (with certain exemptions), all new industrial chemicals and existing chemicals must be notified and/or assessed to NICNAS on a priority basis prior to their import, manufacture or use in Australia. NICNAS operates under Commonwealth legislation known as the Industrial Chemicals (Notification and Assessment) Act 1989 (the Act). NICNAS aims to ensure the safe use of chemicals by making information on chemicals and their potential occupational health and safety, public health and environmental risk widely available to workers, the public, industry, and other State, Territory and Commonwealth government agencies; o The public health risk assessment for pesticides, veterinary medicines and other chemicals to which the public may be exposed; o Secretariat support and preparation of the Standard for the Uniform Scheduling of Drugs and Poisons (SUSDP), an Australian Government standard used by all jurisdictions in Australia for the classification, labelling, packaging and general control of drugs and poisons. The standard forms a major component of Poisons legislation in Australia, and assists in providing national uniformity in the levels of control of drugs and poisons;

72 Chapter 2

o Monitoring and compliance activities in relation to Australia’s obligations for international drug treaty arrangements and other prohibited/controlled substances.

The Office also has responsibilities for giving effect to Australia's obligations under international agreements relating to the regulation of chemicals, and for collecting statistics about chemicals x Australian Pesticides and Veterinary Medicines Authority: The APVMA (formerly the National Registration Authority for Agricultural and Veterinary Chemicals) was established in 1993 as an independent Statutory Authority in the Australian Government Portfolio of Agriculture, Fisheries and Forestry. The APVMA administers legislation established under the National Registration Scheme on behalf of the Australian and State/Territory Governments and is responsible for the assessment, registration and regulation of agricultural and veterinary chemical products (including some domestic and household products such as insect sprays) up to, and including, the point of retail sale. Controlling the use of agricultural and veterinary chemicals is the responsibility of the relevant State/Territory authority. The Australian system of agricultural and veterinary chemical registration (including standards and standard setting processes) is aligned closely with international best practice. Before an agricultural or veterinary chemical product can be sold in Australia, it must be assessed and registered by the APVMA. Chemical companies are required to provide extensive data to demonstrate that a product will be effective for the uses described on the label, will be safe for humans and non-target species, and will not pose unacceptable risks to the environment or trade with other nations. When products are evaluated, the APVMA takes full account of the nature of the product, the amount and completeness of data for consideration, and the extent of consultation required between the APVMA, manufacturers, advisory agencies, and State/Territories Governments.

For specialist advice during the assessment process, the APVMA receives input from a number of Australian Government Agencies:

73 Chapter 2

x The TGA’s Office of Chemical Safety within the Australian Government Department of Health and Ageing (DHA) evaluates toxicology data submitted by applicants to determine if any health risk may be posed to the community. x The Australian Government Department of Environment and Heritage evaluates the environmental implications of products submitted for registration and recommends measures to avoid or minimise adverse environmental effects. x The National Occupational Health and Safety Commission (NOHSC), now the Australian Safety and Compensation Commission (ASCC) conduct occupational health and safety assessments to ensure that any risks arising out of workers’ exposure to agricultural and veterinary chemical products are minimised. x Food Standards Australia New Zealand assesses the dietary intake implications of residues in food and in cooperation with the APVMA sets maximum residue limits. x The Office of the Gene Technology Regulator provides advice in relation to products of gene technology. x The National Health and Medical Research Council’s Expert Advisory Group on Antimicrobial Resistance addresses the implications of the use of antibiotics in agriculture. x The Australian Quarantine and Inspection Service advises on quarantine safety matters associated with imported biological products.

The APVMA operates several programs that monitor agricultural and veterinary chemicals after registration. The Chemical Review Program reconsiders the registration of agricultural and veterinary chemicals in the marketplace where potential risks to safety and performance have been identified. The APVMA’s Manufacturers’ Licensing Scheme requires all Australian based manufacturers of veterinary chemical products to be licensed and to meet standards described in a Code of Good Manufacturing Practice. In addition, the APVMA monitors compliance as well as the reporting of adverse experiences resulting from the use of agricultural and veterinary chemicals. It is these features that underpin the credibility of Australia’s agricultural and veterinary chemicals management system. x Federal Office of Road Safety (FORS), Commonwealth Department of Transport and Regional Services: FORS promotes best practice and development of

74 Chapter 2

harmonised legislation for the transport of dangerous goods and explosives in Australia. It also coordinates implementation of the recommendations of United Nations Committee of Experts on the Transport of Dangerous Goods (UNCETDG) in Australia. The Department provides secretariat support and preparation of the Australian Dangerous Goods Code (ADG Code), an Australian Government standard used by all jurisdictions in Australia for the classification, labelling, packaging and general control of dangerous goods. The standard forms a major component of Dangerous Goods legislation in Australia, and assists in providing national uniformity in the levels of control of Dangerous Goods. x National Occupational Health and Safety Commission (NOHSC): NOHSC was abolished in 2003. Its functions have been transferred to the Australian Safety and Compensation Commission. NOHSC is a tripartite statutory body, with government, employer and employee representation. It provides a forum for consultation and development and formulation of policies and strategies relating to occupational health and safety matters. There are three NOHSC regulatory instruments for the control of chemicals in the workplace: o The Hazardous Substances Regulatory Package; o The National Standard and Code of Practice for the Storage and Handling of Dangerous Goods; o The Major Hazards Facility Standard. x Food Standards Australia New Zealand (FSANZ): FSANZ, formerly Australia New Zealand Food Authority (ANZFA) is a binational Statutory Authority that in cooperation with the Australian Commonwealth, State and Territory governments and the New Zealand Government, develops food standards and other regulatory measures for Australia and New Zealand. These standards are published in the Food Standards Code once they are approved by the FSANZ Board and then considered by the Australia and New Zealand Food Regulation Ministerial Council (ANZFRMC; formerly the Australia New Zealand Food Standards Council (ANZFSC). ANZFRMC can request a review of any standard developed by FSANZ.

75 Chapter 2

FSANZ develops food standards under the Food Standards Code which include all food matters. In particular, in relation to chemicals that may be included in foods for a specific technological purpose or which may enter food products as a result of natural or accidental contamination, there are standards for food additives, processing aids and contaminants in the Food Standards Code with maximum levels set in various foods where appropriate. FSANZ develops the food standards but does not enforce them; this is the responsibility of the State, Territory and New Zealand governments who adopt the standards in the Food Standards Code into their respective State, Territory and New Zealand Food Acts or Regulations. However, FSANZ has a coordination role to harmonise consistent interpretation of enforcement requirements. x Australian Competition and Consumer Commission (ACCC): ACCC is an independent statutory authority formed in 1995. It administers the Trade Practices Act 1974 and the Price Surveillance Act 1983. The Act covers anti-competitive and unfair market prices, product safety/liability, mergers and acquisitions of companies and third party access to facilities of national significance. The ACCC has been involved in product liability cases in the past.

2.7.4 International or multinational bodies

National initiatives are carried out by sovereign nations to deal with specific initiatives within their jurisdictional areas and responsibilities. However, for chemicals, some activities, for example classification and labelling, do not require constant duplication by numerous national agencies. Therefore, chemicals hazard identification, assessment and control is an area that may be better managed at the international or multinational level (IOMC 1996).

A number of international bodies have taken on chemicals related activity within the framework of the IFCS: x United Nations: The UN has been at the forefront of chemicals control activities since its formation in 1945. Examples of activities include:

76 Chapter 2

o International Code of Conduct on the Distribution and Use of Pesticides: The International Code of Conduct was developed to address a number of difficulties associated with the use of pesticides in developing countries where adequate regulatory infrastructures are frequently lacking; o United Nations Environmental Program (UN-EP), London Guidelines for the Exchange of Information on Chemicals in International Trade: This set of guidelines is addressed to governments with a view to assisting them in the process of increasing chemical safety in all countries through exchange of information on chemicals in international trade. These guidelines are general in nature and are aimed at enhancing the proper management of chemicals through the exchange of scientific, technical, economic and legal information; o United Nations Transport of Dangerous Goods (UN-TDG): UN-TDG provides a basis for the development of harmonised regulations for all modes of transport, in order to facilitate trade and the safe, efficient transport of hazardous materials. It covers all aspects of transportation necessary to provide international uniformity. The regulations include a comprehensive criteria based classification system. Hazards regulated include explosivity, flammability, toxicity (oral, dermal, and inhalation), corrosivity, reactivity, radioactivity, infectious substance hazards and environmental hazards. This also includes a system of hazard communication such as cover labelling, documentation and emergency response information. Many of the national and international regulations governing the transport of dangerous goods are based on the UN recommendations (including the USA and Australia), therefore facilitating compliance and decreasing confusion. Previously, some of the regulations were structured differently requiring transporters to be familiar with the unique structure of all applicable regulations. The lack of harmony of regulations can lead to frustration in compliance resulting in non-compliance that is detrimental to safety. o The Basel Convention on the Transboundary Transport of Hazardous Wastes.

77 Chapter 2

x OECD: The OECD is an international organisation grouping of about thirty industrialised countries and country groupings. Member nations include the United States, Australia, New Zealand, Canada, Mexico, South Korea, Turkey and many of the developed nations of the European Economic Community. The OECD has formed expert groups to review toxicity testing requirements for the member nations and formulate testing guidelines, which would be acceptable to all members. These guidelines are a collection of the most relevant internationally agreed testing methods used by government, industry and independent laboratories to assess potential hazards of new and existing chemical substances and mixtures. It incorporates some procedures that are designed to reduce numbers of animals used in experiments and to limit the amount of pain they experience. OECD has published over 100 guidelines that are constantly updated or renewed. An example of a published OECD guideline specific to acute oral toxicity is Testing Guideline 423, Acute Oral Toxicity- Acute Toxic Class Method, published on March 22, 1996 and updated December 2001 (OECD, 2001a).

2.7.5 The Globally Harmonised System (GHS)

The presence of many hazard classification schemes management regimes for chemicals, both nationally and internationally, makes for a confused picture and makes it difficult to implement suitable chemicals control management. Calls for a harmonised system for chemical hazards classification and hazard communication began in the 1980s (ILO 1990). The need for the GHS was identified as countries had differing abilities to identify and systematically regulate every hazardous chemical. Most countries have developed systems that require the transmission of information through labels and/or safety data sheets (Pratt 2002). However, most of these countries have few, if any, requirements to communicate the hazard of chemicals. There also exist inconsistencies in the classification and labelling of the same chemical between the different countries, or within different sectors in the same country or manufacturers. For example, some chemicals are classified as flammable or carcinogenic in one country and not in another. These differences in classification have a strong impact on the protection to human health, the environment, and on trade.

78 Chapter 2

The GHS provides the infrastructure for a globalised, consistent approach to the classification of chemicals and a coherent and consistent approach to defining and classifying chemical hazards and communicating information on labels and safety data sheets (Pratt 2002). Obviously, with systems already in place for classification and labelling of hazardous chemicals, the GHS serves as a focus for convergence of the existing systems. The GHS can then be used for the establishment of a comprehensive chemical safety program at the national or regional level.

Work on the GHS began in 1989, when the International Labour Organisation (ILO) adopted a resolution concerning the harmonisation of systems of classification and labelling (ILO 1990). In the early development of a globally harmonised system, the existing chemical classification and labelling systems of the following international organisations and countries were considered: x OECD Chemicals Program; x ILO Chemical Safety Tools; x UN Recommendations for Transport; x FAO Recommendations on Pesticides; x UN Transport Recommendations; x EU directives on Dangerous Substances and Preparations; x US requirements for Workplace, Consumers and Pesticides; and x Canadian Requirements for Workplace, Consumers and Pesticides.

The process of harmonisation fell under the umbrella of the Interorganisational Programme for the Sound Management of Chemicals (IOMC). The GHS covers all hazardous chemical substances, dilute solutions and mixtures but it does not cover pharmaceuticals, cosmetics, food additives, and pesticide residues in food except when workers are exposed and in transport (OECD 2001a; OECD 2001e). The GHS considers that classification of a chemical substance depends on the criteria and on the reliability of the test methods underpinning the criteria. Tests that determine hazardous properties, which are conducted according to internationally recognized scientific principles, can be used for the purposes of a hazard determination for health and environmental hazards. The GHS criteria are test-method neutral, and are performance based, in the sense that they allow for any approach as long as it is scientifically sound

79 Chapter 2

and validated according to international procedures. Criteria for physical hazards are linked to specific test methods for hazard classes such as flammability and explosivity.

The new GHS, which was adopted in December 2002, now moves to the front of the list of major regulatory issues facing virtually all government agencies with responsibility for regulating chemicals, as well as industry and unions over the coming years. This new system, the outcome of collaborative efforts of the WHO, the ILO, the OECD, and the UN, as well as member countries of the above organisations, has broad support from the chemical industry because of its promise to harmonize at international level the manner in which chemicals are classified according to their hazards and labelled using universally understandable pictograms, as well as a uniform system of safety data sheets (SDS).

The main GHS elements are classification criteria for substances and mixtures (for physical effects; toxic (health) effects; environmental effects) and requirements for hazard communication for chemicals (labels and SDS).

2.7.5.1 Classification criteria

A full explanation of classification criteria for physical, health and environmental effects can be found at: http://www.unece.org/trans/danger/publi/ghs/ghs_text- pdf/GHS-ANNEX-2.pdf

A summary of one of the main elements of the classification system is provided below. The communication of hazard in the GHS is based on the provision of signal words, hazard statements and pictograms, all of which are linked to the specific hazard of the substance or mixture. x Health effects

An assigned pictogram, signal word and hazard statement are given for each hazard category of the hazard class (Tables 2-8 to 2-12). The symbols presented in Table 2-8

80 Chapter 2

to Table 2-12 are the standard symbols which should be used in the GHS. Two new symbols have been introduced, the symbol used for health hazard and the exclamation mark are part of the standard set used in the “UN Recommendations on the Transport of Dangerous Goods, Model Regulations”.

Toxicity (health) hazards are: x Single dose toxicity, covering a range of toxicity endpoints by various routes of exposure (see Table 2-8). The skull and cross bones symbol for acute oral toxicity represents a signal word “danger”; the exclamation mark represents a signal word “Warning”. There are no symbols for category 5 and a signal word “Warning” is given for that category. x Skin irritation and corrosion: Category 1 is for corrosive effects and Categories 2 and 3 for irritation (see Table 2-9) x Eye irritation and serious eye damage: Category 1 is for corrosive effects and Categories 2A and 2B for irritation (see Table 2 -10) x Skin or respiratory sensitisation: If evidence is available to allow a classification of sensitisation, both skin and respiratory sensitisation are in category 1, but note the new symbol for serious effects for respiratory sensitisation (see Table 2- 11) x Single or repeated dose target organ systemic toxicity (TOST): This uses similar criteria for both single and repeated exposures (see Table 2 -12) x Genotoxicity and germ cell toxicity: As with the EC criteria, there are two categories (see Table 2- 13) x Reproductive toxicity: Reproductive toxicity includes adverse effects on sexual function and fertility in adult males and females, as well as developmental toxicity in the offspring. (see Table 2- 14) x Carcinogenicity: As with the EC criteria, there are two broad categories (see Table 2 -15).

81 Chapter 2

Table 2-8 GHS, criteria for single dose toxicity

Toxicity category 1 2 3 4 Category 5

Oral (mg/kg) 5 50 300 2,000 Oral LD50 greater than 5000 mg/kg Dermal (mg/kg) 50 200 1,000 2,000 Indication of significant effect in Gases (ppm) 100 500 2,500 5,000 humans Any mortality in Category 4 Vapours (mg/L) 0.5 2 10 20 Indications from other studies Dusts and mists (mg/L) 0.05 0.5 1 5

Table 2-9 GHS criteria, skin corrosion/irritation

Category 1 Category 2 Category 3

Destruction of skin tissue; Reversible adverse effects Visible necrosis in one or more of 3 animals in skin tissue

Sub-category 1A Sub-category 1B Sub-category 1C Exposure: less than 4 hours Observations: less than 14 days

Exposure: 3 mins Exposure: between Exposure: between Mean irritation Mean irritation or less 3 minutes and 1 1 and 4 hours score of 2.3 to 4 score of 1.5 to 2.3 Observations: up hour Observations: up for erythema/ for erythema/ to 60 mins Observations: up to 14 days eschar or for eschar or for to 14 days oedema at 24, 48 oedema in at least and 72h in at least 2 of 3 tested 2 of 3 tested animals at 24, 48 animals persistent and 72h inflammation at the end of the observation period

82 Chapter 2

Table 2-10 GHS criteria, serious eye damage/ eye irritation

Category 1 Category 2

Adverse effects on conjunctiva, cornea, iris Reversible adverse effects on conjunctiva, that have not reversed within the observation cornea, iris period (normally 21 days after exposure) in at Mean irritation score in at least 2 of 3 tested least one animal, and/or animals of 1 or more for corneal opacity and or In at least 2 of 3 tested animals, a positive 1 or more for iritis, and/or mean scores of 2 or response of corneal opacity with a mean score more for redness and/or 2 or more for of 3 or above, and/or a mean score of more conjunctival oedema (chemosis) than 1.5 for iritis, at 24, 48 and 72h Subcategory 2A Subcategory 2B

Reversible in 21 days Reversible in 7 days

Table 2-11 GHS criteria, respiratory or skin sensitisation

Respiratory Category 1 sensitisation

Evidence in humans of specific respiratory sensitivity and/or Results of respiratory sensitivity from animal studies Skin sensitisation Category 1

Evidence in humans of sensitisation by skin contact in a substantial number of persons, or Results of skin sensitivity from appropriate animal studies

83 Chapter 2

Table 2-12 GHS criteria, single or repeated target organ systemic toxicity (TOST)

Category 1 Category 2

Significant toxicity in humans Presumed to be harmful to human health Significant toxicity in humans: Animal studies with significant toxic effects Reliable, good quality human case studies or relevant to humans at generally moderate (as a epidemiological studies guide) exposure Presumed significant toxicity in humans: Human evidence in exceptional cases Animal studies with significant and/or severe toxic effects relevant to humans at a generally (as a guide) low exposures

Table 2-13 GHS criteria, germ cell mutagenicity

Category 1 Category 2

May induce heritable mutations in human germ cells Subcategory 1A Subcategory 1B Known to produce heritable Should be regarded as if they Positive evidence from tests in mutations in human germ cells produce heritable mutations in mammals and somatic cell tests Positive evidence from human the germ cells of humans In vivo somatic genotoxicity epidemiological studies Positive results in: supported by in vitro Human germ cell tests mutagenicity In vivo heritable germ cell tests in mammals In vivo somatic mutagenicity tests, combined with some evidence of germ cell mutagenicity

84 Chapter 2

Table 2-14 GHS criteria, reproductive and developmental effects

Category 2 Additional category Category 1 effects on lactation Or effects via lactation

Suspected to cause Known or presumed to cause effects on effects on human human reproductive ability/capacity or on reproductive ability/ development capacity or on development Subcategory 1A Subcategory 1B Known (based on Presumed (based on human data) animal data)

Table 2-15 GHS criteria, carcinogenicity

Category 1 Category 2

Known or presumed human carcinogen Suspected human carcinogen Subcategory 1A Subcategory 1B Known human carcinogen Presumed human carcinogen Limited evidence of human based on human evidence based on demonstrated animal or animal carcinogenicity carcinogenicity

2.7.5.2 Classification of mixtures

Mixtures will be classified according to a tiered approach, and are dependent on the amount of information available for the mixture itself and for its ingredients (OECD 2001e). If available, test data for the mixture is used, and then the mixture is classified to the same criteria as for substances; if test data is not available for the mixture, “bridging principles” will be used, if applicable; examples of bridging principles include the use of available data on individual ingredients of a mixture or a similar tested mixture, provided that sufficient data is present. This ensures that the classification uses available data without necessity for additional animal testing; if no other information is available, estimate hazards based on the available information on the known ingredients, using toxicity additivity approaches based on the calculation of an acute

85 Chapter 2

toxicity estimate (ATE) of ingredients (Equation 2-1). This does not cover possible synergistic or antagonistic effects.

100 Ci ¦ [2 -1] ATEmixn ATEi

Ci represents concentration of ingredient i n number of ingredients and is running from 1 to n ATEi represents Acute Toxicity Estimate of ingredient i

The development of the GHS is largely complete. The next phase is implementation. The Intergovernmental Forum for Chemical Safety recommends that all countries implement the GHS as soon as possible with a view to have the system fully operational by 2008. The Australian Federal Government has given this commitment at the World Summit on Sustainable Development (WSSD) in Johannesburg 2002, and government agencies are now addressing GHS issues.

Ultimately, it is hoped that the GHS will provide all countries with a structure to classify and label hazardous chemicals, and ensure suitably understandable information is available for all manufactured, imported and exported chemicals. In this way, a system will be established that will form the basis of one international system for the sound worldwide management of chemicals.

On a final note, the GHS criteria for determining health hazards are test method neutral, allowing for the introduction of different approaches, including alternative to animal methods, as long as they are scientifically sound and validated according to international procedures and criteria. Wherever available, testing schemes involving in vitro methodology is currently used for the classification of certain hazards (skin irritation/corrosion and eye irritation/serious eye damage). For other hazards, such as acute toxicity, validated alternatives that completely replace animal testing are not yet available, and acute tests that use fewer animals and or cause less suffering are internationally accepted and preferred to the conventional LD50 test.

86 Chapter 3

Chapter 3. Aims and Objectives

From the regulatory point of view, one objective of acute toxicity testing is to classify chemicals according to their intrinsic toxicity. This is conventionally conducted on the bases of the median lethal dose (LD50) (Section 2.2). LD50 values are used by government, regulatory authorities to place chemicals in hazard categories, which dictate appropriate labelling, exposure limits, handling, disposal and transport requirements of chemical materials (Section 2.2.1). However, since the end of the 1970’s this test has been widely criticised for scientific, animal welfare and regulatory reasons (Section 2.2.1.1).

Recent legislation in Europe calling for the ban of most animal testing for substances used in cosmetic products by 2009 and a total ban of all animal testing by 2013 has added to the urgency of developing and validating alternatives to animal testing (Daston et al., 2005). In addition, the implementation of the new EU chemicals policy and the development of the REACH system (Section 2.4.2) will result in the need for further testing of up to 30,100 existing chemicals and for which essential human health data are lacking. If testing is to be based on traditional methods, very large number of laboratory animals could be needed in response to the REACH system, causing ethical, scientific and logistical problems (Combes et al., 2003). Alternatives to animal testing are required to rapidly identify hazardous chemicals and to permit classification and labelling (Langley 2005).

The aim of this research is to demonstrate the validity of using in vitro toxicity testing with human cultures for setting safety standards for industrial/commercial products. Over the past decades, an increasing number of in vitro test systems for evaluating the possible toxicological hazards of compounds have been developed. Methods of in vitro cytotoxicity have demonstrated their applicability to toxicity testing and a number of in vitro methods have been validated, with a regulatory acceptance and are currently used in routine toxicity testing of chemicals (Section 2.2.3 and Section 2.2.4). These tests have the potential to provide a mechanistic basis for toxicity testing where toxicants act

87 Chapter 3

at a molecular cellular level, and they may permit the use of tissues from more- appropriate target species and individuals, including humans (Section 2.2.2). They are ethical, cost effective, and scientifically accurate and after validation are relevant to human risk assessment (Section 2.2.2).

This study will evaluate the ability of in vitro cytotoxicity tests to provide an accurate estimate of acute toxicity hazard according to current and internationally harmonised schemes (e.g. GHS). More over, there is a need for in vitro methods to help increase the accuracy of the estimates of chemical mixtures and their interaction (Gennari et al., 2004) (Section 2.3.3). The cytotoxicity tests will then be applied to a set of chemical mixtures, to evaluate possible toxicological interactions between chemicals. As a result potentially influencing the way risk assessments currently classify and label single chemicals and mixtures.

In order to achieve the goal of this study a set of test single chemicals and mixtures with a wide range of toxicity and physicochemical properties need to be selected and tested with various cytotoxicity assays, measuring different endpoints. Currently there is a move towards the implementation of the GHS and in the area of acute toxicity, the GHS currently relies on data from animal tests but need not be test-specific (Section 2.4.5). This system could be adapted to use non-animal data to inform classification and labelling decisions. Therefore a set of test chemicals were chosen to represent the 5 GHS categories and tested using a battery of in vitro cytotoxicity assays and human cell culture. A large number of in vitro models have been developed which are based on either a colorimetric or bioluminescence reaction. A variety of cytotoxicity assays exist which measure different endpoints, including the MTS, NRU, LDH and ATP assays (Section 2.2.5). These assays have an extensive use among researchers and are found to be reliable and sensitive assays (Section 2.2.5). The data will then be used to evaluate the predictive performance of the methods for correctly estimating all of the GHS hazard categories. The study will also be used to evaluate the reproducibility of the assays, and effectiveness of the cytotoxicity to predict in vivo rodent data with the ultimate goal of predicting human toxicity, as well as the detection of mixture interaction.

88 Chapter 3

Research questions for this study were divided into the following components:

1. What are the criteria for choosing these particular in vitro cytotoxicity assays? 2. How do the assays chosen compare with each other in terms of reproducibility, rapidity, simplicity and accuracy? 3. Are the cytotoxicity assays able to reliably distinguish chemicals in adjacent hazard categories, similar to the cut-off concentrations used by the GHS? 4. How does the performance of the in vitro cytotoxicity assays compare to established in vitro and in vivo toxicological methods? 5. What is the experimental design to be used for combining and testing chemical mixtures that is consistent, reliable and scientifically accurate? 6. What prediction models and analysis need to be used to accurately identify possible toxicological interactions, if they exist, between the chemical mixtures? 7. Can the results of these tests be used as a basis for ranking new chemicals and mixtures?

While the use of in vitro techniques in toxicological research is widespread relatively minor progress has been made in applying this knowledge to toxicity testing for regulatory purposes. Results from this study, will increase our knowledge in the field of in vitro toxicity testing as applied to chemicals and mixtures with the potential use of a battery of in vitro tests for regulatory purposes. The study outlines new methods for testing toxicological interaction between chemical mixtures. In addition, data produced from this study could potentially be added to an in vitro database for future use in the development, refinement and validation of other methods that could be a part of a new testing strategy for assessing the acute systemic toxicity of chemicals.

89 Chapter 4

Chapter 4. Materials and Methods

4.1 Test Materials

4.1.1 Chemical selection

The chemicals used in this study were selected to represent a wide range of toxicity from highly toxic (LD50 d 5 mg/kg) to unclassified toxicity (LD50 > 5000 mg/kg) (OECD 2001a) (Table 4-1). The existence of toxicity data and human exposure was also a major criterion for chemical selection.

Table 4-1 GHS categories for acute oral classification

Category Oral LD50 Signal word

1 d 5 mg/kg Danger

2 > 5 - d 50 mg/kg Danger

3 > 50 - d 300 mg/kg Danger

4 > 300 - d 2000 mg/kg Warning

5 > 2000 - d 5000 mg/kg Warning

Unclassified > 5000 mg/kg Unclassified

Twenty one chemicals were selected for testing across the six acute oral toxicity classification groups of the GHS (OECD 2001a) (Table 4-2). Chemicals were selected from a database of 72 chemicals compiled by a joint effort of

NICEATM and ECVAM (http://iccvam.niehs.nih.gov/ivcytoval/alt_chem.pdf27H83 ). This original database was used for an in vitro cytotoxicity validation study to generate in vitro toxicity data using NRU assays with rodent (mouse fibroblast 3T3) and human (normal human keratinocyte) cells.

90 Chapter 4

Table 4-2 Selected chemicals and their respective GHS categories for acute oral toxicity

GHS unclassified Category/ Chemicals 1 2 3 4 5

Mercuric Colchicine Cupric Phenol Potassium Ethanol chloride sulfate chloride

Sodium Potassium Cadmium Lithium Sodium Methanol Selenate cyanide chloride sulfate chloride

Strychnine Sodium Sodium SDS Irgasan Saccharin dichromate fluoride Caffeine Cobalt Glycerol chloride

4.1.2 Chemical information and mechanisms of toxicity

Unless otherwise specified, information on test chemicals tested in this project were taken from the Registry of Toxic Effects of Chemical Substances (RTECS, 2002); Hazardous Substances Database (HSDB 2002); Merck Index (Budavari 2000); Material Safety Data Sheets (MSDS) provided by manufacturers; and Sax’s Dangerous Properties of Industrial Materials (Lewis 2000).

Evaluations of carcinogenicity to humans was taken from the International Agency for Research on Cancer (IARC). This agency coordinates and conducts research on the causes of human cancer and develops strategies for cancer control (IARC, 2005). The classification scheme employed by IARC is shown in Table 4-3.

91 Chapter 4

Table 4-3 Evaluation scheme of carcinogenicity to humans taken from IARC (2005)

Characteristic Evaluation Scheme Group Carcinogenic to humans 1 Probably carcinogenic to humans 2A Possibly carcinogenic to humans 2B Unclassifiable as to carcinogenicity to humans 3 Probably not carcinogenic to humans 4

In total 21 chemicals representing different toxicity groups were tested, details of each chemical is summarised below:

x Cadmium II chloride hemi-pentahydrate (CdCl2 5/2 H2O) Exposure to cadmium chloride is a common occurrence in industry. Cadmium chloride is incorporated into a variety of alloys and metal plating. The average population is exposed to cadmium through food, water, air and smoking. It is listed as a suspected carcinogen and is a known nephrotoxicant. Cadmium chloride causes damage to the proximal tubular epithelium of the mammalian kidney both in vivo and in vitro (Gennari et al. 2003).

The LD50 for cadmium chloride is 88 mg/kg (oral rat) (NIOSH 2002). The GHS category for acute oral toxicity is 3.

x Caffeine (C8H10N4O2) Caffeine is a weakly basic alkaloid that occurs naturally in coffee and cocoa beans, kola nuts and tea leaves in amounts up to 2% by weight. The chemical name is 1, 3, 7- trimethylxanthine. An average cup of coffee or tea can contain from 40-100 mg of the drug (Stavric 1988). It is therapeutically used in humans as a mild central nervous system stimulant. It also acts on the kidney to produce diuresis, stimulates contractility of cardiac and skeletal muscle directly, and inhibits the contractility of the smooth muscle. Target organs in humans are the central nervous system and heart (HSDB 2002).

92 Chapter 4

The LD50 for caffeine is 192 mg/kg (oral rat) (NIOSH 2002). There is inadequate evidence for the carcinogenicity in humans of caffeine. There is inadequate evidence for the carcinogenicity in experimental animals of caffeine. Overall evaluation: Caffeine is not classifiable as to its carcinogenicity to humans (Group 3) (IARC 2004). The GHS category for caffeine acute toxicity is 3.

x Cobalt chloride hexahydrate (CoCl2.6H2O) Cobalt chloride is found in certain grades of steel and in tungsten carbide tools. It is also used as paint pigments and occasionally as therapeutic agents. Exposure to cobalt chloride has produced allergic dermatitis in workers. Its target organs in humans are the respiratory tracts and cardiovascular system. Features of chronic toxicity include effects on the heart, thyroid and possibly the kidney (Carson et al. 1986). It is thought that cobalt chloride and ethanol have an additive effect, since cardiomypathy was reported in heavy beer drinkers in the 1960s as a result of the use of cobalt chloride as a foam stabiliser (present in beer at concentration 1 – 1.5 ppm) (Sheghizzi et al. 1994). Cobalt without alcohol caused damage to the heart muscle, whilst both alcohol and cobalt reduced blood flow significantly. It has been suggested that the anoxia caused by the combination of alcohol and cobalt exacerbated to the cardiotoxic effects of cobalt (Sheghizzi et al. 1994). Since cobalt chloride is no longer used as an additive in beer, no further cases have been reported (EVM 2003).

The LD50 of cobalt chloride is 766 mg/kg (oral rat) (NIOSH 2002). It is possibly carcinogenic to humans; no sufficient evidence has yet been produced in humans or animals (IARC evaluation for cobalt chloride: 2B) (IARC 2004). The GHS category for its acute toxicity is 4.

x Colchicine (C22H25NO6) Colchicine is a naturally occurring alkaloid, found in flowers of the meadow saffron (Colchicium autumnale) at a concentration of about 0.1%. It is an antimitotic agent by its ability to arrest cell division in the metaphase resulting in death of the cell and is used in the treatment of cancer (Harris and Gillett 1998). Colchicine has an anti- inflammatory effect and is the drug of choice for acute gouty arthritis. It may have a direct toxic effect on muscle, peripheral nervous system and liver (HSDB 2002).

93 Chapter 4

The LD50 for colchicine is 25.53 mg/kg (oral mouse) (NIOSH 2002). The GHS category for acute toxicity for colchicine is 2.

x Copper sulfate pentahydrate (CuSO4.5H2O) Copper sulfate occurs naturally as an essential trace metal and has for a long time been used in clinical medicine as a therapeutic drug. Nowadays, its main use is in agriculture as a fungicide (Budavari 2000). Target organs in humans for cupric sulfate are the kidneys and liver. Copper functions as a constituent of a number of enzyme systems including ascorbic acid oxidase, lactase, and tyrosinase (HSDB 2002). It exerts its toxicity through a number of mechanisms: increasing red blood cell permeability through increasing oxidation of haemoglobin sulfhydryl groups leading to hemolysis and inhibiting enzymes important in protecting the cell from oxygen free radicals (HSDB 2002).

The LD50 for copper sulfate is 300 mg/kg (oral rat) (NIOSH 2002). The GHS acute toxicity category rating for cupric sulfate is 3.

x Ethanol (C2H6O) Ethanol is commonly used as an antiseptic, and solvent in the manufacture of pharmaceuticals and alcoholic beverages. Like caffeine and nicotine, ethanol is mainly a social drug and rarely used in therapeutics. Target organs for ethanol human toxicity are the central nervous system, kidneys and liver. Ethanol is a central nervous system depressant that causes stupor, coma and eventually death if ingested in excessive quantitities. A comprehensive review article on ethanol and its action on neurotransmitters, their receptors, and their metabolism in the brain was published (Dietrich et al., 1989). Ethanol is biotransformed to acetaldehyde and then to acetic acid in the liver. it is hepatotoxic and can also potentiate the hepatotoxicity of some chemicals and/or alter their clearance due to combined hepatotoxicity of ethanol and these chemicals. Increased bioactivation induced by ethanol, and other ethanol-induced effects on hepatic enzymes involved in detoxification, bioactivation or cellular protection, are shown in animals and in vitro (Snyder and Andrews 2001).

94 Chapter 4

The LD50 for ethanol is 7,060 mg/kg (oral rat) (NIOSH 2002). Human carcinogenicity of alcoholic beverages has been proven (IARC evaluation of alcoholic beverages: Group1) (IARC 2004), but ethanol is suggested not to act as an initiator but more as a cocarcinogen which interacts with other carcinogens to cause tumorigenic responses (Snyder and Andrews 2001). According to the American Conference of Governmental Industrial Hygienists (ACGIH®) ethanol is not classifiable as a human carcinogen (A4) (Lewis 2000; HSDB 2002) . The GHS acute oral toxicity category for ethanol is “unclassified”.

x Glycerol (C3H5(OH)3) Glycerol has a wide range of usage and application. Glycerol is one of the main components in the production of nitroglycerin (dynamite)(Budavari 2000; Lewis 2000). it may be produced from various fats and oils. Due to its hygroscopic properties it is a main ingredient in skin lotions, cosmetics and soaps. Glycerol is used as a sweetener in liqueurs and confectioneries (Lewis 2000). Glycerol is also used as a solvent and emulsifier and may be found in inks, adhesives, and pharmaceuticals. It can also be used in the manufacture of plastics and resins. Glycerol is an irritant to the eye, skin and mucous membrane. Its target organs in humans are the central nervous system, kidneys and liver. It is non-toxic after ingestion, except with very large doses, when it can cause headache, dizziness, nausea, vomiting, thirst, diarrhoea, and confusion (HSDB 2002). Glycerol had no systemic effects in humans after skin application (HSDB 2002).

The LD50 for glycerol is 4,090 mg/kg (oral mouse) (NIOSH 2002). The GHS category for acute toxicity for glycerol is category 5.

x Irgasan (C12H7Cl3O2) Irgasan is also known as triclosan or tetrachlorosalicylanilide is commonly used as a bacteriostat and preservative for cosmetic and detergent preparations (Budavari 2000). It is also medicinally used as an antiseptic and disinfectant, and can be found in toothpastes as an antimicrobial agent that reduces dental plaque (Babich and Babich 1997; Zuckrebraun et al. 1998; Budavari 2000; Lewis 2000). Human toxicity effects for this chemical have been limited to mild sensitization and mild itching to irgasan found in soap bars or deodorant foot powder. Irgasan also causes photodermatitis, with light

95 Chapter 4

causing the original material to be converted to a contact allergen (HSDB 2002). In animals optical and neural toxicity have been observed (HSDB 2002).

The LD50 for irgasan is 3,700 mg/kg (oral rat). The GHS category for acute toxicity is 5.

x Lithium sulfate hydrate (Li2SO4 H2O) Lithium sulfate hydrate occurs naturally as a mineral and is widely used in industrial and chemical processes. It is therapeutically used as antidepressant (Budavari 2000). Target organs in humans of lithium sulfate are the central nervous system and kidneys.

LD50 for lithium sulfate is 613 mg/kg (oral rat) (NIOSH 2002). The GHS acute toxicity category for lithium sulfate is 4.

x Mercury II chloride (HgCl2) Mercuric chloride is a highly toxic corrosive substance, used for preserving wood and anatomical specimens as well as in electroplating and other metal processes (Budavari 2000). Mercuric chloride is also used as topical antiseptic and disinfectant (Budavari 2000). Mercuric chloride is one of the most toxic inorganic mercury salts. Fatalities have occurred from exposure to as little as 0.5 grams. Mercuric chloride is a severe eye and skin irritant; it is corrosive to the eyes and throat (HSDB 2002) . It can be toxic or fatal by the inhalation, dermal, oral, or intrauterine exposure routes (HSDB 2002). Target organs in humans for mercuric chloride are the central and peripheral nervous systems (Lewis 2000; HSDB 2002). Mercuric chloride has caused similar effects in laboratory animals. In addition it has caused kidney damage in rats and horses (HSDB 2002). Mercuric chloride has been classified as a possible carcinogen (IARC evaluation value: 2B) (IARC 2004).

The LD50 for mercuric chloride is 1 mg/kg (oral rat) (NIOSH 2002). The GHS acute toxicity category for mercuric chloride is 1.

96 Chapter 4

x Methanol (CH3OH) Methanol occurs naturally in some woods, however, traces have been found in engine exhausts and cigarette smoke (Budavari 2000). Methanol is primarily used as solvent, anti-freeze, and fuel but it can also be utilized for organic synthesis in the chemical industry (Lewis 2000). Methanol is first converted in the human liver to formaldehyde and then formic acid. The latter two metabolites rather than methanol itself, that is highly toxic and produce severe metabolic acidosis, ocular symptoms, neurotoxicity and other effects of acute methanol poisoning (HSDB 2002). The main target in humans for methanol toxicity is the retina. It can cause reversible or permanent blindness at high doses (Snyder and Andrews 2001).

The LD50 for methanol is 5,630 mg/kg (oral rat) (NIOSH 2002). The GHS acute toxicity category for methanol is “unclassified”.

x Phenol (C6H6O) Phenol is mainly obtained from coal tar and mainly commercially used as a disinfectant and as an intermediate in chemical syntheses (Budavari 2000). Phenol in concentrations of 5% or greater will rapidly denature all proteins with which it comes in contact. Phenol also causes severe irritation and corrosion on contact with skin or other tissue. It is also rapidly absorbed upon dermal contact and can lead to death (Lewis 2000; HSDB 2002). Target organs of phenol in humans are the central nervous system, kidneys, spleen and respiratory tract (HSDB 2002). Incidences of human carcinogenic and teratogenic effects have been related to this substance. However, a classification of this substance on human carcinogenicity could not be agreed on (IARC evaluation value: 3) (IARC 2004).

The LD50 for phenol is 317 mg/kg (oral rat) (NIOSH 2002). The GHS acute category rating for phenol is 4. x Potassium chloride (KCl) Potassium chloride occurs naturally as a mineral and is used as fertilizer, electrolyte, and photographic developer. Potassium is also the chief intracellular cation and its bioregulation in the mammalian organism is critical to maintenance of proper function

97 Chapter 4

(Budavari 2000). Potassium chloride is also used in pharmaceutical preparations. Target organ of potassium chloride is the gastrointestinal tract (HSDB 2002).

The LD50 for potassium chloride is 2,600 mg/kg (oral rat) (NIOSH 2002). The GHS category for acute oral toxicity testing is 5. x Potassium cyanide (KCN) Potassium cyanide is industrially used as fumigants, insecticides, metal polishes and in electroplating solutions (Budavari 2000). Ingestion of as little as 50 milligrams of potassium cyanide has caused significant poisoning in humans. About 80% of cyanide dose is detoxified by conversion via the liver enzyme rhodanase to thiocyanate, and is subsequently secreted in the urine. Potassium cyanide causes hypoxia by the inhibition of cytochrome oxidase (HSDB 2002). In humans, acute cyanide exposure results primarily on the central nervous system (CNS), has cardiovascular and respiratory effects. While in vitro KCN inhibits cytochrome oxidase, and there by, cell respiration (Clemedson et al. 2002).

The LD50 for potassium cyanide is 10 mg/kg (oral rodent). The GHS category for acute toxicity of potassium cyanide is 2.

x Saccharin (C7H5NO3S) Saccharin is mostly used as a non-caloric synthetic sweetener in foods, pharmaceutical products, and toothpaste (Budavari 2000; Lewis 2000). Saccharin is also used in industry as an electroplating bath additive, and as cattle feed additive (Budavari 2000; Lewis 2000).

Saccharin has been a controversial chemical in terms of its carcinogenicity. The Human Health Assessment Group in EPA's Office of Health and Environmental Assessment has evaluated saccharin for carcinogenicity. According to their analysis, the weight-of- evidence for saccharin is group C, which is based on inadequate evidence in humans and limited evidence in animals. As a group C chemical, saccharin is considered to be possibly carcinogenic to humans. IARC evaluation of saccharin has placed it in group 3 (unclassifiable as to carcinogenicity in humans) (IARC 2004). Sodium saccharin

98 Chapter 4

produced urothelial bladder tumours in rats, but the mechanism was found not to be relevant for humans (IARC 2004).

The LD50 for saccharin is 14,200 mg/kg (oral rat). The GHS category for saccharin acute oral toxicity is “unclassified”.

x Sodium dichromate dihydrate (Na2Cr2O7 . 2H2O) Sodium dichromate is used in a number of industrial applications and as an oxidant in the manufacture of a number of substances: steel and non-ferrous alloys; metal-plating; refractory materials; chromate pigments and chromate preservatives (Budavari 2000). Hexavalent chromium salts are strong oxidizing agents which may cause corrosive burns by denaturation of tissue protein (HSDB 2002). In general, soluble hexavalent salts are less toxic in the presence of metabolic activation system, however, they are powerful inhibitors of respiratory chain activity (HSDB 2002). Soluble hexavalent chromium compounds are taken up into the cells by simple diffusion through the plasma membrane and the sulfate transport system. Rapid reduction to the trivalent form then occurs with the action of enzymatically mobilized electrons. The detoxification of hexavalent chromium compounds occurs in the saliva, red blood cells, epithelial lining fluid, liver cells, and pulmonary alveolar macrophages (Clayton and Clayton 1994). Hexavalent chromium is reduced to trivalent form within the skin by methionine, cystine, and cysteine. Within the skin, trivalent chromium is cleared at a slow rate and can combine with proteins or other skin components to form whole skin allergens (HSDB 2002). Hexavalent chromium poisoning of the kidneys can lead to a decrease of ascorbic acid content. Such decrease may reduce the oxidative protection to the tissue and increase the susceptibility to toxicity (HSDB 2002). Target organs of sodium dichromate are the respiratory tract, kidney, eyes, and skin (Goyer 2001).Hexavalent chromium has been confirmed as a human carcinogen, with an ACGIH grouping A1 (HSDB 2002) and IARC classification is Group 1 (IARC 2004).

The LD50 for sodium dichromate is 50 mg/kg (oral rat) (NIOSH 2002). The GHS category for acute toxicity of sodium dichromate is 2. x Sodium chloride (NaCl)

99 Chapter 4

Sodium chloride is a common salt that naturally occurs as a mineral. It is used in various industrial and chemical processes as well as an electrolyte replenisher for therapeutic purposes (Budavari 2000). Together with water, NaCl constitutes the basic internal milieu of mammals. Sodium chloride contributes to stability of extracellular and intracellular osmolarity; the sodium ion is distributed virtually exclusively in the extracellular compartment. An ingestion of 0.5-1 g/kg of NaCl can be toxic to most patients (HSDB 2002). It can cause skin, eye and respiratory irritations (HSDB 2002). Acute salt poisoning produces central nervous system damage when brain cells become dehydrated following the acute osmotic shift of intracellular fluids to the extracellular space.

The LD50 for sodium chloride is 3,000 mg/kg (oral rat) (NIOSH 2002). The GHS category for acute toxicity is 5. x Sodium fluoride (NaF) Sodium fluoride is commonly used in electroplating, disinfecting fermentation apparatus in breweries and distilleries, and in dental labs (Budavari 2000). Sodium fluoride is a direct cellular poison and interferes with calcium metabolism and enzyme mechanisms. It also diminishes tissue respiration, decreases oxygen consumption and carbon dioxide production in muscle(HSDB 2002). The fluoride ion can act as an anticoagulant by lowering the plasma calcium concentration in the blood.

The LD50 for sodium fluoride is 180 mg/kg (oral rodent) (NIOSH 2002). The GHS category for sodium fluoride acute oral toxicity is 3.

+ x Sodium lauryl sulfate (SLS) or (SDS) (CH3-(CH2)11-O-SO3-Na ) SLS is commonly found in shampoos for cleansing of the scalp in the treatment of dandruff, seborrheic dermatitis and psoriasis, in skin lotions and dental toothpastes (Babich and Babich 1997; Budavari 2000). SLS is also known as sodium dodecyl sulfate (SDS). SLS acts as an emulsifying, wetting dispersing agent in creams, lotions and medical preparations, and as surfactant in foods (Babich and Babich 1997). SLS also has the ability to increase the permeability of the stratum corneum, which is helpful for medicaments but it will also allow permeability to noxious agents and thus may

100 Chapter 4

directly or indirectly produce irritation (HSDB 2002). SLS can produce sensitive allergic sensitivity reactions; skin dryness; eye irritation (HSDB 2002).

The LD50 for SLS is 1288 mg/kg (oral rat) (NIOSH 2002). The GHS category for acute oral toxicity is 4.

x Sodium selenate (Na2SeO4) Sodium selenate is commonly used as a feed additive. The minimum human lethal dose has not yet been delineated (HSDB 2002).

The LD50 for sodium selenate is 1.6 mg/kg (oral rodents) (NIOSH 2002). The GHS category for sodium selenate acute oral toxicity is 1.

x Strychnine (C21H22N2O2) Strychnine is an alkaloid that is found in seeds of Strychnos nux-vomica, a tree indigenous to India. It is mainly used as a pesticide, and it has also been used as a tonic to improve circulation and muscle-tone (Budavari 2000). Strychnine is a potent central nervous system stimulant and convulsant by selectively blocking of postsynaptic neuronal inhibition (INCHEM 2005). Strychnine competitively blocks the binding of the inhibitory neurotransmitter glycine at its postsynaptic receptor sites on motor neurons of the ventral horn of the spinal cord (INCHEM 2005).

The LD50 of strychnine is 2 mg/kg (oral rodent) (NIOSH 2002). The GHS category for strychnine acute oral toxicity is 1.

101 Chapter 4

Table 4-4 Test chemicals

Test chemical CAS RN1 MW2 Supplier Purity (%) Cadmium dichloride 7790-78-5 228.36 May & Baker Ltd. 98.0 Caffeine 58-08-2 194.20 Sigma 99.0 Cobalt chloride 7791-13-1 237.90 Sigma 98.1 Colchicine 64-86-8 399.45 Sigma 98.0 Cupric sulfate 7758-99-8 249.70 Sigma 99.2 Ethanol 64-17-5 46.07 Ajax 100 Glycerol 56-81-5 92.09 Chem. Supply 99.0 Irgasan 3380-34-5 289.54 Fluka 97.0 Lithium sulfate 10102-25-7 127.95 Merck 99.0

Mercuric chloride 7487-94-7 271.50 AnalaR 99.5

Methanol 67-56-1 32.04 Sigma-aldrich 99.9 Phenol 108-95-2 94.11 Sigma 99.0

Potassium chloride 7447-40-7 74.55 Ajax >99.8

Potassium cyanide 151-50-8 65.12 Ajax >96.0

Saccharin 81-07-2 205.20 Sigma 99.0

Sodium chloride 7647-14-5 58.44 Chem.-supply 99.0

Sodium dichromate 7789-12-0 298.00 Riedel-de Haen >99.5

Sodium fluoride 7681-49-4 41.99 AnalaR 99.0

Sodium lauryl sulfate 151-21-3 288.40 Sigma 99.0

Sodium selenate 13410-01-0 188.90 Sigma-Aldrich 99.9

Strychnine 57-24-9 334.40 Sigma 99.0

1CAS RN: Chemical Abstracts Services Registry Number. 2MW: Molecular Weight (g/mol.)

102 Chapter 4

4.2 In vitro Cytotoxicity Analysis

4.2.1 Human cell culture

Results obtained in general cytotoxicity tests have shown to be independent on the choice of cell type (Reinhardt et al. 1985; Seibert et al. 1996; Spielmann et al. 1999; Gad 2000; Clemedson et al. 2002; Gribaldo et al. 2005). Several studies, such as the MEIC study, have shown that cell culture tests gave similar results irrespective of the cell type, therefore almost any cell type could be used for measurement of basal cytotoxicity (Ekwall et al. 1989; Ekwall 1998; Clemedson and Ekwall 1999; Ekwall 1999a; Clemedson et al. 2002; Gribaldo et al. 2005). The basic cellular functions and structures, which are fundamental to all types of cells, have been found to form a common basis for the evaluation of general toxic effects of chemicals (Ekwall et al. 1989; Balls and Fentem 1992). However, the MEIC study also indicated that human cell lines are more suitable for detecting cytotoxicity than cells from animal origin. A long term advantage of using human cells is that cytotoxicity results can be added to human toxicity databases to facilitate development of methods to predict acute human lethality (ICCVAM 2001a). It was beyond the scope of this study to find the most sensitive cell line to single or different groups of chemicals.

The human cells used for this study were primary skin fibroblasts. Fibroblasts are one of the human cells most frequently used for basal cytotoxicity with good results in validation studies (Gettings et al. 1996; Harbell et al. 1997; Malich et al. 1997; ICCVAM 2001a; ICCVAM 2001b). Primary cell cultures of human skin fibroblasts were obtained from fresh skin biopsies taken from the arms of healthy individuals undergoing genetic screening who returned normal results (Cytogenetics Department, The New Children’s Hospital, Westmead, Sydney, Australia).

103 Chapter 4

4.2.2 Culturing method

Cells were cultured in phenol red free culture medium DMEM/F12 (Dulbecco’s modified essential medium/Ham’s 12 nutrient mixture) (Gibco), supplemented with 5% (v/v) foetal calf serum (JSBioscience, Australia) and containing 0.5% of an antibiotic mix: L-glutamine (2mM), penicillin (100 U/ml) and streptomycin (0.1 mg/ml) (Sigma,

USA). Cultures were maintained at 37qC in a humidified 5% CO2 incubator. Cells were in the exponential growth phase at the time of testing, and once confluence was reached the flasks were rinsed three times with sterile HBSS (Hank’s Balanced Salt Solution) (Gibco). Cells were removed from the culture surface using standard methods by the addition of trypsin-EDTA (Gibco). The viability of the cells exceeded 95% on all occasions as determined by the Trypan Blue dye exclusion method (trypan blue solution 0.4% (w/v), Sigma) using a Neubauer Haemacytometer (Weber, UK) and a light microscope (Leitz Laborlux 12 pol). Trypan blue (50 μl) are added to an equal amount of cell dilution (50μl) concentration. Proliferation curves were derived to determine the seeding densities of the cells to performed adequate cytotoxicity test (details in Chapter 5: Section 5.2.5). The concentrations of skin fibroblasts resulted in cell suspension concentrations of (250 000 cells/ml; MTS); (250 000cells/ml; NRU); (100 000 cells/ml; ATP) and (350 000 cells/ml; LDH).

4.2.3 Preparation of test chemicals

Test chemicals were freshly prepared each time, immediately before use. Test chemicals were prepared in stocks by dissolving in 10ml of culture medium DMEM/F12 and sterilised through 0.22 ȝm filters. This sterile stock solution was then transferred into 96 well microplates (Greiner bio-one), and serial dilutions were prepared according to the appropriate dilution protocol (Section 4.2.4). A range finding experiment was undertaken to determine the range of test chemical concentration that will cause an effect prior to performing a more specific assay at the range of dilutions around the IC50.

When the IC50 was not detected in the dosage range used, the experiment was repeated and the test chemical concentrations were adjusted accordingly until the complete dose

104 Chapter 4

response curve was generated. An example of the range of chemical concentrations is displayed in Table 4-5 for each experiment.

Non soluble chemicals (irgasan; mercuric chloride and strychnine) were dissolved in ethanol such that the final application to the cells was 0.5% (v/v) in the medium only column and in all the ten dilution concentrations. This concentration is considered to be non-toxic to the cells (NTP and NICEATM 2003b). All other chemicals were dissolved in culture medium DMEM/F12.

4.2.4 Dilution protocol

The dilution protocols were adopted from previously published papers and thesis (Malich et al. 1997; Malich 1998; Malich et al. 1998; Zarei and Markovic 2000; Hayes and Markovic 2002).

4.2.4.1 Dilution protocol (i)

In general, 40Pl of chemical was removed from chemical stock solutions and added to 60Pl of media in a serial dilution (column 3-12) ( Figure 4-1). Cells were then added in a ratio of 40:60 (cell suspension: test chemical solution), such that the total volume inside the wells of the 96 well microplates was a constant 100Pl.

4.2.4.2 Dilution protocol (ii)

In general, 25ȝl of chemical was removed from chemical stock solutions and added to 75ȝl of media in a serial dilution (column 3-12) (Figure 4-1). Cells were then added in a ratio of 25:75 (cell suspension: test chemical solution), such that the total volume inside the wells of the 96 well microplates was a constant 100ȝl.

105 Chapter 4

Table 4-5 Test chemical concentrations for the study of 21 GHS reference chemicals

Chemical C2chemical [mM] [mg/l] Cadmium chloride 10.09 - 0.01 2400 -1.00 Caffeine 24.72 - < 0.01 4800 -1.26 Cobalt chloride 21.02 - <0.01 5000 - 0.04 Colchicine 4.69 - <0.0011 1875 - 0.01 Cupric sulfate 2.88 - <0.0011 720 - 0.18 Ethanol 6511- 0.68 300 000- 31.45 Glycerol 1563 - 0.41 144 000 - 37.74 Irgasan1 10.36 - <0.01 3000-0.01 Lithium sulfate 37.51-0.01 4800-1.00 Mercuric chloride1 2.83 -<0.01 768-0.20 Methanol 9363 - 0.98 300 000 - 31.45 Phenol 127.5-0.03 12000 - 3.14 Potassium chloride 321.9-0.08 24000 - 6.29 Potassium cyanide 36.86 - 0.01 2400 - 0.629 Saccharin 116.96 - 0.03 24000 - 6.29 Sodium chloride 410.6 - 0.10 24000-6.29 Sodium dichromate 3.15 - <0.01 937.5 - 0.01 Sodium fluoride 57.16 - 0.01 2400 - 0.63 Sodium lauryl sulfate 8.32 - 0.01 2400 - 0.63 Sodium selenate 25.41 - 0.02 12000 - 3.14

Strychnine1 7.18 - 0.01 2400 - 0.63

1: chemical dissolved in ethanol, remaining chemicals were dissolved in culture medium. 2: chemical concentrations in cell solution.

106 Chapter 4

1 2 3 4 5 6 7 8 9 10 11 12

A Medium Cells Cells and dilution series of chemicals B Medium Cells

C Medium Cells

D Medium Cells

E Medium Cells Background absorbance of chemicals

F Medium Cells

G Medium Cells

H Medium Cells

Figure 4-1 Microtitre plate design for MTS, NRU and ATP assays

107 Chapter 4

4.2.5 Controls

Three internal controls were set up for each experiment (Figure 4-1), they consisted of: (1) Medium only control as a measure of 0% cell viability; (2) Cells only control as a measure of 100% cell viability; (3) background control in case of interference between chemical and the performed adequate cytotoxicity test.

Experiments for each chemical were repeated on three separate occasions, to assess for inter-test variations and reproducibility.

4.2.6 In vitro cytotoxicity protocols

Measurable endpoints for cytotoxicity are well established and have been used to assess cytotoxicity (Spielmann et al. 1999; ICCVAM 2001a). Four cytotoxicity indicators were developed in this study to form a battery of tests. These included: MTS (Promega, USA), Neutral Red Uptake (NRU; Sigma, USA), Adenosine Triphosphate (ATP; Promega, USA) and Lactate Dehydrogenase release (LDH; Promega, USA) assays. Details on each of the assays can be found in the Literature Review Section 2.2.1. The MTS and ATP assays are a measure of cell viability as metabolic indicators; NRU is a measure of decreased cell viability as a membrane marker and the LDH release assay is a more specific assay determining the release of intracellular components, typically detecting damage of the outer cell membrane (ICCVAM 2001a). All four assays have been extensively used in cytotoxicity studies (refer to Section 2.2.1).

4.2.6.1 The MTS assay

® The MTS assay was performed using the CellTiter 96 AQueous Non-Radioactive Cell Proliferation Assay (Promega). It is a colorimetric assay that measures formazan production in viable cells by the mitochondria (conversion of a tetrazolium salt to formazan product) in conjunction with the addition of an electron-coupling reagent PMS, (Sigma, USA) (Promega 2001a) (Section 2.2.1.1).

108 Chapter 4

Preliminary studies and published in vitro studies using the MTS assay, have established a linear relationship between formazan production and incubation time and between formazan production and cell number (Malich et al. 1997; Zarei and Markovic 2000).

Three internal controls were set up for each experiment using the MTS assay, as seen per Section 4.2.5. The results obtained from background absorbance values were then deleted from the MTS absorbance values to account for potential chemical interference.

Incubation and exposure period: Test chemicals were prepared as stock solutions as per Section 4.2.4. This sterile stock solution was then transferred into 96 well microplates as per dilution protocol (i) (Section 4.2.4.1). Cells were then added in a ratio of 40:60 (cell suspension: test chemical solution), resulting in a concentration of cells of 10,000 cells/well. The activated MTS/PMS working solution (20ȝl) (details in Appendix A1) was then added into a 100ȝl of serial dilutions of test chemicals and cell culture solutions (preserving a ration of 1:5) into the prepared microtitre plates. The 96 well microtitre plates were then incubated for 4 hours at 37°C in a humidified incubator with 5% CO2. An incubation period of 4 hours was chosen to reflect acute toxicity for determining in vitro cytotoxicity of test compounds, based on previous studies (Malich et al. 1997; Malich et al. 1998; Promega 2001a). This exposure period, was chosen to represent acute exposure conditions, and minimised the loss of volatile test chemicals during incubation.

Measurement of the colorimetric assay: After the 4 hours exposure period, absorbance was read using a Multiskan MS plate reader (Labsystem, Finland) at an absorbance of 492nm. Absorbance values for each test compound and its dilutions were plotted on a dose response curve, either in terms of absorbance as per Figure 4-2 or in terms of % cell viability against chemical concentration as per Figure 4-3. In order to determine the IC50 values (50% Inhibitory Concentration that causes reduction by half of the activity of mitochondrial dehydrogenases), NOEC (No Observable Effective Concentration) values and TLC (Total Lethal Concentrations).

109 Chapter 4

0.9 y = 0.632e-0.001x NOEC R2 = 0.986

0.6

IC50

0.3

Absorbance (492nm) Absorbance TLC

0 1 10 100 1000 10000 Concentration (mg/L)

Figure 4-2 MTS dose-response curve (phenol) (Absorbance vs. Concentration)

120

100

80 y = 96.552e-0.0011x R2 = 0.9864 60

40 % Cell Viability

0 1 10 100 1000 10000 Concentration (mg/L)

Figure 4-3 MTS dose-response curve (phenol) (% cell viability vs concentration) Figures (4.2 – 4.3): Each point represents the m ± SD of 4 replicates for each chemical concentration.

110 Chapter 4

4.2.6.2 The NRU assay

The neutral red (NR) cytotoxicity assay is a cell survival/viability assay, based on the ability of viable cells to incorporate and bind neutral red, a supravital dye (Basic Red, Toluene red) (NTP and NICEATM 2003). Neutral Red Assay measures lysosomal integrity, as it is a weak cationic dye that will penetrate into the cell membranes through non-ionic diffusion and accumulate intracellularly. Hence any changes to the cell surface or the sensitive lysosomal membrane will lead to lysosomal fragility and other irreversible changes (Section 2.5.2).

The protocol used was adopted and modified from Borenfreund and Babich (1992). After 24 hours incubation of cells, the medium was aspirated from the incubated 96- microtitre well plate. Chemical stock solutions were prepared as per Section 4.2.3 and added to a separate 96 well microtitre plate in serial dilution using dilution protocol (i) (Section 4.2.4.1). The dilution was then transferred to the washed 96-well plate containing the seeded cells. Microtitre plates containing seeded cells and chemical dilution were then placed at 37°C in a 5% CO2 humidified incubator and left for an exposure period of 4 hours. After 4 hours exposure time, the medium was removed and the microtitre plates were rinsed with HBSS (200Pl). An aliquot of 200Pl NR medium (Appendix A1) was then added to the wells. The plates were placed in a humidified incubator at 37°C, for 2 hours. This was found to be enough for NR uptake to take place. During the incubation, cells were briefly observed for neutral red crystal formation, and rejected if excessive neutral red crystallization had occurred.

After incubation period was completed, the neutral red medium was removed, and cells were washed with a fixative solution (100ȝl) (Appendix A1). The wells were then rinsed with HBSS (200Pl). The neutral red solubilisation (100ȝl) (Appendix A1) was then added to all wells.

96 well microtitre plates were then left to stand for 10 minutes at room temperature, and then placed in a gyratory shaker for 20 minutes to extract the neutral red from the cells and form a homogeneous solution.

111 Chapter 4

Absorbance was then read using a Multiskan MS plate reader at a dual absorbance of 540nm, and 690nm. Measurements for the 690nm represented the background absorbance and were subtracted from the 540nm measurements.

Three internal controls were set up for each experiment using the NRU assay, as seen per Section 4.2.5 and Figure 4-1. Coloured compounds did not pose a particular problem with this assay because the medium with the test agent(s) was removed and the cells were washed prior to the addition of the neutral red medium.

Experiments for each chemical were repeated on three separate occasions, to assess for inter-test variations and reproducibility.

4.2.6.3 The LDH assay

The LDH assay was performed using the Cytotox 96 Non-Radioactive cytotoxicity assay, provided by Promega. It is a colorimetric assay that quantitatively measures the release of lactate dehydrogenase (LDH); a cytosolic enzyme present in all mammalian cells (Promega 2004). LDH is not released through the plasma membrane in to the extracellular fluid unless damage has occurred to the membrane. In vitro release of LDH from cells provides an accurate measure of cell membrane integrity and cell viability (Allen and Rushton 1994).

In order to assess the effects of a chemical on mammalian cell culture, cells were exposed to the chemicals for 24 hours (see Section 5.2.7). LDH in culture supernatants was then measured using a 30-minute coupled enzymatic assay (details for reagents Appendix A1), resulting in the conversion of a tetrazolium salt (INT) into a red formazan product. The amount of colour formed was proportional to the number of lyzed cells. Absorbance was calculated at 492nm using a 96-well plate reader. This method was used as an alternative to the traditional radioactive 51Cr release assay (Promega 2004). The protocol for the LDH assay was adopted and modified from Allen et al., (1994) and Promega (2004).

112 Chapter 4

Preparation of standard cytotoxicity plate: Cells were cultured in culture medium

(DMEM/F12) (as per Section 4.2.2) and maintained at 37°C, in humidified 5% CO2 incubator. Cells were added in quadruplets from wells 1-12 as can be seen in Figure 4-4. Post-incubation, microtitre plate was removed, and the medium was aspirated. Stock solutions of chemicals were prepared in stocks by dissolving in 10ml culture medium DMEM/F12, modified from Section 4.2.2, in this case the culture medium was supplemented with only 1% Charcoal Stripped Bovine Serum (JSBioscience, USA), to minimize background absorbance due to lactate dehydrogenase enzyme already present in the serum (detailed explanation found in Section 5.2.6). Chemicals were then added to a separate 96-well microtitre plate in serial dilution as per dilution protocol (ii) (Section 4.2.4.2). The dilution was then transferred to the washed 96-well plates containing the seeded cells. Microtitre plates containing the seeded cells and chemical dilution were then placed at 37°C in a 5% CO2 humidified incubator for an exposure period of 24 hours (see Section 5.2.7). Three internal controls were set up for each experiment as per Section 4.2.5, they consisted of: (1) Medium only control as a measure of 0% cell viability; (2) Cells only control as a measure of 100% cell viability; (3) background control in case of interference between chemical and the lactate dehydrogenase enzyme. Experiments for each chemical were repeated on three separate occasions, to assess for inter-test variations and reproducibility.

1 2 3 4 5 6 7 8 9 10 11 12

A Medium Cells

B Medium Cells

C Medium Cells

D Medium Cells Background absorbance of chemicals

Figure 4-4 Preparation of standard cytotoxicity plate for the LDH assay

113 Chapter 4

LDH measurement: After the defined exposure time, media containing the chemical was aspirated (culture supernatant) and transferred onto a new 96-well microtitre plate, for the measurement of extracellular LDH released due to chemical induced damage on cell membranes. The original plate was then rinsed with HBSS (200ȝl). Fresh culture medium (100ȝl) was added to the plates to which a lysis solution (10ȝl) consisting of Triton X-100 (Appendix A1) was then added to lyse the cells. The plate was then placed in 5% CO2 humidified incubator for 45 minutes. The intracellular LDH (cell lysate), that was still present in the viable cells was then collected upon cell lysis, and added in quadruplets from columns 1-12 in the lower part of the 96-well plate containing the extracellular LDH, as can be seen in Figure 4-5. Reconstituted enzymatic substrate mixture (details in Appendix A1) was then added to the plate, covered with foil and left for 30 minutes at room temperature. Absorbance was read at 492nm; using a Multiskan MS plate reader. In order to correct for any endogenous LDH activity in animal serum, the A492nm value of the medium only control column was subtracted from all other values.

LDH levels were measured in both the culture supernatant and the cell lysate. The percentage of LDH released was calculated by dividing the level of LDH in the medium by the total amount of LDH in the medium and cell lysate. This is shown in the following Equations (4.1- 4.2) (Allen and Rushton 1994):

Test A492 % LDH release u 100 [4-1] Total A492

Test A492 % LDH release = u 100 [4-2] Test A492+cell lysate A492

Test A492 is a measure of extracellular LDH released upon chemical induced damaged on the cell membrane; Cell lysate A492 is a measure of the intracellular LDH that remained inside viable cells.

114 Chapter 4

1 2 3 4 5 6 7 8 9 10 11 12

Medium Cells A Extracellular LDH

B Medium Cells

C Medium Cells

D Medium Cells Background absorbance of chemicals E Medium Cells Intracellular LDH F Medium Cells

G Medium Cells

Medium Cells H Background absorbance of chemicals

Figure 4-5 Microtitre plate design for LDH measurement

115 Chapter 4

120 100 0.001x 80 y = 16.548e 2 60 R = 0.979 40

% LDH release 20 0 0 0 0 1 10 100 1000 10000 Concentration (mg/L)

Figure 4-6 Dose-response curve (Irgasan) of plotted % LDH released % LDH released (Equation 4-1) is plotted against chemical concentration.

1 y = 0.3027e0.001x 2 0 R = 0.9932 0 0 0 1 10 100 1000 10000 Extracellular LDH (492 nm) Concentration (mg/L)

Figure 4-7 Dose-response curve (Irgasan) for extracellular LDH released Extracellular LDH (Test A492) is plotted against chemical concentration.

y = 1.6455e-0.0015x R2 = 0.9737

Cell lysate nm) (492 0 0 0 0 1 10 100 1000 10000 Concentration (mg/L)

Figure 4-8 Dose-response curve (Irgasan) for intracellular LDH (cell lysate) Intracellular LDH (cell lysate A492) is plotted against chemical concentration.

Figures (4.6 – 4.8) represent an example of LDH release dose-response curves for chemical (Irgasan). Each point represents the m ± SD of 3 replicates for each chemical concentration.

116 Chapter 4

4.2.6.4 The ATP assay

The ATP assay was conducted using the CellTiter-Glo Luminescent Cell Viability Assay (Promega 2001b), which quantifies the amount of ATP levels in viable cells. The amount of ATP in cells correlates with cell viability. Once damage has occurred to the cell membranes, within minutes, the cell will lose its ability to synthesize ATP, and endogenous ATPases destroy any remaining ATP, thus the levels of ATP fall rapidly (Riss et al. 2003). This bioluminescent assay depends on the ability of a thermostable luciferase (Ultra-GloTM recombinant Luciferase), which catalyses the formation of light from ATP and luciferin (Promega 2001b). The “glow-type” signal can be recorded with a luminometer, CCD camera or modified fluorometer. It generally has a half-life of 5 hours, providing a consistent signal across large batches of plates (Riss et al. 2003). The protocol was performed on opaque 96-well micro plates (Nunc, Denmark).

Cells and serial chemical dilutions were prepared as per Sections 4.2.3 and 4.2.4. Three internal controls were set up for each experiment as per Section 4.2.5 and Figure 4-1. Microtitre plates containing seeded cells and chemical dilution were then placed at 37°C in a 5% CO2 humidified incubator and left for an exposure period of 4 hours. Post exposure, and incubation period opaque plates were removed from incubator and equilibrated for 30 minutes at room temperature. The CellTiter-Glo® Reagent (100ȝl) was added at an equivalent volume to the cell culture medium present in each well. Plates were then placed in a gyratory shaker for 2 minutes to induce cell lysis. The plate was then left to stand for 10 minutes at room temperature to stabilize luminescence signal. The luminescence signal was read using an Orion Luminometer (Berthold Detection System, Australia).

117 Chapter 4

4.2.7 Statistical analysis used for the calculation of cytotoxicity values

Cytotoxicity values (NOAEC, IC50 and TLC) were calculated from dose-response curves generated using 4 assays (MTS, NRU, LDH and ATP). Since cells and serial dilution of chemicals were added in quadruplets within the same experiment standard errors were first calculated based for the replications, for each chemical at each concentration versus absorbance. Cytotoxicity values were then extrapolated using exponential regression analysis based on a derivative equation derived from exponential equation (Y = ae -bx) for MTS, NRU and ATP assays, and an exponential equation (Y = ae +bx) for the LDH assay. NOAEC values were determined at the point where the normalized cell viability showed no cytotoxic effect (flat region of the curve); IC50 was determined at 50% of the maximum response; and TLC values were determined at the maximum response of the normalized cell viability.

4.2.7.1 Calculation of standard error for MTS, NRU and ATP dose response curves

The net mean of Absorbance (Ǔi) as a function of concentration (i), was obtained by using Equation 4-3:

Yyeiii [4-3]

ǔi is the mean of the signals, as a measure of the number of absorbance (nsi) wells containing cells and dilution series of chemicals (Figure 4-1: columns 3-12);

Ɲi represented the mean of the signals as a measure of background absorbance of

chemicals obtained through (nbi = 4) repeated measures (Figure 4-1: column 1).

The variance of the mean (s) was then calculated using Equation 4-4 (the equations required to derive Equation 4-4 are found in Appendix B1).

118 Chapter 4

22 sY iii 1/2 ¬ª sy se ¼º [4-4]

4.2.7.2 Calculation of standard error for LDH dose response curve

In the case of the LDH assay, % LDH released (pi) was plotted against concentration of the test chemical (ci) (Figure 4-6). It was calculated as a fraction of the absorbance of the extracellular LDH released (Ǔi) over total absorbance (ǓTi) such that:

pYY100iiTI [4-5]

And the variation (s) of pi was then obtained by Equation 4-6 (method used to derive Equation 4-6 are found in Appendix B1):

22 2 222 2 spiiiiiiiI1 pª s(y) s(e) Y s(y) s(e) s(y)iT Yiº 2 ¬ ¼

[4-6]

ǓIi represents the intracellular LDH (cell lysate)

ei represented the mean of the signals as a measure of background

absorbance of chemicals obtained through nbi repeated measures Yi absorbance of extracellular LDH

YTI total absorbance

119 Chapter 4

4.2.7.3 Calculation of IC50 values (MTS, NRU, ATP and LDH)

(i) Calculation of IC50 values for MTS, NRU and ATP assays

For dose-response curve using the regression equation (Y = ae -bx) for the exponential curves, the following equations were derived for the calculation of IC50 values:

ln y lna bx [4-7]

ln a / y x [4-8] b

ln 2 IC50 [4-9] b y is the absorbance values or expressed as percentage cell viability; a is the maximum response of normalized cell viability; such that a = ymax b is the curve slope; x is the concentration on normalized cell viability;

IC50 values were determined in which y was 50% of the maximum response on normalized cell viability. a = ymax (or the value of y when x = 0). In this case IC50 is the x obtained when y = ymax/2 = a/2 leading to Equation 4-9.

(ii) Calculation of IC50 values for the LDH assay

For dose-response curve using the regression equation (Y = ae +bx) for the exponential curves (see Figure 4-6), the following equations were derived for the calculation of IC50 values:

ln y lna bx [4-10]

120 Chapter 4

ln y /a x [4-11] b

ln 2 IC50 [4-12] b y is the absorbance values or expressed as percentage LDH released; a is the maximum response of normalized cell viability; such that a = ymin b is the curve slope; x is the concentration on normalized cell viability;

IC50 is the x obtained when y = 2a, resulting in Equation 4-12.

4.2.7.4 Calculation of NOEC

Independent of the method followed, upon observation of the mean absorbance values

Ǔi, the means remain constant for a certain range of concentrations that have similar levels of high cell viability. Therefore the graph was delimited in this range of concentrations, to calculate the mean ǓN of signals obtained in this range, which is then

used to calculate NOEC value from the regression equation. As an example, Figure56H 4-9 is the concentration-effect curve of copper sulfate derived using MTS assay. We -0.0325*x obtained the Ǔi and the regression equation of the form (y = 1.4552*e ). Upon observation Ǔi is practically constant for concentrations of CuSO4 1.5 mg/L. The calculated average in this range is ǓN = 1.4 mg/L and modifying Equation 4-8 for the calculation of NOEC to derive Equation 4-13.

NOEC Ln a YN b [4-13]

121 Chapter 4

y = 1.4552e-0.0325x R2 = 0.9827 1.6 1.4 1.2 1 0.8 ǓN 0.6 0.4 Absorbance 492 nm 0.2 0 0.00 0.01 0.10 1.00 10.00 100.00 Concentration (mg/L)

Figure 4-9 Dose-response curve for copper sulfate (MTS assay)

4.2.7.5 Calculation of TLC values

(i) TLC values for dose-response curves (MTS, NRU and ATP)

The absorbance measured for these curves decreased with increased chemical concentration. The regression was in the form of (y = a*e-b*x). As indicated in Section 4.2.4.1, it is obtained from the means of absorbance for each concentration ‘i’ (Equation 4-3). The total lethal concentration (TLC) was taken as the concentration that corresponded to the signal that differed significantly from the noise (background) of the analytical method. Therefore it corresponds to the limit of the detection of the method. Equation 4-8 was modified to calculate the total lethal concentration (TLC) using Equation 4-11.

122 Chapter 4

TLC Ln a / YTLC b [4-14]

ȏTLC represents net mean absorbance of total lethal concentration (derivation method found in Appendix B1)

(ii) Calculation of TLC for curves generated using LDH assay

LDH has a regression equation in the form of (y = a*e+b*x). In the case of the LDH assay, the absorbance (signal) value increased with the toxicity of the chemical (Figure 4-6). Such that 100% LDH release is indicative of total cell death. The TLC is then calculated for y = 100 and Equation 4-11 is modified:

TLC Ln(100 / a) b [4-15]

4.3 Study of the Cytotoxicity of Chemical Mixtures

While conventional toxicology testing focuses on single chemicals, human exposures are usually to more than one chemical. Most regulatory agencies use the default assumption that the risk of exposures to more than one chemical is treated in an additive manner. This approach may under- or over-estimates the risk of chemicals depending on their different modes of action or interaction. Therefore, one objective of combination toxicology is to establish whether exposure to a mixture of chemicals will result in an effect similar to that predicted on the basis of additivity. The predictive ability of toxicological information on single substances is limited when applied for risk assessments of chemical mixtures (Teuschler et al. 2002). In vitro assays are a useful approach to assess risks posed by chemical mixtures since their value in order to detect and rank the toxicity of chemicals has already been established (Galli et al. 1993; Marinovich et al. 1996).

Cytotoxicity data of the mixtures was obtained using both the MTS and NRU assays. The data obtained was compared to the individual independent effect of each component, and to data calculated with predictive mathematical algorithms. The

123 Chapter 4

observed mixture toxicity was then compared with the predictions by applying the concentration addition concept (details in Section 2.3.1). This concept rests on the assumption that the components of a mixture act in a similar way, such that one can be replaced by an equal fraction of an equi-effective concentration of another, without diminishing the overall mixture effect.

4.3.1 Test materials

4.3.1.1 Chemical selection

Ternary and binary chemical mixtures were composed of test chemicals that covered a broad toxicity range. Individual IC50 toxicity values for 15 chemicals whose in vivo toxicity was spread over the five Globally Harmonised System (GHS) categories for acute oral toxicity were used to create 5 ternary chemical mixtures and their respective binary combinations (15) (Table 4-6). The 15 chemicals were chosen from the list of 21 chemicals already tested using 4 in vitro cytotoxicity assays (MTS, NRU, LDH and ATP) (refer to Sections 4.1 and 4.2).

Table 4-6 Selected chemical mixtures and their respective GHS categories Mixtures Chemical GHS category

Mixture 1 Sodium Fluoride 3 SLS 4 Irgasan 5 Mixture 2 Sodium dichromate 2 Copper sulfate 3 Phenol 4 Mixture 3 Cadmium chloride 3 Lithium sulfate 4 Ethanol Unclassified Mixture 4 Mercuric chloride 1 Potassium cyanide 2 Cadmium chloride 3 Mixture 5 Sodium fluoride 3 Phenol 4 Potassium chloride 5

124 Chapter 4

4.3.2 In vitro cytotoxicity analysis for chemical mixtures

4.3.2.1 Human cell culture

Human skin fibroblasts were used for this study and cultured as per Section 4.2.1 and 4.2.2.

4.3.2.2 Preparation of test chemicals

Concentrations for the individual chemicals in mixtures were chosen based on an estimation of equi-toxicity, ensuring that no chemical contributed disproportionately to the overall combination effect.

In general for any binary or ternary mixture of 3 toxicants, A and B and C the previously experimentally derived IC50 of each toxicant in the mixture was used as an index of toxicity. The toxicants were prepared such that their fractional effects individually in the mixture was calculated in proportion of their concentration to the total concentrations of all toxicants in the mixture and such that the sum of all ratio combinations equalled to a theoretical additive value of 1 (Equation 4-16). For chemicals A, B and C then

IC50 (A) IC50 (B) IC50 (C) 1 [4-16] IC50 (mix) IC50 (mix) IC50 (mix)

IC50 (A, B or C) denotes 50% inhibitory concentration for chemical A, B or C when administered individually and calculated with the respective in vitro tests.

125 Chapter 4

IC50 (mix) is the sum of the total individual IC50 concentrations for the chemicals. For chemicals A, B and C, for e.g. the mixture ratio of chemical A = (1 –

(IC50B/IC50mix- IC50C/IC50 mix) and so forth the remaining chemicals.

Equation 4-16 can then be modified to calculate starting concentrations for mixture combinations for n components and for an effect x (Equation 4-17).

n Ci ¦ 1 [4-17] i1 Cmix

Ci denotes the concentration of an agent in a mixture yielding an effect x, produced experimentally on its own.

Cmix is the sum of the concentration of n components yielding an effect x when administered independently.

For each chemical Xi denotes the fraction of the component i in the mixture such that

Xi = Ci/Cmix.

Tables 4-7 and 4-8 display the ratios of concentrations (Xi) for the test chemicals in weight per volume for the relative assays (MTS, NRU), in order to obtain the strength of each chemical when in mixture. A multiplication factor ranging from 10 000 to 100 000 was applied to the calculated concentrations for ease of preparation.

126 Chapter 4

Table 4-7 Selected ternary mixture combinations

Ternary mixture components C* chemical

MTS assay NRU assay

Sodium Fluoride 0.770 0.598 Sodium Lauryl Sulfate 0.085 0.051 Irgasan 0.145 0.351 Sodium dichromate 0.087 0.240 Cupric sulfate 0.019 0.259 Phenol 0.894 0.502 Cadmium dichloride 0.032 0.001 Lithium sulfate 0.323 0.182 Ethanol 0.645 0.817 Mercuric chloride 0.002 0.016 Potassium cyanide 0.111 0.907 Cadmium dichloride 0.887 0.077 Sodium fluoride 0.063 0.116 Phenol 0.085 0.315 Potassium chloride 0.852 0.569

C*: stock concentrations in g/10ml derived according to Equation 4-13 used to combine the ternary mixtures

127 Chapter 4

Table 4-8 Selected binary mixture combinations

Binary mixture components C* chemical MTS assay NRU assay Sodium fluoride 0.875 0.921 Sodium lauryl sulfate 0.125 0.079 Sodium fluoride 0.805 0.630 Irgasan 0.195 0.370 Sodium lauryl sulfate 0.370 0.127 Irgasan 0.630 0.873 Sodium dichromate 0.817 0.481 Cupric sulfate 0.183 0.519 Sodium dichromate 0.089 0.323 Phenol 0.911 0.677 Cupric Sulfate 0.021 0.340 Phenol 0.979 0.660 Cadmium dichloride 0.091 0.005 Lithium sulfate 0.909 0.995 Cadmium dichloride 0.048 N/A Ethanol 0.952 N/A Lithium sulfate 0.333 0.182 Ethanol 0.667 0.818 Mercuric chloride 0.021 0.018 Potassium cyanide 0.979 0.982 Mercuric chloride 0.003 0.177 Cadmium dichloride 0.997 0.823 Potassium cyanide 0.111 0.922 Cadmium dichloride 0.889 0.078 Sodium fluoride 0.424 0.269 Phenol 0.576 0.731 Sodium fluoride 0.068 0.170 Potassium chloride 0.932 0.830 Phenol 0.091 0.356 Potassium chloride 0.909 0.644

C*: stock concentrations in g/10ml derived according to Equation 4-16 used to combine the mixtures N/A: not applicable due to the huge difference in IC50 between CdCl2 and Ethanol

128 Chapter 4

The test chemicals were prepared as stock solutions (as per Section 4.2.3). Irgasan was not soluble in culture medium and was first dissolved in ethanol so that the final concentration of this solvent did not exceed 0.5% in the maximum chemical solution. This ethanol concentration is considered non-toxic (NTP and NICEATM 2003b). Test chemicals were then filter sterilized (0.22Pm), and added in serial dilution to 96-well microtitre plates. In general 50ȝl of chemical mixture was removed from stock and added to 60ȝl media in serial dilution, and then the cells were added in a ratio of 40:60 (cell suspension: test chemical solution), such that the total volume inside the wells was a constant 100ȝl and a dilution factor of 0.5.

Parameters for the determination of quantitative cytotoxicity data of chemical mixtures was adopted from the study of the 21 GHS reference chemicals using human skin fibroblast cell cultures. They are summarized in Table 4.9. Derivations of cell number values are taken from Chapter 5 (Section 5.2.5).

Table 4-9 Parameters for the study of the cytotoxicity of chemical mixtures

Parameters Assays

MTS NRU

Cell number (/well) 10 - 20 x 103 15 - 20 x 103

Incubation of cells (h) 4 24

Exposure (h) 4 4

Incubation with reagent (h) 4 4

4.3.2.3 Protocol for MTS and NRU Assays

The protocol for the study of mixture toxicity using both cytoxicity assays (MTS and NRU), was similar to that used for the individual chemicals, and as described in Sections 4.2.6.1 and 4.2.6.2.

129 Chapter 5

Chapter 5. Results and Discussions: Cytotoxicity of Test Chemicals

5.1 Introduction

In this study, a battery of in vitro cytotoxicity tests were used to evaluate the cytotoxicity of selected substances spread across the GHS (categories 1-5 and unclassified), representing different classes of chemicals with diverse physicochemical properties. cytotoxicity studies were evaluated with reference to the GHS categories as the GHS will be adopted globally in the next few years (Gennari et al. 2004). Any validation efforts for acute toxicity methods should also evaluate the predictive performance of the methods for correctly estimating all of the GHS hazard categories (Gennari et al. 2004).

The aim of this study was to evaluate the use of four different types of in vitro assays for detecting the cytotoxicity of chemicals. The study will be used to evaluate the reproducibility of the assays, and effectiveness of cytotoxicity to predict in vivo rodent data with the ultimate goal of predicting human toxicity.

These assays included the colourimetric MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3- carboxymethoxy-phenyl)-2-(4-sulfophenyl)-2H-tetrazolium; Promega®), NRU (Neutral Red Uptake; Sigma), enzymatic LDH (Lactate Dehydrogenase; Promega®) and the ATP (CellTiter-Glo® Luminescent Cell Viability; Promega ®) Assays. Results were expressed as NOEC (no observable effect concentration), IC50 (50% inhibitory concentration) and TLC (total lethal concentration) values. A comparison between the assays’ sensitivity and specificity, and reproducibility was undertaken. Experimentally derived in vitro data were compared to in vitro values published in the literature and databases. Multiple in vitro – in vivo comparisons were performed between cytotoxicity values (NOEC, IC50) against published rodent LD50 values, and published human

130 Chapter 5

toxicity values: Lethal Dose Lowest (LDL0) and Lethal Plasma Concentration (LC). A correlation of NOEC values was also performed against published occupational exposure limit values (TLV: Threshold Limit Values). The predictive performance of the methods for correctly estimating all of the GHS hazard categories was also studied.

5.2 Experimental Design

5.2.1 Human cell culture

The use of human cells has the potential to provide more accurate data to humans and eliminate the need to extrapolate from animal experimentation to humans. Primary human skin fibroblast cell cultures were used for this study (Section 4.2.1). Fibroblasts have been extensively used for basal cytotoxicity with good results in validation studies (Chapter 4; Section 4.2.1). Fibroblasts were cultured and maintained as per Section 4.2.2.

5.2.2 Test chemicals

Twenty one chemicals were selected for testing across the six acute oral toxicity classification groups of the GHS (Chapter 4; Table 4-2). Chemical information and mechanism of toxicity for the selected chemicals are presented in Chapter 4 (Section 4.1.2). Chemical dilutions were prepared for experiments as per Section 4.2.3.

5.2.3 Assays

The four assays (MTS, NRU, LDH and ATP) used throughout this study were based on protocols detailed in Chapter 4 as per Section 4.2.6. Table 5-1 provides a summary of the mechanism of the cytotoxicity assays.

131 Chapter 5

Table 5-1 Cytotoxicity assays endpoints

Assay Endpoint Measured Type of cytotoxicity information obtained

MTS Measures metabolic activity in the cell as a function of its Cell viability and cell number mitochondrial activity

NRU Measures uptake of dye into lysosomes of living cells; active Cell viability and cell number endocytosis. LDH Measures membrane integrity through the release of the LDH Damage to cell plasma enzyme due to cell damage. membrane ATP Measures metabolic activity in the cell as a function of its Cell viability, cell number and cellular ATP levels. estimation of biomass in growing cell lines

5.2.4 Dose response curve

Dose response curves were plotted for selected test chemicals after correction by subtracting the background absorbance from the controls (Section 4.2.5). NOEC, IC50 values and TLC were derived from the plotted absorbance data based on statistical method detailed in Chapter 4 (Section 4.2.7).

5.2.5 Optimisation of cell numbers

(i) Optimisation of cell number for the MTS assay

The objective of this study was to develop and refine the test conditions which are adequate for toxicological studies for the MTS assay. Initial experiments were carried out with primary human cell culture of skin fibroblast to establish the relationship between optimal absorbance and cell number for the MTS assay. The dependence of formazan production on cell concentration was determined. A linear relationship was established between cell number and formazan production (Figure 5-1). A linear correlation between absorbance and cell concentration was found at a range of 0 to 200,

132 Chapter 5

000 cells/ml. For this range linear regression analysis reveals R2 = 0.9984. Coefficients of determination (R2) tended to decrease for cells greater than 500,000 skin fibroblasts/ml (R2 < 0.91) resulting in a deviation from a trend reaching a plateau (Appendix A2). The latter indicated that the cell numbers were too high in order to provide constant rates at which the MTS detection reagents were transformed into a coloured formazan product.

2.0

1.5

1.0

0.5 y = 0.1035x + 0.0253 R2 = 0.9984

Absorbance (492 nm) (492 Absorbance 0.0 0 5 10 15 20 -0.5 Cell number (x104 cells/ml)

Figure 5-1 Linear range for skin fibroblast cell number against absorbance (MTS assay).

The effect of cell number on formazan production was examined. Each point represents the m ± SD of 4 replicates for each cell number. A linear relationship was established between cell number and absorbance (R2 = 0.99), for 4h incubation and exposure to reagent. The ultimate range of cell number was determined between 150,000 and 200,000 skin fibroblasts/ml.

(ii) Optimisation of cell number for the NRU assay

The neutral red assay is a cell viability assay based on the ability of viable cells to incorporate and bind neutral red (NR), a supravital dye. Alterations of the cell surface result in a decreased uptake and binding of NR. An increase in cell number generally resulted in increased colour formation. A linear correlation between absorbance and cell concentration was found at a range of 0 to 250, 000 cells/ml. For this range linear regression analysis reveals R2 = 0.9859 (Figure 5-2). The correlation tended to decrease

133 Chapter 5

between 250,000 – 300, 000 skin fibroblasts/ml (R2 < 0.97). Both a cell number greater than 300,000 cells/ml resulted in a plateau of the dose response curve or lower than 50,000 cells/ml indicating the cell numbers were either too high or too low to provide consistent rates (Appendix A2). It is also interesting to note that whereas the curve with the MTS assay does not deviate from linearity even at an optical density of 2.0, the ultimate absorption reached with the plate-reader using the NRU assay is at a maximum optical density 1.0. This is consistent with other studies conducted with the NRU assay, where absorbance was found to correlate linearly with cell number over a specific optical density range of 0.2 to 1.0 (Barile 1994a).

0.5

y = 0.0443x - 0.0626 R2 = 0.9859 Absorbance (540 nm) (540 Absorbance

0 0 5 10 15 20 25

Cell number (x104 cells/ml)

Figure 5-2 Linear range for skin fibroblast cell number against absorbance (NRU assay)

The effect of cell number on neutral red dye incorporation into cell lysosomes was examined. Each point represents the m ± SD of 4 replicates for each cell number. A linear relationship was established between cell number and absorbance (R2 = 0.98).The ultimate range of cell number was determined between 100,000 and 250,000 skin fibroblasts/ml.

134 Chapter 5

(iii) Optimisation of cell number for the ATP assay

The Promega CellTiter-Glo® Luminescent Cell Viability Assay kit was used to quantitatively measure the presence of ATP. The luminescent signal produced was proportional to the presence of metabolically active cells. In Figure 5-3 a direct relationship was seen with luminescence measured, which was expressed in relative light units (RLU) and the number of cells in culture. An optimal coefficient of determination (R2 > 0.99) was determined for a cell number up to 100 000 cells/ml. This correlation decreased for cell numbers greater than 200 000 cells/ml (R2 = 0.97) (Figure 5-3). In comparison with MTS, NRU, and LDH assay, the ATP assay measured luminescence at cell number even below the detection limits of those assays. With the luminescent signal for 40,000 skin fibroblasts/ml being 32 times above the background signal resulting from serum supplemented medium without cells. This was consistent with published studies where ATP was found to be more sensitive, with a much lower cell number detectable (Petty et al. 1995; Promega 2001b; Eirheim et al. 2004; Mueller et al. 2004).

150000

100000

y = 8061.4x + 2809.4 50000 R2 = 0.9948 Luminiscence (RLU) 0 0 5 10 15 20 Cell number (x 104 cells/ml)

Figure 5-3 Skin fibroblast cell number against absorbance (ATP assay)

A direct relationship exists between luminescence measured and the number of cells in culture (R2 > 0.99) for 4h incubation and exposure to reagent. Each value represents the m ± SD of 4 replicates for each cell number. The ultimate range of cell number was determined between 40,000 to 200,000 skin fibroblasts/ml.

135 Chapter 5

(iv) Optimisation of cell number for the LDH assay

The correlation between the initial concentration of skin fibroblast cells and the LDH activity of culture supernatant after 24 hours incubation was determined (Appendix A2). The background LDH activity increased with increasing cell concentration due to continuous LDH release from the cells. The relationship was limited to 350,000 cells/ml (R2 > 0.99) (Figure 5-4). Test points plateaued at concentrations greater than that, with a decrease in the coefficient of determination (R2 < 0.80) (Appendix A2).

1.6

1.2

0.8

y = 0.0307x + 0.0396 0.4 R2 = 0.9986 Absorbance (492 nm) 0 02040 Cell number (x 104 cells/ml)

Figure 5-4 Skin fibroblast cell number against absorbance (LDH assay)

Skin fibroblast cells were seeded at the various concentrations and incubated for 24h. All data represent m ± SD for 4 replicates at each cell concentration. A direct relationship exists between cell number and absorbance (LDH activity) (R2 > 0.99).

5.2.6 Optimisation of serum concentration

The effect of the serum addition in the culture medium needed to be further studied in the case of the LDH assay. Culture medium for MTS, NRU and ATP assays was prepared as per Section 4.2.2. However, in the case of the LDH release assay, any serum used for cell culture experiments also had LDH activity (Baba et al. 2005). This could cause a considerable effect on the results of the LDH assay leading to a high

136 Chapter 5

background (Absorbance > 0.450 nm). An appropriate serum concentration was determined. Cells were cultured in culture medium supplemented with 1% Animal Bovine Serum (ABS) or 5% Foetal Calf Serum (FCS). After an incubation of 24 hours, the LDH activity in culture medium was measured by absorbance. Absorbance of the culture medium containing 5% FCS was high and could not be used for measurement (Appendix A3). In contrast, the medium containing 1% ABS had low LDH activity as background, and was found appropriate for the assay. Triton X-100 was also added to completely lyse the cells, and for the complete release of the LDH enzyme (Appendix A3). In general, in order to correct for any endogenous LDH activity in animal serum, generated background absorbance values in the medium only column (as per Section 4.2.5) were subtracted from all other values.

. 5.2.7 Optimisation of exposure time

Cells were exposed for 4 hours to chemicals as an indication of acute exposure conditions (refer to Chapter 4, Sections 4.2.6.1, 4.2.6.2 and 4.2.6.4). Based on previously published studies and protocols the 4 hour exposure represented acute exposure conditions, and minimized the loss of volatile test chemicals during incubation (Malich et al. 1997; Malich et al. 1998). However, in the case of the LDH assay one of the important conditions that considerably affect the results of enzymatic assays is the length of exposure of the chemical. Therefore, the LDH activity of culture medium alone, and medium supplemented with randomly selected test chemicals (KCl, NaF and SDS) from the set of 21 of chemicals was determined for 4 hours and 24 hours exposure periods. With the exception for the culture medium alone, the LDH activity increased with increasing the exposure period, reaching 100% LDH release in the 24 hours chemical exposure. Better absorbance was observed with the 24 hours chemical exposure, with no significant change in the slope of the curve (Figures 5-5 and Appendix A4). The noticeable change was in the total amount of LDH released at the high concentrations of chemical exposure. The literature contains many studies, with 24 hours exposure of cells to chemicals such as studies by Lobner (2000) and Issa et al., (2004). The LDH assay is a measure of membrane integrity (Chapter 2; Section 2.2.5.3), and the cell may reach necrosis or secondary necrosis (if the chemical induced damage

137 Chapter 5

through apoptosis), and this usually occurs at 24 hour exposure (Riss et al. 2003). In addition, the LDH molecule is large (140,000 MW) and in general more damage needs to be caused to the cell for LDH leakage to occur (Niederau et al. 1995).

100

24hrs exposure 50 4hrs exposure % LDH leakage

0 0 1 100 10000 Concentration (mg/L)

Figure 5-5 Effect of 4h and 24h exposure for KCl on %LDH leakage in culture supernatants

Skin fibroblast cells were treated with serial diluted concentrations of KCl. % LDH leakage was determined at 4h and 24h exposure. All data represent m ± SD for 4 replicates at each concentration.

5.2.8 Inactivation of LDH activity with certain chemicals

In theory IC50 values determined by the extracellular assay of LDH (LDH released upon chemical exposure) should be comparable to the assay measuring intracellular LDH (LDH remaining in the cell) after exposure to the chemical (Eirham et al., 2004). However, partial inactivation of the LDH activity was observed in the case of mercury chloride, copper sulfate and cobalt dichloride. The chemicals appeared to inhibit the release of the lactate dehydrogenase enzyme.

A possible explanation for the phenomena was that the chemicals were not cleaving the plasma membrane of the skin fibroblasts, and therefore the enzyme was not being released. But then, when the treated cells were lysed using Triton x-100, there was a decrease in lactate dehydrogenase remaining in the cells, compared to the cells that had been treated with a lower concentration of the chemical (Table 5-2).

138 Chapter 5

Table 5-2 Extracellular and intracellular LDH absorbance values for cells treated with CuSO4,

HgCl2 and CoCl2.

Chemical Concentration Extracellular LDH Abs Intracellular LDH Abs (mg/L) (492 nm) (492 nm) 2+ Cu SO4 562.50 0.06 0.15 2+ Cu SO4 0.002 0.30 1.40 2+ Hg Cl2 56.25 0.06 0.07 2+ Hg Cl2 0.001 0.26 1.35 2+ Co Cl2 375.00 0.15 0.18 2+ Co Cl2 0.001 0.20 0.86

Concentration values represent both highest tested chemical concentrations to which the skin fibroblast cells were exposed to, and the lowest chemical concentration. Each absorbance value represents the mean absorbance of 4 replicates for 3 separate experiments.

In the case of CuSO4 the cells were treated with a serial dilution (n = 4 replicates at each concentration) of the chemical; highest concentration of 562.5 mg/L and lowest 0.002 mg/L. The experimental protocol for the LDH assay is explained in detail in Chapter 4 (Section 4.2.6.3). After 24 hours chemical exposure, the culture supernatant was removed from the 96-well plate and the substrate reagent added to it, to allow for the enzymatic reaction to take place, and the measurement of the extracellular LDH. In the original plate, cells were washed to remove traces of chemical, and then treated with Triton X-100, to produce cell lysis, for the measurement of intracellular LDH (amount of LDH remaining in the cell post chemical treatment). When absorbance was measured for the extracellular LDH, there was no colour formation, and the absorbance was quite low (Absorbance = 0.06 nm; Table 5-2). But when absorbance was taken for the measurement of intracellular LDH that remained in the cell after exposure to chemical, the absorbance was also low (Absorbance = 0.15 nm) and a dose-response relationship could be seen. This demonstrated that in the wells where cells had prior exposure to higher chemical concentration, there was minimum LDH remaining in the cell. This helped in the deduction that an extracellular release of LDH had occurred, but that the formazan product was not being produced, therefore no colour formation.

The LDH enzymatic reaction equation involves the conversion of a tetrazolium salt, 2- p-(iodophenyl)-3-(p- nitrophenyl)-5-phenyltetrazolium chloride (INT), into a formazan

139 Chapter 5

product. The reaction (Figure 5-6) is catalysed by LDH released from the cells and diaphorase present in the assay substrate mixture.

LDH Lactate + NAD+ ĺ Pyruvate + NADH Diaphorase NADH + INT ĺ NAD+ + Formazan (red) Figure 5-6 LDH enzymatic reaction

Copper, Mercury and Cobalt are present in the compounds as cationic electrophiles (Cu2+, Hg2+, Co2+), and they can form a covalent bond with the nicotinamide (Lewis structure) present in the assay substrate mixture (Figure 5-7) (Carey 1996). This covalent binding is practically irreversible, and in this case can take the following form 2+ 2+ [Cu (NR3)] . Hence, deactivating the diaphorase reaction required for the conversion of a tetrazolium salt into a coloured formazan product.

Cu+2 or Hg+2 or Co+2

Figure 5-7 Covalent bond between NADH and selected metals Possible covalent bond that could be occurring between the cationic ions (Cu2+, Hg2+ or Co2+) present in copper sulfate, mercuric chloride and cobalt chloride, therefore inhibiting the following reaction to take place with INT to allow the formation of the coloured formazan.

A similar problem was faced in a study conducted by Eirheim et al., 2004. The study evaluated different toxicity assays on proliferating cells, and compared these results to the permeability enhancing effect of adding glycocholate (GC). The researchers in the study found that they could not use data from extracellular LDH, because they found that GC was inhibiting the released enzyme.

140 Chapter 5

5.3 Cytotoxicity Results

The in vitro cytotoxicity result (IC50) on human skin fibroblast cell culture as determined with the MTS, NRU, LDH and ATP assays are presented in Table 5-3. The remaining in vitro cytotoxicity results (NOEC and TLC) are summarized in Tables (Appendix C1). All values (mM) were expressed as mean and ± standard deviation (m ± SD). The most toxic chemical was found be mercuric chloride for all assays. Mercuric chloride is an extremely strong toxicant and classified in category 1 of the GHS. On the other hand, the weakest toxicant as measured with the cytoxicity assays was glycerol and unclassified in the GHS.

5.3.1 Ranking of test chemicals using the in vitro assays

The IC50 data values for each chemical, which were the result of a repeat of at least 3 separate experiments were compared with each other. The chemicals tested were ranked in order of their toxicity concentrations required to produce 50% cytotoxicity (IC50 values) (Table 5-4), and Spearman’s rank correlation coefficient was calculated to assess the correlation between the particular results and their respective assays (Table 5- 5). The closest relationship was found for the 21 ranked chemicals with the NRU and LDH assays (r = 0.922, p<0.01) A weaker but still strongly significant relationship existed between NRU and ATP (r = 0.899; p<0.01) followed by LDH and MTS (r = 0.838, p<0.01) then by NRU and MTS (r = 0.828, p<0.01) and LDH and ATP (r = 0.816, p<0.01). The weakest relationship but still in a relatively good agreement was between MTS and ATP (r = 0.698, p<0.01). In Figures 5-8 to 5-10 assays were correlated with each other, and the coefficient of determination (R2) was determined. R2 is the percentage of the results that can be explained by the regression (y = a * x + b) obtained between the two sets of data. Such that the correlation coefficient between the two sets of data r = ¥R. This also allowed us to determine visually, the outlier chemicals in certain conditions.

141 Chapter 5

Table 5-3 Cytoxicity data of test chemicals as determined by MTS, NRU, LDH and ATP assays

Chemicals MTS IC50 NRU IC50 LDH IC50 ATP IC50 mmol/L 1 mmol/L1 mmol/L 1 mmol/L

Cadmium chloride 6.58 r 0.88 0.04 r 0.01 0.01±0.00 1.44 ± 0.97 Caffeine 6.94 r 1.72 15.86 r 3.43 9.52 r 3.37 71.38 r 0.15 Cobalt chloride 0.06 r 0.01 4.95 r 0.84 1.23 r 0.56 29.14r 0.35 Colchicine 3.10 r 1.07 5.30 r 3.01 6.75 r 1.67 21.69 r 0.00 Cupric sulfate 0.11 r 0.04 0.81 r 0.22 0.58 r 0.18 1.53 r 0.29 Ethanol 668.6 r 144.7 255.5 r 43.18 168.5±25.61 150.4±0.07 Glycerol 1045 r 295.6 1505 r 0.00 108.5 r 0.00 1075r0.00 Irgasan 1.00 r 0.24 1.29 r 0.61 2.68 r 0.30 1.79 r 1.23 Lithium I sulfate 195.6 r 68.96 15.05 r 2.61 63.92 r 12.05 65.78 r 16.42 Mercuric chloride 0.03 r 0.01 0.01 r 0.00 0.04 r 0.03 0.01 r 0.00 Methanol 2644 r 832.6 432.6 r 0.00 309.0±0.00 1081±76.4 Phenol 18.41 r 0.00 4.51 r 0.16 10.52 r 0.00 5.26 r 0.00 Potassium cyanide 8.71 r 6.02 5.48 r 0.74 10.32±0.55 9.41±1.74 Potassium chloride 232.4 r 77.48 25.83 r 4.47 100.7 r 13.42 309.9 r 0.00 Saccharin 121.9 r 43.00 33.78 r0.00 39.88 r 3.32 84.45 r 17.9 Sodium Chloride 296.5 r 98.83 39.53 r 0.00 118.6 r 0.00 125.2 r 9.32 Sodium chromate 0.10 r 0.01 0.72 r 0.32 0.02 r 0.00 1.80 r 0.20 SLS 0.10 r 0.02 0.15 r 0.03 0.05 r 0.01 0.16 r 0.02 Sodium fluoride 24.89 r 2.27 6.71 r 1.95 17.73 r 1.06 8.93 r 0.34 Sodium Selenate 67.27 r 28.03 3.67 r 0.00 188.9±1.79 41.11±6.25 Strychnine 22.46 r 2.99 3.42 r 2.21 2.31 r 0.58 27.64 r 9.77

1 IC50 data for skin fibroblasts represent the m ± SD for at least three separate experiments.

142 Chapter 5

Table 5-4 Ranking of chemicals based on their respective cytotoxicity assays

MTS NRU LDH ATP

Mercuric chloride 1 1 2 1 Cobalt dichloride 2 11 6 12 Copper sulfate 3 4 5 5 SLS 4 3 4 2 Sodium dichromate 5 5 3 8 Irgasan 6 8 11 7 Potassium Cyanide 7 7 8 9 Sodium fluoride 8 6 9 4 Colchicine 9 12 14 18 Caffeine 10 17 13 17 Cadmium chloride 11 2 1 3 Phenol 12 9 12 6 Strychnine 13 13 10 16 Sodium Selenate 14 10 7 10 Potassium chloride 15 15 15 19 Sodium chloride 16 16 16 13 Lithium sulfate 17 14 18 14 Saccharin 18 18 19 18 Ethanol 19 19 17 11 Methanol 20 20 20 20 Glycerol 21 21 21 21

Chemicals were ranked from 1 to 21, in descending order of toxicity, based on mean IC50 values of the respective assays.

Table 5-5 Correlation coefficient (r) of ranked chemicals MTS NRU LDH ATP MTS r 1.000 0.829 0.838 0.698 p (2-tailed) N/A <0.01 <0.01 <0.01 N 21 21 21 21 NRU r 0.829 1.000 0.922 0.899 p (2-tailed) <0.01 N/A <0.01 <0.01 N 21 21 21 21 LDH r 0.838 0.922 1.000 0.816 p (2-tailed) <0.01 <0.01 N/A <0.01 N 21 21 21 21 ATP r 0.698 0.899 0.816 1.000 p (2-tailed) <0.01 <0.01 <0.01 N/A N 21 21 21 21

The chemicals tested were ranked in order of their toxicity, and Spearman’s rank correlation coefficient to assess the correlation between the different assays was determined (SPSS 10.0, Inc.). The correlation was found to be significant at the 0.01 level (2-tailed).

143 Chapter 5

In Figure 58, chemicals ranked with the NRU assay were correlated against the LDH and MTS assay (R2 = 0.70 and R2 = 0.75) respectively. In Figure 5-9, chemicals ranked with the ATP assay were correlated against results from MTS and NRU assays (R2 = 0.50; R2 = 0.80) respectively. Finally, Figure 5-10, chemicals ranked with the LDH assay were correlated against results from NRU and ATP assays (R2 = 0.85; R2 = 0.65) respectively.

30 g 20 rankin H

, LD NRU vs MTS 10 LDH vs MTS NRU

0 0 102030 MTS ranking

Figure 5-8 Correlation of chemicals based on ranking: NRU and LDH against MTS.

This is a regression plot between chemicals ranked from 1 to 21 based on their mean NRU and LDH IC50 (mg/l) values against the MTS assay. The Coefficient of determination (R2) was determined for NRU vs. MTS, LDH vs. MTS (R2 = 0.70 and R2 = 0.75) respectively. The line represents a 1:1 correlation.

15 MTS vs ATP NRU vs ATP 10

MTS, NRU ranking 5

0 0 5 10 15 20 25 ATP ranking

Figure 5-9 Correlation of chemicals based on ranking: MTS, NRU against ATP.

This is a regression plot between chemicals ranked from 1 to 21 based on their mean MTS and NRU IC50 (mg/l) values against the ATP assay. The Coefficient of determination (R2) was determined for MTS vs. ATP, and NRU vs. ATP (R2 = 0.50 and R2 = 0.80) respectively. The line represents 1:1 correlation.

144 Chapter 5

15 NRU vs LDH ATP vs LDH 10

NRU, ATPranking 5

0 0 5 10 15 20 25 LDH ranking

Figure 5-10 Correlation of chemicals based on ranking: NRU, ATP against LDH.

This is a regression plot between chemicals ranked from 1 to 21 based on their mean NRU and ATP IC50 (mg/l) values against the LDH assay. The Coefficient of determination (R2) was determined for NRU vs. LDH, and ATP vs. LDH (R2 = 0.85 and R2 = 0.65) respectively. The line represents 1:1 correlation.

5.3.2 Correlation between assays

The individual IC50 values were assessed between the four different assays using one way analysis of variance (ANOVA) at the significance 95%, Į = 0.05 to determine the differences among the assays for each chemical (p < 0.05). Further analysis was done to evaluate if any assay caused the significant differences in toxicity. Results from the one way ANOVA (at significance level 95%, Į = 0.05) are summarised in Table 5-6. The assays presented significant difference (p < 0.05) for most tested chemicals with the exception of: mercuric chloride, irgasan, glycerol and potassium cyanide (Table 5-6). In addition certain chemicals gave similar results between the LDH and NRU assay (cadmium chloride and copper sulfate); NRU and MTS assays (sodium dichromate and colchicine); NRU and ATP assays (phenol, SLS and sodium fluoride); ATP and LDH assays (ethanol, lithium sulfate and sodium chloride); ATP and MTS assays (sodium selenate and saccharin) and only one chemical was not significantly different between the MTS and LDH assays (caffeine).

145 Chapter 5

Table 5-6 One way ANOVA and comparison of means Chemicals One-Waya Description Comparison of meansb ANOVA Cadmium chloride p< 0.05 S.D LDH § NRU > ATP > MTS

Caffeine p < 0.05 S.D MTS § LDH > NRU > ATP

Cobalt chloride p < 0.05 S.D MTS > LDH > NRU > ATP

Colchicine p < 0.05 S.D MTS § NRU > LDH > ATP

Copper sulfate p < 0.05 S.D MTS > LDH § NRU > ATP

Ethanol p < 0.05 S.D ATP § LDH > NRU > MTS

Glycerol p > 0.05 N.S.D NRU § MTS § LDH § ATP

Irgasan p > 0.05 N.S.D NRU § MTS § LDH § ATP

Lithium sulfate p < 0.05 S.D NRU > ATP § LDH > MTS

Mercuric chloride p > 0.05 N.S.D NRU § MTS § LDH § ATP

Methanol p < 0.05 S.D LDH > NRU > ATP > MTS

Phenol p < 0.05 S.D NRU § ATP > LDH > MTS

Potassium cyanide p > 0.05 N.S.D NRU § MTS § LDH § ATP

Potassium chloride p < 0.05 S.D NRU > LDH > MTS > ATP

Saccharin p < 0.05 S.D NRU > LDH > ATP § MTS

Sodium chloride p < 0.05 S.D NRU > LDH § ATP > MTS

Sodium dichromate p < 0.05 S.D LDH > MTS § NRU > ATP

SLS p < 0.05 S.D LDH > MTS > NRU § ATP

Sodium fluoride p < 0.05 S.D NRU § ATP > LDH > MTS

Sodium selenate p < 0.05 S.D LDH > NRU > ATP § MTS

Strychnine p < 0.05 S.D LDH > NRU > MTS > ATP a One- way analysis of variance b Comparison of means based on individual IC50 values and ANOVA S.D represents statistically significant difference N.S.D represents not statistically significant difference

146 Chapter 5

Correlation between IC50 values between the assays was also performed to establish an in vitro – in vitro correlation. In general, a good correlation was observed, between the assays (Table 5-7). Highest correlation was observed between NRU and ATP assays (r = 0.91, p<0.01), followed by NRU and LDH (r = 0.88, p<0.01), and the weakest but still significant relationship was between NRU and MTS (r = 0.73, p<0.01). The coefficient of determination for the latter was weaker at (R2 = 0.58). Two chemicals affected the weak correlation, and were considered outliers. These chemicals were cadmium chloride, and cobalt chloride. The IC50 of CoCl2 with the MTS and NRU assay was:

14.83 mg/L; 1177.25 mg/L respectively, and the IC50 of CdCl2 for MTS and NRU were: 1501.83 mg/L; 5.63 mg/L. When both chemicals were removed, both the coefficient of correlation and coefficient of determination improved for both assays (r = 0.80, R2 = 0.89) (Figure 5-11). For the remaining correlation coefficients, outlier chemicals could not be easily visually detected. Compared to the other 3 assays, the NRU and LDH assays were very sensitive to cadmium chloride toxicity while MTS assay was sensitive to cobalt chloride toxicity. The IC50 for cadmium chloride was at least 300 fold more toxic compared with both the MTS and ATP assays for the chemical. Cobalt chloride was 20-fold more toxic than depicted with the LDH assays and 80-fold more toxic than with the NRU assay and 400 folds more toxic than results with the ATP assay. Cadmium chloride belongs to the group of heavy metals. Cadmium chloride seems to have an effect on membrane integrity as both the LDH and NRU assays are a measure of membrane integrity. Any chemical having a localized effect upon lysosomes will result in an artificial reflection of low cell number and viability; therefore it will have a greater effect on neutral red uptake than most chemicals. In the literature, cadmium chloride toxicity was measured using the NRU assay on renal epithelial cells and found to be highly potent toxicant (Barrouillet et al. 2001). Cadmium chloride toxicity has also been measured using the LDH assays on LLC-PK1 cells originating from pig proximal tubule, and found that at 50 ȝmol/ L a statistically significant decrease of cell viability (Gennari et al. 2003). Both studies found that cadmium chloride damages the intactness of the epithelial barrier in vitro.

Cobalt chloride appeared to either have a direct toxicity on the mitochondrial activity of the cell, or to interact with the tetrazolium salt reducing formazan production and therefore showing a higher toxicity with the MTS assay as compared to the other assays.

147 Chapter 5

Table 5-7 Correlation matrix between mean IC50 values of chemicals (21) using the MTS, NRU, LDH and ATP assays.

NRU MTS LDH ATP

NRU 1 0.725 0.881 0.912

MTS 0.725 1 0.776 0.796

LDH 0.881 0.776 1 0.830

ATP 0.912 0.796 0.830 1

The data points represent Pearson’s correlation coefficient (r) values for mean IC50 values (three separate experiments) of the respective assays against each other for the 21 chemicals tested. They represent significant values at the level of significance (0.05) (2-tailed test).

y = 0.7057x + 0.6685 6 y = 0.6241x + 0.8855

NRU vs MTS LDH vs MTS Linear (LDH vs MTS) Linear (NRU vs MTS) Log NRU and LDH IC50 (mg/L) 0 036 Log MTS IC50 (mg/L)

Figure 5-11 Regression plot using NRU, LDH IC50 values against MTS IC50 values for the 21 chemicals.

Values represent mean IC50 values from three independent experiments of 21 tested chemicals of NRU and LDH values against MTS values. Correlation coefficient for NRU vs. MTS (R2 = 0.80) and LDH vs. MTS (R2 = 0.76).

148 Chapter 5

y = 0.9748x - 0.0379 6 y = 0.925x - 0.1668 (mg/L) 50

MTS vs. ATP NRU vs. ATP

Log MTS, NRU IC NRU MTS, Log Linear (MTS vs. ATP) Linear (NRU vs. ATP) 0 036 Log ATP IC50 (mg/L)

Figure 5-12 Regression plot using MTS, NRU IC50 values against ATP IC50 values for the 21 chemicals.

Values represent mean IC50 values from three independent experiments of 21 tested chemicals of MTS and NRU values against ATP values. Correlation coefficient for MTS vs. ATP (R2 = 0.63) and NRU vs. ATP (R2 = 0.85).

y = 0.7372x + 1.2232 6 y = 0.8014x + 0.6279

NRU vs. LDH ATP vs. LDH Linear (ATP vs. LDH)

Log NRU, ATP IC50 ATP Log NRU, (mg/L) Linear (NRU vs. LDH)

0 02.55 Log LDH IC50 (mg/L)

Figure 5-13 Regression plot using NRU, ATP IC50 values against LDH IC50 values for the 21 chemicals.

Values represent mean IC50 values from three independent experiments of 21 tested chemicals of NRU and ATP values against LDH values. Correlation coefficient for NRU vs. LDH (R2 = 0.80) and ATP vs. LDH (R2 = 0.70).

149 Chapter 5

The NRU assay was found to be a more sensitive measure of toxicity for most chemicals than the MTS, LDH, and ATP assays (Figure 5-14). From the chemicals tested 40% had lower IC50 values with the NRU assay than the remaining assays (Figure 5-14). It was noted that NRU seemed to be a better indicator of toxicity for chemicals that generally fall in category 4 (LD50 >300 mg/kg), following GHS classification for acute toxicity. Examples were: phenol, lithium sulfate, potassium chloride, sodium chloride. A general ranking of assay sensitivity to chemical toxicity based on Figure 5- 14 was as follows NRU > LDH > ATP > MTS. It has been previously shown that the ATP based assay is more sensitive than the MTT assay (of which MTS is a modification) (Petty et al. 1995). MTT, NRU and LDH assays have been reported to correlate well together, when tested on surfactants (Arechabala et al. 1999). A reason for the lower sensitivity of the MTT assay was because the assay measures the cell’s ability to convert tetrazolium salt to formazan product. As the cell loses viability (metabolic activity) and decreases mitochondrial function its ability to perform a conversion decreases. However some chemicals increase the metabolic activity in the cell, resulting in increased mitochondrial succinate dehydrogenase activity. This enzyme is susceptible to local changes in ion concentration and ion flux. Therefore its activity can be increased with the appropriate ionic conditions (Putnam et al. 2002). The MTS assay assumes that the ability of a cell to reduce the tetrazolium salt will remain constant throughout the cell culture period. The mitochondria being stronger than the cells that contain them, can continue to reduce MTS, while other parts of the cells are injured, leading to a right-ward shift of the concentration-response curves (Petty et al. 1995; Mueller et al. 2004). For chemicals that fell in category 4 and above (GHS classification): IC50 (MTS) > IC50 (NRU). For example potassium chloride has a much lower IC50 value using the NRU assay (IC50 = 1925 mg/L) than using the MTS assay

(IC50 = 17 000 mg/ L).

150 Chapter 5

NRU LDH ATP MTS

rol

o gly

methan

ethanol in

ine

sac fe

loride

dium c dium

ulfa

e s

hnin

potassium chloride potassium

lithium

icine

Str h e

Colc lorid

e h

nat

t d t

l l

no Chemicals Test

m s m

Phe

a diu

s los

e c

nid Tri

ide

c r

e ium

romat

pota

dium fluo dium h

e dium dic dium

d o

copper sulfate copper

ori

chl

1 di m

u i

1000 a c

mercuric chloride mercuric 1000000

(mg/L) (Log scale) (Log (mg/L) IC 50

Figure 5-14 bar plot using mean IC50 values of tested chemicals with MTS, NRU, LDH and ATP assays. Error bars represent m ± SD for 3 separate experiments

151 Chapter 5

5.3.2.1 Primary PCA modelling

The individual IC50 results (mean of 3 separate experiments) from the 4 assays, on the selected chemicals (21), were analysed using principal component analysis (PCA) (XLSTAT, US). PCA was used to show multiple intercorrelations between the assays and thus the equality of the results. PCA and factorial analysis are the most widely used statistical methods to reduce the number of dimensions in data analysis and to investigate multiple intercorrelations between variables (Massart et al. 2003; Mueller et al. 2004). 96% of the total variance of all variables was explained by 1 principal component. Factor loadings were 0.919, with MTS the weakest, followed by LDH (0.946), ATP (0.949) and NRU (0.954). The results indicated the similarity of the 4 assays. Results of the intercorrelation between the data are found in Table 5-7 (correlation matrix). The correlation matrix gave a 1st overview on strong and weak correlation between the 4 assays. Figure 5-15 and Figure 5-16 are graphical illustrations of the variables along two axes F1 and F2. The MTS assay lies the furthest away on the axis in comparison with the other assays (Figure 5-15).

Variables (axes F1 and F2: 95 %)

1.5

0.5 MTS

--> MTS 0 LDH ATP LDH NRU ATP -0.5 Axis F2 (8%) NRU

-1

-1.5 -1.5 -1 -0.5 0 0.5 1 1.5 -- Axis F1 (87 %) -->

Figure 5-15 Graphical illustration of the positions of the assays using PCA modelling

152 Chapter 5

Variables and observations (axes F1 and F2: 95 %)

1.5

Cadmium chloride 1 MTS

LiSO4 NaSe NaCl LDH 0.5 EthanolMTS NaF Strychnine ATP Mercuric chloride Phenol KCl Methanol Saccharin 0 LDH NRU SLS KCN ATP Irgasan Glycerol Colchicine NRU

-- Axis F1 (8%) CrVI Caffeine -0.5 CuSO4

-1 Cobalt Chloride

-1.5 -1.5 -1 -0.5 0 0.5 1 1.5 -- Axis F1 (87 %) -->

Figure 5-16 Graphical illustration of the observations and variables.

5.3.3 Reproducibility of the in vitro assays

Reproducibility of results was given as an average coefficient or variation (CoV) for three separate experiments for each assay and calculated using Equation 5-1.

standard deviation CoV [5-1] mean (%)

The reproducibility of the ATP assay (CoV = 16.85%) was found to be greater than the remaining assays (Table 5-8). In general ranking of the reproducibility of the assays was: ATP > LDH > NRU > MTS. The reproducibility of both LDH and NRU assays were similar (CoV = 18.02; CoV = 19.61, respectively). This result was comparable to a study where both MTT and ATP were compared and the ATP assay was found to be

153 Chapter 5

much more sensitive (required less cell number for detection) and had a better reproducibility than the MTT assay (Petty et al. 1995).

Table 5-8 Calculated Coefficient of variation (CoV) for the assays

MTS NRU LDH ATP

CoV (%) 26.65 19.61 18.02 16.85

The data represents the calculated CoV (standard deviation/mean %), for the 21 chemicals tested in 3 separate experiments.

5.4 Comparison of In vitro – In vivo Toxicity

The performance of the MTS, NRU, LDH and ATP cytotoxicity assays was compared to published in vivo toxicological test methods. Different comparisons were undertaken:

Method one: a comparison of the mean IC50 data, from the cytotoxicity assays, with rodent oral LD50 data. Selected rodent LD50 data was taken from the GHS list of selected and alternate chemicals that they used to derive the classification of acute

toxicity of chemicals (http://iccvam.niehs.nih.gov/ivcytoval/alt_chem.pdf278H4 ). A few chemicals could not be found with the GHS selection of chemicals, and rodent LD50 values, were taken as an average of both oral LD50 rat and mice from the Registry of Cytotoxicity Data Bank, ZEBET (cobalt chloride, sodium dodecyl sulfate and saccharin) and irgasan from RTECS (NIOSH 2002).

Method 2: average IC50 data were compared with in vivo human toxicity data: LDL0

(Lethal Dose Lowest), this represented the lowest dose (other than LD50) of a chemical introduced by any route, over any given period of time, reported to have caused deaths in humans. In the case where the LDL0 was not available, TDL0 (Toxic Dose Lowest), was used. The sources of published in vivo human toxicity data were derived from RTECS (NIOSH 2002), Sax’s Dangerous Properties of Industrial Materials (Lewis 2000), Hazardous Substances Databank (HSDB 2002), and National Toxicology

154 Chapter 5

Program (NTP and NICEATM 2003). Mean IC50 values were also compared to published lethal blood concentrations in humans (LC). LC values were taken from (Barile et al. 1994b; NTP and NICEATM 2000) .

Method 3: average NOEC values determined for the assays were correlated with threshold limit values – time weighted average (TLV-TWA) (ACGIH, 2000), or if data was missing then Permissible Exposure Limits were used (PEL) (OSHA, 2000). The American Conference of Governmental Industrial Hygienists (ACGIH) is a trade organisation that annually publishes exposure limits in the form of TLVs for chemicals and for physical agents. They take the form of TLV. These exposure limits are adopted as guidelines and are not enforceable. According to ACGIH, TLVs refer to “airborne concentrations of substances and represent conditions under which it is believed that nearly all workers may be repeatedly exposed day after day without adverse health effects” (ACGIH, 2000). TLV-TWA is an occupational exposure limit for exposures averaged over an 8-hour day, 5-day week work regimen. The Occupational Safety and Health Administration (OSHA), publishes legally enforceable PELs. They are designed to apply the best scientific evidence to ensure as reasonably as possible that no worker will suffer health or functional impairment with regular exposure for the duration of his/her working life.

5.4.1 Correlation of in vitro – in vivo rodent data

Comparison of in vitro – in vivo data was of particular interest since in vivo studies are still common in risk assessment procedures for predicting human toxicity (Chapter 2: Section 2.2.1). Results from linear regression analysis of the 21 chemicals initially, only gave weak correlations between the assays and rodent LD50. The assays gave weak positive statistically significant result: NRU (r = 0.56, p < 0.05) > LDH (r = 0.55, p < 0.05) > ATP (r = 0.52, p< 0.05) > MTS (r = 0.50, P<0.05). R2 values were very weak 2 (R < 0.40). The lethality indices (IC50/LD50) were then taken for the compounds, and subsequently adjustment factors were applied to the chemicals whose lethality indices were significantly different from the other chemicals (Table 5-9). In a study on 44 diverse compounds, Halle and Spielmann, 1992 eliminated compounds based on

155 Chapter 5

lethality indices and subsequently the correlation between IC50 values and LD50 values increased from r = 0.55 to r = 0.73. Lestari et al., 2005 also followed a similar approach on the study of selected substances commonly produced during fires, in which IC50 values were correlated against published human toxicity data. This study showed a significantly positive increase in correlation after the application of an appropriate adjustment factor, calculated from the ratio of their IC50 values with the human toxicity data.

In order to improve development and incorporation of in vitro data in a consistent fashion in risk assessments for regulatory purposes, it has been argued that it is acceptable to use adjustment factors, to quantitate interspecies differences, when extrapolating from in vitro data (e.g. cell culture systems) to whole animals or humans (Worth and Balls 2002; Meek et al., 2002; Combes et al., 2003). In addition adjustment factors are used to account for human variability in kinetics and dynamics (Meek et al. 2002; Combes et al. 2003). In risk assessments, chemical specific adjustment factors (CSAFs) are developed for the consideration of kinetic and dynamic data, as a response for interspecies differences and human response variability in dose response assessment. A factor is applied to allow for the incorporation of quantitative chemical-specific data, relating to either toxicokinetics or toxicodynamics(Meek et al. 2002). A factor of 100 is used to account for interspecies difference for toxicodynamics as the ratio of the effective concentration in animal versus human tissues (EC10 animal / EC10 human). Based on this consideration, an appropriate adjustment factor was applied which was calculated from the ration of IC50/LD50. A list of the modifying factors applied is found in Table 5-9. The correlation improved greatly, as can also be seen from Figure 5-17. A list of the correlation coefficients is found in Table 5-10. The most positive significant correlation based on Pearson’s r was between NRU assay against rodent LD50 (r = 0.94, R2 = 0.83, p < 0.01), followed by the ATP assay (r = 0.92, R2 = 0.75, p < 0.01), then the MTS assay (r = 0.91, R2 = 0.79, p < 0.01), while the weakest but still positively significant was with the LDH assay (r = 0.89, R2 = 0.82, p < 0.01).

156 Chapter 5

Table 5-9 Ratio of IC50/LD50 and adjustment factors implemented for correlations between in vitro – in vivo data

MTS NRU LDH ATP Chemicals Ratio Ratio Ratio Ratio AF AF AF AF IC50/LD50 IC50/LD50 IC50/LD50 IC50/LD50 Caffeine 7.02 N/A 16.0 ÷10 9.63 N/A 72.2 ÷10 Cobalt 0.18 N/A 14.6 ÷10 3.62 N/A 86.1 ÷10 chloride Colchicine 216 ÷ 100 241 ÷100 449 ÷100 1440 ÷103 Cupric sulfate 0.09 N/A 0.68 N/A 0.49 N/A 1.27 N/A Glycerol 7.59 N/A 10.92 N/A 7.88 N/A 7.80 N/A Irgasan 7.0 ÷ 10 0.09 N/A 0.19 N/A 0.13 N/A Lithium 21.1 ÷ 10 1.62 N/A 6.89 N/A 7.09 N/A sulfate Mercuric 7.04 N/A 3.08 N/A 3.62 N/A 3.93 N/A chloride Phenol 4.19 N/A 1.03 N/A 2.39 N/A 1.20 N/A Potassium 6.66 N/A 0.74 N/A 2.89 N/A 8.88 N/A chloride Saccharin 1.47 N/A 0.41 N/A 0.48 N/A 1.02 N/A Sodium 5.78 N/A 0.77 N/A 2.31 N/A 2.44 N/A chloride Sodium 0.58 N/A 4.28 N/A 0.11 N/A 10.7 N/A dichromate SLS 0.02 × 10 0.03 ×10 0.01 ×10 0.03 ×10 Sodium 5.81 N/A 1.57 N/A 4.14 N/A 2.08 N/A fluoride Strychnine 3750 ÷ 1000 78.0 ÷10 38.6 ÷10 4.62 N/A Cadmium 17.1 ÷ 10 0.06 ×10 0.03 ×10 3.75 N/A chloride Ethanol 2.20 N/A 0.84 N/A 0.55 N/A 0.49 N/A Methanol 6.51 N/A 1.06 N/A 0.76 N/A 3.99 N/A Potassium 56.7 ÷ 10 35.7 ÷10 67.21 ÷10 61.31 ÷10 cyanide Sodium 7940 ÷ 1000 1440 ÷103 212.26 ÷100 4853 ÷103 selenate

AF adjustment factor N/A Not Applicable

157 Chapter 5

MTS vs LD50

(mg/L) NRU vs LD50

50 3 LDH vs LD50 ATP vs LD50 Log IC

0 02.55

Log LD50 (mg/kg)

Figure 5-17 Regression between cytotoxicity (IC50) (MTS, NRU, LDH and ATP) and acute oral

rodent LD50 values. Correlations were made for the 21 chemicals, tested using the different assays (MTS, NRU, LDH and ATP), with the application of adjustment factors. Coefficient of determination R2 were determined for 2 2 2 2 LD50 data against MTS, NRU, LDH and ATP (R = 0.79; R = 0.83; R = 0.82 and R = 0.75), respectively.

Table 5-10 Correlation matrix in vitro – in vivo with application of adjustment factors (AF).

MTS NRU LDH ATP LD50 MTS 1 0.907 0.910 0.905 0.909 NRU 0.907 1 0.921 0.958 0.940 LDH 0.910 0.921 1 0.911 0.893 ATP 0.905 0.958 0.911 1 0.926 LD50 0.909 0.940 0.893 0.926 1 r values are significant at the 0.05 level (2-tailed).

Adjustment factors were mostly applied to the following chemicals across the assays: colchicine, strychnine, sodium dodecyl sulfate (SDS), potassium cyanide (KCN) and sodium selenate. Possible reasons for this disparity between in vitro toxicity of the chemicals and the rodent LD50 data are outlined below. In general, most chemicals having neurotoxic effects have been found to be outliers when tested for the disruption of basal cytotoxicity functions. Most basal cytotoxicity assays underestimate toxicity for chemicals known to act on specific receptors, or cells, or following bioactivation

158 Chapter 5

(Clemedson et al. 2002). Some chemicals needed different adjustment factors depending on the sensitivity of the assay to the chemical’s toxicity and on the calculated

IC50/LD50 ratio. In the case of strychnine the ATP assay was the most sensitive and no adjustment factors were needed. On the other hand, the MTS assay was the least sensitive and adjustment factor of 1000 was used. NRU and LDH needed a similar adjustment factor of 10.

SDS: showed a higher toxicity in vitro, compared to in vivo, could be because it is an anionic surfactant, allowing easier spreading and penetration through the cellular system.

KCN: Studies conducted on KCN, found that it could not be studied in cell lines (Clemedson et al. 2002). In humans, acute cyanide exposure results primarily on the central nervous system (CNS), has cardiovascular and respiratory effects. While in vitro KCN inhibits cytochrome oxidase, and thereby, cell respiration. At the same time, cell lines can alternate between aerobic and anaerobic metabolism, therefore making it hard to be able to study the effect of KCN.

Strychnine: Causes excitation of all parts of the CNS, through a specific function of increasing the level of neuronal excitability by interfering with inhibitory influences (e.g. glycine) on motor neurons. It has a specific site of action which is the postsynaptic membrane. It interferes with the postsynaptic inhibition that is mediated by glycine. Glycine is an inhibitory transmitter to motor neurons in the spinal cord. And strychnine acts as a selective competitive antagonist to block the effect of glycine at the glycine receptors (INCHEM 2005).

Colchicine: Even though it exerts a multiorgan toxicity, its main toxic effects are exerted on cellular division. This mitosis blockade, will lead in vivo to diarrhoea, bone marrow depression, and alopecia. Colchicine may also have toxic effect on muscle, peripheral nervous system and liver. It would be suggested that a further study might need to be conducted on colchicine effect on HepG2 cell line, instead of the skin fibroblasts currently used in the study, to check if direct cytotoxicity will then be detected (NIOSH 2002).

159 Chapter 5

Sodium selenate: the mechanism of selenium toxicity remains unclear, and no unifying hypothesis regarding the toxic effects of selenium has been presented, it is possible that selenite may interfere with glutathione metabolism and that this may affect enzyme activities. Animal studies suggest that the cytotoxicity of selenium results from the pro- oxidant catalytic activity of the selenide anions, which produce superoxide anions, hydrogen peroxide and other reactive metabolites.

In recent years, several large research groups such as CAAT, MEIC, FRAME (Chapter 2: Section 2.2.3), have evaluated various in vitro assays, and have recommended the use of cytotoxicity assays in toxicology. For e.g. the MEIC, examined a large number of cytotoxicity endpoints using 50 diverse chemicals with different mechanisms of actions. The studies showed a good correlation with in vitro cytotoxicity and acute lethal potency and irritant potential (Chapter 2: Section 2.2.4). But they also found that specific toxic mechanisms exist, which can only be measured using organotypic in vitro toxicity tests (Clemedson et al., 2002). Additional mechanistic information, toxicokinetics parameters, needed to be incorporated into the study for the proper detection of chemical toxicity. The Integrated Toxicity Testing Scheme (ICITTS) has been recommended as a step-wise procedure, where chemical toxicity is predicted with integrating knowledge from in vitro tests, and also physiologically based biokinetic models (Blaauboer et al., 2001). Further studies; have shown the need to also integrate blood brain barrier passage (BBB) and neurotoxic effects (Prieto et al. 2004). An extensive in vitro study with culture human lung and dermal cells was conducted to test the basal cytotoxicity concept (Barile and Cardona 1998). 29 chemicals from the MEIC list of chemicals were tested with each cell line, using the MTT assay. The study compared the in vitro data to LD50 values already established for these chemicals in order to examine the ability of their methods to screen for human toxicity. Barile et al., 1998, concluded that relatively rapid in vitro cytotoxicity tests such as MTT have the potential for screening cytotoxic compounds as alternative methods to the traditional animal testing protocols. Some chemicals, though, such as digoxin eluded the cytotoxic screen. Several other studies have also been undertaken using MEIC chemicals to compare their results with animal and human data (Shrivastava et al. 1992; Dierickx 2000; Scheers et al. 2001). Garle et al., 1994, summarised the important published

160 Chapter 5

cytotoxicity tests and concluded that significant correlations exist between cytotoxicity and animal lethality on numerous occasions.

Sauvant et al., 1997, compared the relative toxicity of 16 environmental pollutants measured using 4 bioassays (RNA, MTT, NRI and DTP) on ciliated protozoa. The in vitro results were compared with the rat model. The study found significant but low correlations for all the assays (0.50 < r < 0.69). The researchers explained the relatively low correlations determined by the study was caused by many possible factors such as: the low number of substances tested; the lack of power of the statistical tests performed and the protozoan cell lines used were less able to predict toxicity than mammalian or eukaryotic cell lines (Sauvant et al. 1997).

Only one study, currently underway, has used the GHS, as a basis for selecting their chemicals for validation of cytotoxicity testing. A validation study is being conducted by the NTP NICEATM, to evaluate the usefulness of two cytotoxicity methods: the NRU assay using mouse fibroblasts (BALB/c) 3T3 cells and the NRU assay using normal human keratinocytes (NHK). One of their study objectives is to address the accuracy of the assays for estimating rodent oral LD50 values and human lethal concentrations across the five GHS (GHS; OECD 2001) categories of acute oral toxicity as well as unclassified toxicity. All 72 chemicals of the GHS were selected, with 12 chemicals from each of the 6 acute oral toxicity classification groups.

5.4.2 Correlation of in vitro – in vivo human toxicity (LDL0)

Ultimately, validated in vitro tests need to replace or compliment animal testing for the prediction of human acute toxicity. Hence, the development and validation of in vitro methods should concentrate on predicting human acute toxicity, rather than rodent, acute toxicity. In this study a correlation of IC50 values against published human toxicity data was established to determine the predictability of the in vitro results compared to in vivo human toxicity. Mean IC50 values of the tested chemicals were compared with

161 Chapter 5

published LDL0 and where this value was not available it was replaced with TDL0.

Table 5-11 summarises the published in vivo LDL0 and the different sources.

Initial correlation between mean IC50 values of the assays and human lethal dose, was quite weak, but it is important to note that when rodent LD50 data was correlated against

LDL0 values, it did not give a better correlation, but was actually weaker (Table 5-13). NRU and ATP were the only assays to give a significant correlation for (N = 21), with (r = 0.51; r = 0.47; p < 0.05), but the correlation of determination was still very weak 2 2 (R = 0.30 and R = 0.25). Specificity of the variables with LDL0 was as such NRU >

ATP > LDH > MTS > LD50 (Table 5-13). The ratio between IC50/LDL0 for the various assays was established and an adjustment factor was applied as per Section 5.4.1. The degree of correlation based on the r value and regression coefficient (R2) was determined. There was a significant increase in correlation with LDH having the highest correlation (r = 0.97, p < 0.05), with the weakest but still significant correlation with the

MTS assay (r = 0.85, p < 0.05) (Table 5-14). Table 5-12 presents the ratio IC50/LDL0 and the appropriate adjustment factors implemented for correlation based on the lethality indice ratio. Table 5-13 and Table 5-14 list the correlation variables with and without adjustment factors. The correlation of IC50 values against in vivo data with the inclusion of the appropriate adjustment factors are shown in Figure 5-18.

In general, the ratio between in vitro data to LDL0 varied within the same assay and between assays. A good comparison between IC50/LDL0 was observed for the LDH assay, with most chemicals having ratios between 0.5–7, an adjustment factor was applied for 6 chemicals (colchicine, lithium sulfate, potassium chloride, SDS, cadmium chloride and potassium cyanide). Reasons for the weak ratio for colchicine, and potassium cyanide can be explained as per Section 5.4.1. Lithium sulfate and potassium chloride seem to cause toxicity in humans at much lower doses than detected with basal cytotoxicity. The TDL0 for SDS was derived after direct application on human skin (24 hour exposure), a more direct testing for the toxicity of SDS than a basal cytotoxicity test. Overall, a good agreement was also observed for IC50/LDL0 for the NRU and ATP assay, followed by the MTS assays (Table 5-12).

162 Chapter 5

Table 5-11 Published in vivo human Lowest Lethal Dose (LDL0) Chemicals Toxicity Dose (mg/kg) Exposure Route Reference data

Cadmium chloride LDL0 3000 Oral woman RTECS

Caffeine LDL0 192 oral human Sax’s

Cobalt chloride LDL0 1500 Oral child RTECS

Colchicine LDL0 11 Oral man RTECS

Cupric sulfate LDL0 221 Unknown man Sax’s ; RTECS

Ethanol LDL0 1400 Oral human Sax’s ; RTECS

Glycerol TDL0 1428 Oral human Sax’s ; RTECS

Irgasan LDL0 750 Skin exposure (3d) HSDB

Lithium sulfate TDL0 6 Oral woman Sax’s ; RTECS

Mercury chloride LDL0 14.3 Oral man RTECS

Methanol LDL0 6422 Oral man Sax’s ; RTECS

Phenol LDL0 140 Oral human Sax’s ; RTECS

Potassium cyanide LDL0 2.9 Oral human Sax’s ; RTECS

Potassium chloride LDL0 20 Oral man Sax’s ; RTECS

Saccharin TDL0 1000 Oral man HSDB

Sodium chloride TDL0 12357 Oral human (23d) Sax’s ; RTECS

Sodium dichromate (as Cr) LDL0 50 Oral child RTECS

SLS TDL0 0.14 Human skin (24h) RTECS

Sodium fluoride (as F) LDL0 90 Oral woman Sax’s ; RTECS

Sodium selenate TDL0 53 Oral man Sax’s ; RTECS

Strychnine LDL0 30 Oral human Sax’s ; RTECS 5143 Oral man Sax’s ; RTECS

LDL0 represents Lethal Dose Lowest (mg/kg) TDL0 represents Toxic Dose Lowest (mg/kg) RTECS Registry of Toxic Effective Concentrations Sax Sax’s Dangerous Properties of Industrial Materials HSDB Hazardous Substances Databank

163 Chapter 5

Table 5-12 Ratio of in vitro to in vivo human toxicity and adjustment factors implemented for correlation

MTS NRU LDH ATP Chemicals Ratio Ratio Ratio Ratio AF AF AF AF IC50/LDL0 IC50/LDL0 IC50/LDL0 IC50/LDL0 Caffeine 7.02 N/A 16.0 ÷ 10 9.63 N/A 72.20 ÷ 10 Cobalt 0.01 ×10 0.78 N/A 0.19 N/A 4.62 N/A chloride Colchicine 112 ÷ 100 131.2 ÷ 100 245.0 ÷100 787 ÷ 100 Cupric sulfate 0.13 N/A 0.92 N/A 0.65 N/A 1.72 N/A Glycerol 67.41 ÷ 10 97.07 ÷ 10 7.00 N/A 68.34 ÷ 10 Irgasan 0.38 N/A 0.49 N/A 1.03 N/A 0.69 N/A Lithium 4171 ÷ 1000 320.8 ÷ 100 1363 ÷103 1650 ÷ 103 sulfate Mercuric 0.49 N/A 0.21 N/A 0.25 N/A 0.27 N/A chloride Phenol 12.37 ÷ 10 3.03 N/A 7.07 N/A 3.53 N/A Potassium 866.4 ÷ 100 96.27 ÷ 10 375.4 ÷ 100 1155 ÷103 chloride Saccharin 25.03 ÷ 10 6.93 N/A 8.18 N/A 17.33 ÷ 10 Sodium 1.40 N/A 0.18 N/A 0.56 N/A 0.59 N/A chloride Sodium 3.2 N/A 4.28 N/A 0.11 N/A 10.72 ÷ 10 dichromate SLS 198.6 ÷ 100 299.2 ÷ 100 93.01 ÷ 10 330.3 ÷ 100 Sodium 11.61 ÷ 10 3.13 N/A 8.27 N/A 4.16 N/A fluoride Strychnine 1.46 N/A 0.30 N/A 0.15 N/A 1.79 N/A Cadmium 0.50 N/A <0.01 ×100 0.01 ×100 0.11 N/A chloride Ethanol 22.00 ÷ 10 8.40 N/A 5.54 N/A 4.95 N/A Methanol 13.19 ÷ 10 2.15 N/A 1.54 N/A 8.09 N/A Potassium 122.4 ÷ 100 123.1 ÷ 100 231.7 ÷ 100 211.3 ÷ 100 cyanide Sodium 239.7 ÷ 100 13.08 ÷ 10 7.54 N/A 22.26 ÷ 10 selenate

AF adjustment factor N/A Not Applicable

164 Chapter 5

Table 5-13 Correlation Table between cytotoxicity data (IC50), acute oral rodent LD50 values and

human lethal dose values (LDL0)

Human Lethal Dose MTS NRU LDH ATP LD50

LDL0 r 0.41 0.51 0.421 0.473 0.297 p (2-tailed) 0.60 0.04 0.10 0.032 0.190 R2 0.21 0.30 0.20 0.25 0.096

Table 5-14 Correlation Table between in vitro – in vivo human toxicity with adjustment factor

Human Lethal Dose MTS NRU LDH ATP

LDL0 r 0.85 0.90 0.97 0.93 p (2-tailed) < 0.01 < 0.01 < 0.01 < 0.01 R2 0.73 0.82 0.76 0.86

3 NRU vs LDL0 LDH vs LDL0 2

(mg/L) ATP vs LDL0 50 1 MTS vs LDL0

Log IC 0

-1

-2 -2-1012345

Log LDL0 (mg/kg)

Figure 5-18 Correlation of IC50 (MTS, NRU, LDH and ATP) assays against LDL0 values Correlations were made for the 21 chemicals, tested using the different assays (MTS, NRU, LDH and ATP), with the application of adjustment factors. Coefficient of determination R2 were determined for 2 2 2 2 LDL0 data against MTS, NRU, LDH and ATP (R = 0.73; R = 0.82; R = 0.76 and R = 0.86, respectively). The line represents a 1:1 linear correlation.

165 Chapter 5

One possible reason for the weak correlation observed between cytotoxicity in vitro and published in vivo human toxicity data (LDL0) was due to a wide variation of published human toxicity data from numerous sources. The results from this study are comparable to a study conducted by Weiss et al., 1993, in which they measured the toxicity of 50 MEIC test chemicals in primary cultures of chick embryo forebrain neurons using the MTT and NRU assays. The study found that the predictive value of the in vitro data for oral human toxicity was generally poor (Weiss and Sawyer 1993). A number of studies also found poor correlations between in vitro data and human published data (Romert et al. 1994; Ponsoda et al. 1995).

Data derived from in vitro toxicity studies lacks knowledge of the processes of absorption, distribution, metabolism and excretion (i.e. biokinetics) behaviour in vivo (Blaauboer 2002a). There is a need to integrate in vitro biokinetic modelling, with data on mechanism of action of chemicals (Blaauboer 2002b), in addition to the inclusion of blood brain barrier (BBB) data, and target organ toxicity (Prieto et al. 2004).

There is a need to take into consideration the kinetic and dynamic data of the chemical, inter-species and human variability as a basis for using uncertainty factors in dose response analysis (Meek et al., 2002).It is also possible to include in vitro toxicodynamics measurements (or the assessment of the sensitivity of the target tissue to the presence of chemical) as part of the uncertainty factor used in risk assessments (Paine 1996). Certain studies have shown that by including biokinetic data, the ability of in vitro data to predict acute human toxicity, significantly increased. An example is the inclusion of BBB data, distribution volume, protein binding, intestinal absorption to the MEIC chemicals, increased their ability to predict human lethal concentration (Ekwall et al. 1989; Clemedson and Ekwall 1999; Ekwall 1999a; Clemedson et al. 2003).

Lestari et al., 2005 compared IC50 values using three different cell lines, on selected substances and using MTS assay, against LDL0 and LCL0 (lethal concentration lowest). The study found very weak correlations when results were directly compared without including adjustment factors.

In addition, previous independent findings have found that the IC50 values are more accurate predictors of human toxicity than data derived from rodent LD50 data (Barile et

166 Chapter 5

al. 1994b). Furthermore a comparative study of the prediction of human toxicity by rodent LD50 values and by results from in vitro toxicity assays and three independent sets of human acute lethal blood concentrations were better than between rodent LD50 and human acute lethal concentrations (Ekwall et al. 2000).

While this information indicates the predictive potential of the assays currently used in this study for acute human toxicity, this study also proves that they can be compared favourably to established in vitro cytotoxicity assays in terms of in vitro – in vivo correlations.

5.4.3 Correlation of in vitro – in vivo human plasma (LC)

Comparative analysis was performed between in vitro IC50 and rodent LD50 values’ ability to predict human lethal plasma concentration (LC). To facilitate comparison, rodent LD50 values were first converted to human equivalent toxic blood concentrations (HETC) according to Equation 5-2 (Barile et al. 1994b; Barile and Cardona 1998).

§LD50 · 3 HETC ¨ ¸ u10 [5-2] ©Vd ¹

HETC represents estimated human equivalent toxic concentration in plasma (mg/ml).

LD50 represents 50% lethal dose in rodents (mg/kg) Vd represents volume of distribution (L/Kg), and 10-3 represents constant for conversion to millilitres (L/ml)

Chemicals in this study were limited to the ones whose volume of distribution (Vd) could be found in the literature (Table 5-15). Table 5-15 lists in vitro IC50 values for the cytotoxicity assays converted to mg/ml; Volume of distribution data taken from (Baselt and Cravey 1989); Human Equivalent Toxic Concentration (HETC) values (mg/ml)

167 Chapter 5

derived using Equation 5-2 and Human Lethal Plasma Concentration (LC) values (mg/ml) taken from (Barile et al. 1994b; NTP and NICEATM 2000).

Table 5-15 Cytotoxicity values (IC50), acute oral LD50 values, Human Equivalent Toxic Concentrations (HETC) and Human Lethal Concentration (LC) values are shown. 1 2 3 Chemical MTS NRULDH ATP LD50 Vd HETC LC

Phenol 1.732 0.42 0.99 13.86 530.0 1.00 0.53 0.03 Sodium chloride 17.32 2.30 6.90 7.30 3750 1.00 3.75 N/A Sodium fluoride 1.04 0.28 0.74 0.37 180 0.60 0.30 0.02 Lithium sulfate 25.03 1.93 8.17 9.90 1187 0.90 1.32 0.01 Copper sulfate 0.03 0.20 0.14 0.38 300 1.00 0.30 0.03 Mercuric chloride 0.01 <0.01 0.01 <0.01 1.00 1.00 <0.01 <0.01 Caffeine 1.34 3.08 1.84 13.86 192.0 0.40 0.48 0.18 Potassium chloride 17.33 1.92 7.50 23.10 2602 1.00 2.60 7.00 Methanol 84.71 13.86 9.90 51.97 13012 0.60 21.69 1.90 Ethanol 30.81 11.77 7.76 6.93 14008 0.53 26.43 3.90 Potassium cyanide 0.56 0.35 0.67 0.61 10.00 1.00 0.01 0.01

Cytotoxicity values (IC50) for the assays are shown in mg/ml 1Volume of distribution, values taken from Baselt and Cravey (1989). 2Human equivalent toxic concentration (mg/ml) derived using Equation 5-3. 3Human lethal plasma concentration, (mg/ml), values taken from (Barile et al. 1994b; NTP and NICEATM 2000).

Table 5-16 shows the correlation coefficients (r) and coefficients of determination (R2) from the regression analysis of IC50 data from the assays and the rodent LD50 present in the form of human equivalent toxic concentration (HETC) against human lethal plasma concentration (LC). When all chemicals tested (n = 11) were plotted against LC values, a positively significant correlation was found, and a reasonable coefficient of determination. The highest correlation was with HETC (r = 0.80, R2 = 0.64, p<0.05) followed close by with NRU (r = 0.78, R2 = 0.65, p < 0.05), then MTS (r = 0.71, R2 = 0.50, p < 0.05), with the least significant relationship for LDH and ATP (r = 0.70, R2 = 0.50, p<0.05) (Table 5-16). Even though the number of chemicals is relatively low in this actually study, but it can be safely deduced that in vitro methods have the ability to predict human toxicity at least as accurately as animal procedures.

168 Chapter 5

Table 5-16 correlation and regression analysis between cytotoxicity data (IC50), human equivalent toxic concentrations (HETC) values and human lethal concentration values (LC)

Human Lethal Concentration MTS NRU LDH ATP HETC LC Pearson’s r 0.71 0.78 0.70 0.70 0.80 p (2-tailed) 0.02 0.009 0.02 0.02 0.005 R2 0.50 0.65 0.50 0.50 0.64

5.4.4 Correlation of in vitro - and TLV

Currently lethality data from animals continue to be used by government, regulatory authorities not only to place chemicals in hazard categories, but also for the setting of permissible uses and exposure limits. This is generally done with the use of NOAELs (no observable adverse effect levels) determined from animal experiments and divided by an uncertainty factor. Therefore any in vitro method that needs to be used for regulatory purposes needs to be compared with current exposure limit values. In this study NOEC values for the four assays, extrapolated from the concentration-effect curves, according to statistical method (Chapter 4: Section 4.2.7.4) were correlated against recommended threshold limit values (TLV). The TLVs are recommended exposure levels given by the American Conference of Governmental Industrial Hygienists (ACGIH, 2000). These values include the time-weighted average (TWA) for a normal 8 hour workday. The number of chemicals (13) was limited to those having TLV values (Table 5-17). A safety factor of 10 was included in the NOEC values, to allow for a margin of safety when extrapolating the NOEC value from the concentration-effect curve. TLV values of many chemicals were in (mg/m3) in air, and were converted to (ppm) according to Equation 5-3 with the exception of values for methanol, ethanol and phenol (ACGIH, 2000).

24.45 TLV(ppm) TLV(mg/m3 )u [5-3] MolecularWeight

169 Chapter 5

Significantly positive correlations were found between the NOEC values of tested chemicals against TLV. LDH (r = 0.82; R2 = 0.70; p < 0.01) > NRU (r = 0.80; R2 = 0.65; p < 0.01) > MTS (r = 0.76; R2 = 0.60; p < 0.01) > ATP (r = 0.71; R2 = 0.55; p < 0.01).

Table 5-17 Cytotoxicity values (NOEC) and threshold limit values (TLV)

Chemicals MTS2 NRU2 LDH2 ATP2 TLV1

Cadmium chloride (TLV for Cd) 0.98 0.01 0.02 3.65 <0.01 Cobalt chloride (TLV for Co) 0.05 4.40 0.80 0.35 0.01 Cupric sulfate (TLV for Cu) 0.19 0.36 8.00 1.42 0.38 Ethanol 164.4 400.0 34.90 275.0 1000 Glycerol 248.1 200.0 250.0 64.50 2.65 Mercuric chloride (TLV for Hg) 0.01 0.01 0.08 0.003 0.012 Methanol 243.3 250.0 226.0 290.0 200.0 Phenol 5.50 70.00 17.20 5.00 5.00 Potassium cyanide (TLV for CN3) 0.38 0.73 3.18 10.25 4.70 Sodium dichromate (TLV for CrVI) 0.05 0.50 0.08 0.09 0.01 Sodium fluoride (TLV for F) 5.50 1.37 9.41 0.73 3.22 sodium selenate (TLV for Se) 1.72 2.30 7.90 27.80 0.06 Strychnine 1.39 0.42 1.68 1.50 0.01

1 the occupation exposure levels were taken from the recommendations of the 1999-2000 chemical substances threshold limit values (TLVs) from ACGIH. Most TLV values were converted from mg/m3 in air to ppm based on Equation 5-4, except ethanol, methanol and phenol. 2NOEC values were divided by a factor of 10, as a marginal safety factor, to shift the no observable effect level on the curve further to the left. 3 for CN recommended ceiling value was only found.

170 Chapter 5

Table 5-18 correlation and regression analysis between cytotoxicity data (IC50) and TLV values

Threshold Limit Value MTS NRU LDH ATP TLV r 0.76 0.80 0.82 0.71 p (2-tailed) 0.002 0.001 0.001 0.007 R2 0.60 0.65 0.70 0.55

Correlation is significant at the 0.05 level (p < 0.05).

5.5 Comparison between In vitro – In vitro Toxicity

The performance of the assays was assessed in terms of their sensitivity and specificity in comparison with other published in vitro toxicological test methods. A variety of in vitro toxicity tests have been developed for the prediction of acute toxic effects of chemical substances (Chapter 2: Section 2.2.1). Yet, only a few are suited for pair wise comparison, based on species; exposure time; cell type and end point. In this study two comparisons were performed, one with data provided by a study by Barile et al., 1994, where two laboratories independently tested the 50 MEIC reference chemicals. A second comparison was undertaken with IC50 values listed in the registry of cytotoxicity (RC) database which was compiled by ZEBET (Chapter 2: Section 2.2.3.2). The latter is a comprehensive data base, containing IC50x values (i.e. mean IC50 of several published values) for 347 chemicals, and drugs from a compilation of in vitro cytotoxicity assays.

5.5.1 Correlation of in vitro – in vitro MEIC chemicals

IC50 data derived for MTS, NRU, LDH and ATP assays were compared to the cytotoxicity concentrations of chemicals determined with three independent in vitro cytotoxicity testing protocols (Table 5-19). The protocols used short term, 18 to 24 hour exposures with established mammalian cell lines and tobacco pollen tubes. The three 3 assays were: [ H]Proline assay performed on rat lung epithelial cells; HepG2 assay, using HepG2 cells for the determination of total cell protein after 24 hours initial

171 Chapter 5

seeding and the alcian blue staining method for the photometric quantification of pollen tube mass production was used for the Pollen Tube Growth (PTG) test (Barile et al. 1994b).

Table 5-19 Published and experimentally derived cytotoxicity values (IC50) for chemicals using different assays.

1 3 No Chemical MTS NRU LDH ATP [ H] Proline HepG2 PTG 8 Methanol 84.7 13.86 9.90 51.97 6.60 51.70 35.70 9 Ethanol 30.80 11.77 7.76 6.93 3.80 33.00 14.60 12 Phenol 1.70 0.43 0.99 0.49 0.36 0.94 0.32 13 Sodium chloride 17.32 2.30 6.90 7.30 2.90 8.90 1.70 14 Sodium fluoride 1.04 0.29 0.74 0.37 0.03 0.21 0.11 19 Potassium cyanide 0.56 0.35 0.67 0.61 0.08 0.85 <0.01 20 Lithium sulfate 25.03 1.90 8.17 8.40 2.10 0.71 0.87 27 Copper sulfate 0.03 0.20 0.15 0.38 0.15 0.28 0.28 28 Mercuric chloride 0.01 <0.01 0.01 <0.01 0.03 0.02 0.02 48 Caffeine 1.30 3.00 1.84 13.86 1.90 2.90 2.90 50 Potassium chloride 17.32 1.90 7.50 23.10 3.00 6.80 6.80

1 Number corresponds to MEIC list of chemicals; 2 all values are IC50 values in mg/ml.

A significantly high correlation was observed between the assays (MTS, NRU, LDH and ATP) and the published in vitro results (Table 5-20). The highest correlation was observed between the NRU assay and HepG2 assay (r = 0.93; R2 = 0.86; p<0.05). All assays showed higher correlations with the HepG2 assay, compared to correlations with the [3H] Proline assay and the PTG assay. This could be because both HepG2 assay and MTS, NRU, ATP and LDH assays used human derived cell line and culture.

172 Chapter 5

Table 5-20 Correlation coefficients of the in vitro assays

3 [H] Proline HepG2 PTG assay MTS r 0.84 0.86 0.66 p (2-tailed) 0.001 0.001 0.02 NRU r 0.87 0.93 0.70 p (2-tailed) <0.01 <0.01 0.016 LDH r 0.84 0.85 0.62 p (2-tailed) 0.001 0.001 0.04 ATP r 0.90 0.88 0.71 p (2-tailed) <0.01 0.01 0.015

Positively significant correlations were observed between the assays (p < 0.05).

A graphical illustration of the assays using PCA modelling can be seen in Figure 5-19. This graphically shows that the assays correlated the least with the PTG assay, followed with the [3H] Proline assay, and the HepG2 values was very similar to the assays used in this research (MTS, NRU, LDH and ATP) (same positioning on the axis).

Variables (axes F1 and F2: 94 %)

1.5

PTG MTS 0.5 [3H] LDH Proline 0 NRU HepG2 ATP ATP -0.5 LDH NRU MTS -- Axis F2 (7 %) --> Hep G2 -1 [3H] Proline PTG -1.5 -1.5 -1 -0.5 0 0.5 1 1.5 -- Axis F1 (87 %) -->

Figure 5-19 Representation of the position of the different in vitro assays from each other using PCA modelling.

173 Chapter 5

5.5.2 Correlation of in vitro – in vitro ZEBET (RC)

Experimentally derived IC50 values were compared with values from the registry of cytotoxicity (RC), developed by ZEBET. A positively significant relationship was observed between the assays and the ZEBET values. MTS (r = 0.70, p < 0.05) > LDH (r = 0.66, p < 0.05) > NRU (r = 0.60, p < 0.05) = ATP (r = 0.60, p < 0.05). The ZEBET database is a compilation of IC50 values from different sources and exposure times, this could account for the weak correlation observed.

5.6 Cluster Analysis of In Vitro Data

The aim of this study was to classify the chemicals (21) using the in vitro cytotoxicity assays (NRU, LDH, MTS and ATP) into toxicity categories based on their mean IC50 values. Hierarchical cluster analysis was performed on IC50 values for each chemical with their respective assays to determine the capability of the cytotoxicity tests to distinguish chemicals into hazard categories. The variation between chemicals was determined. The statistical package SPSS for Windows version 10 was used. Parameters for the analysis included: Euclidean distances as distance measurements and Wards method for linkage of clusters. The results of the cluster analysis are shown in Figure 5- 20 for the NRU assay. The results of the cluster analysis of the remaining assays as well as the full statistical analysis are outlined in Appendix C2. The resulting dendrogram for the NRU assay revealed 4 major chemical categories based on IC50 toxicity ((1) – (4)). The chemicals in the resulting categories are outlined in Table 5-21. The characteristic (danger label) of each category is adapted from (Grosselin et al., 1984).

Category 1 chemicals (Table 5-21) consisted of the extremely-toxic chemicals having an IC50 range between 1 and 50 mg/L. The chemicals that appear extremely-toxic as measured by the NRU assay were mercuric chloride, cadmium chloride and SLS. Upon comparison with in vivo rodent LD50 values, the NRU assay is much more sensitive to cadmium chloride and SLS. Category 2 chemicals (Table 5-21) consist of very-toxic

174 Chapter 5

chemicals (50< IC50 500 mg/L). The category consisted of the following chemicals: potassium cyanide, irgasan, phenol, copper sulfate, sodium dichromate, sodium fluoride and sodium selenate. These results were in agreement with classification of chemicals based on in vivo rodent LD50, with the exception of potassium cyanide and sodium selenate. The toxicity of both chemicals was underestimated using the NRU as an indicator of toxicity. Category 3 chemicals (Table 5-21) consisted of moderately toxic chemicals (500 < IC50 5000 mg/L). The category consisted of the following chemicals: lithium sulfate, potassium chloride, sodium chloride, colchicine, strychnine, cobalt chloride and caffeine. These results were in agreement with classification of chemicals based on in vivo rodent LD50, with the exception of colchicine and strychnine. The toxicity of both chemicals was underestimated using the NRU as an indicator of toxicity. Category 4 chemicals (Table 5-21) consisted of slightly toxic chemicals (IC50 > 5000 mg/L). The category consisted of the following chemicals: ethanol, methanol, saccharin and glycerol. These results were in agreement with classification of chemicals based on in vivo rodent LD50.

Classification of chemicals based on cluster analysis for the remaining assays, resulted in only 3 different categories between the chemicals (Appendix C2). Category 1 chemicals consisted of the extremely toxic chemicals (1 < IC50 50 mg/L), category 2 chemicals consisted of a combination of very toxic and moderately toxic chemicals (50

> IC50 2000 mg/L) and category 3 consisted of the slightly toxic chemicals (IC50 > 2000 mg/L). For category 1 chemicals, the MTS assay overestimated the toxicity of cobalt chloride, copper sulfate and SLS when compared to classification based on in vivo rodent LD50 values. The LDH assay, overestimated the toxicity of cadmium chloride, sodium dichromate and SLS. For category 2 chemicals the assays underestimated the toxicity of potassium cyanide, strychnine, colchicine and sodium selenate. The assays also underestimated the toxicity of potassium chloride, sodium chloride, lithium sulfate, with IC50 values for these chemicals exceeding 5000 mg/L.

175 Chapter 5

Dendrogram using Average Linkage (Between Groups)

Rescaled Distance Cluster Combine

C A S E 0 5 10 15 20 25 Label Num +------+------+------+------+------+

LiSO4 14 «± KCl 15 «³ NaCl 16 «³ Colchicine 12 «³ (1) Strychnine 13 ««««± CoCl2 11 «³ ¬ Caffeine 17 «° ¬ KCN 7 «± ²«««««««««««««««± Irgasan 8 «³ ¬ ¬ Phenol 9 «³ ¬ ¬ CuSO4 4 «³(2 ¬) ¬ Cr(VI) 5 ««««° ²«««««««««««««««««««««««««««± NaF 6 «³ ¬ ¬ NaSe 10 «° ¬ ¬ Ethanol 19 «± ¬ ¬ Methanol 20 ««««««««««±(3) ¬ ¬ Saccharin 18 «° ²«««««««««° ¬ Glycerol 21 «««««««««««° ¬ HgCl2 1 «´«««««««±(4) ¬ CdCl2 2 «° ²«««««««««««««««««««««««««««««««««««««««° SLS 3 «««««««««°

Figure 5-20 Hierarchical clustering of selected chemicals based on in vitro IC50 values using NRU assay

176 Chapter 5

The clustering of chemicals measured using the NRU assay, is similar to that used by the US EPA hazard classification categories for acute oral toxicity (US EPA, 2003). On the other hand, the remaining assays and ATP only clustered the chemicals into 3 categories, combining category 2 and 3 of the US EPA hazard classification system into one category.

Table 5-21 Grouping of chemicals measured with NRU assay Category 1 Category 2 Category 3 Category 4

1 < IC50 < 50 (mg/L) 50 < IC50 < 500 (mg/L) 500 < IC50 < 4000 IC50 > 5000 mg/L Mercuric chloride Potassium cyanide Lithium sulfate Ethanol

Cadmium chloride Irgasan Potassium chloride Methanol

SLS Phenol Sodium chloride Saccharin

Copper sulfate Colchicine Glycerol

Sodium dichromate Strychnine

Sodium fluoride Cobalt chloride

Sodium selenate Caffeine

5.6.1 Comparison in vitro – GHS categories

Any validation effort for in vitro tests for acute toxicity should evaluate the predictive performance of the methods for correctly estimating all of the GHS hazard categories (Gennari et al., 2004). The GHS method of classification is outlined in Chapter 4 (Table 4-1). In this study, the toxicity of chemicals measured with the NRU assay, were classified into the 6 acute oral toxicity categories of the GHS (Appendix C3). The MTS and LDH assay chemicals were classified based on their IC50 values into 5 categories.

There were no chemicals in category 5 (2000 > IC50 5000 mg/L). The toxicity of chemicals based on in vitro IC50 values measured with the ATP assay, were only classified into 4 categories. There were no chemicals in category 3 (50 < IC50 300

177 Chapter 5

mg/L) and category 5. All in all, the NRU assay classified the chemicals the closest to the GHS method of classification for acute toxicity.

5.7 Summary of Findings

In conclusion, this study demonstrated the in vitro cytotoxicity using human skin fibroblasts on a selected number of chemicals spread across the GHS categories for acute oral toxicity. Four in vitro cytotoxicity assays were used as a measure of toxicity for the selected chemicals. Mercuric chloride was found to be the most toxic chemical (GHS category 1) with the least toxic chemical being Glycerol (GHS unclassified). Statistical differences were found for certain chemicals among the different assays with the exception of mercuric chloride, irgasan, potassium cyanide and glycerol (Section 5.3.2). These differences were due to the specific mechanisms underlying the toxic response of the assay used as a measurement of toxicity and highlight the need to use a battery of in vitro assays in toxicity studies. The NRU assay was able to distinguish chemicals in adjacent hazard categories in a more precise fashion than the MTS, LDH and ATP assays (Section 5.6). The NRU clustering of chemicals based on hierarchical clustering was similar to the cut-off limits used by the US EPA for classification of chemicals based on acute oral toxicity. In addition, the NRU assay was able to classify chemicals into all of the GHS categories for acute toxicity (Section 5.6.1). There are also considerable differences in cost when choosing an in vitro cytotoxicity assay. In this study, the most expensive being the ATP assay, which uses reagents costing about AUD 83.00/96-well plate. MTS and LDH assays are comparable in price (AUD 32.80/plate and AUD 36.60/plate, respectively). The NRU assay is the cheapest to run, the actual neutral red solution costs about AUD 0.15/plate. Laboratory time differs slightly, the LDH and NRU assays are more time consuming, and require many handling steps. However, the equipment requirements differed between the ATP assay on the one hand, and the three remaining assays. Whereas most labs are equipped with 96-well plate readers, the number of microplate luminometers is much more limited, and more expensive. However, as other luminescence methods such as immunoassays, are becoming more common, luminometer availability is improving and hence cost should decrease (Petty et al., 1995).

178 Chapter 5

The specificity of the assays used, that is their potential to realistically rank the cytotoxicity of chemicals, was also determined when cytotoxicity data produced with this assay were compared to published toxicity data produced with diverse in vitro and in vivo toxicological test methods (Section 5.4 and Section 5.5). The accuracy of the cytotoxicity assays was determined by predicting the performance of the assays against data from a reference method such as the LD50, which is currently used by regulatory authorities to classify hazard. A better comparison in terms of specificity of methods could be determined when the data from the various assays were compared to data from main stream in vitro toxicity assays, using human cell lines (Section 5.5). The correlation between the assays and in vivo studies was found to be as good as in vivo-in vitro correlations described with established in vitro test methods. Good correlation between IC50 against rodent LD50 and human LDL0 was achieved when an appropriate adjustment factor was implemented to account for differences between in vitro and in vivo systems. On the other hand, a good correlation was found with data from the in vitro tests used in this study and human plasma LC without the use of adjustment factors.

Numerous tests for basal cytotoxicity have already been developed but little attention is paid as to how the data in these tests can be applied as a means of making decisions or predictions of human toxicity. In this study, good correlation with in vivo (human and rodent), published in vitro data and exposure limits was found with all the assays, In general, the NRU assay was found to be a more sensitive measure of toxicity for most chemicals than the MTS, LDH and ATP assays, specifically for chemicals that generally fell in Category 4 (LD50> 300 mg/kg) (Section 5.3). Of the chemicals tested 40% had lower IC50 values in comparison to the remaining assays used in this study (Section 5.3). At present, results show that all basal cytoxicity tests are of importance, a battery of these tests can be used in a multi-step testing strategy. The advantage of using a battery of tests identifies possible differences in toxicity of the chemicals found between the assays and possible underlying mechanisms. Data from the assay or assays that do not show false negative or false positive results can be used to predict rodent LD50 data, and adjusted data based on specialised in vitro tests that evaluate toxicokinetic parameters and bioavailability. These results can then be used to predict acute oral human toxicity.

179 Chapter 5

The outcome of this study can be further used to test not only the ability of cytotoxicity assays to determine the toxicity of single chemicals, but chemical mixtures as well. Currently, there is a need for a more scientific, accurate and reliable basis for the evaluation of the toxicological interactions between chemical mixtures (Gennari et al., 2004). This will be expanded in the next chapter.

180 Chapter 6

Chapter 6. Results and Discussions: Cytotoxicity of Chemical Mixtures

6.1 Introduction

When all four assays used were compared with each other using primary PCA, the NRU assay correlated the closest with the LDH and ATP assays, where as the MTS assay had significant but weaker correlations with the assays (as per Section 5.3). The aim of this study was to demonstrate the use of an in vitro system using human skin fibroblasts, to obtain information about the nature of interactive effects (i.e. synergistic, antagonistic, and additive) of selected binary and ternary chemical mixtures. The MTS and NRU cytotoxicity assays were used. These assays were chosen as they correlated significantly with published in vivo and in vitro data (as seen per Section 5.4 and 5.5). In addition, both the NRU and MTS assays, have been applied for the testing of chemical mixtures as can be seen from the literature (Babich and Babich 1997; Malich et al. 1998; Bae et al. 2001).

Binary and ternary mixtures composed of chemicals spread across the GHS categories of acute oral toxicity as well as unclassified toxicities, were tested at fixed ratios of their components (Table 4-6).

The test chemicals were prepared using a mixture ratio based on the relative toxicities of the components. Components mixed in the ratio of their IC50’s of the individual compounds were analysed (as seen per Section 4.3.2.2). Concentration-response analysis was performed for the 20 different mixtures at the fixed mixture ratio. Observed mixture toxicity was then compared with predictions, calculated from a mathematical model. The model was based on quantitative data developed for interpreting experimental data (Konemann 1981; Ribo and Rogers 1990; Malich et al. 1998; Preston et al. 2000). Linear regression analysis was also performed with experimental and predicted data to establish correlation between the results.

181 Chapter 6

A number of concepts have been developed by pharmacologists for the study of mixtures (as seen per Section 2.3.1). In the present study the concept of concentration addition (Altenburger et al. 2000; Faust et al. 2001), was applied. Concentration addition has been defendable as a pragmatic and precautionary default assumption, useful from a regulatory perspective (Faust et al. 2003). A study of the predictive capabilities of the concentration addition concept as a potential promising tool for the risk assessment of chemical mixtures was undertaken.

6.2 Experimental Design

6.2.1 Cell line

Human skin fibroblasts were chosen (Section 4.2.1). The fibroblast culture was cultured and maintained as per Section 4.2.2.

6.2.2 Test chemicals

Ternary and binary chemical mixtures were composed of test chemicals that covered a broad range of toxicity. Fifteen chemicals (Chapter 4: Table 4-6) were chosen from the list of 21 chemicals previously tested using the 4 in vitro cytotoxicity assays. The individual IC50 toxicity values for 15 chemicals whose in vivo toxicity was spread over the 5 Globally Harmonised System (GHS) categories and unclassified for acute oral toxicity were used to create 5 ternary mixtures and their respective binary combinations (15) (Chapter 4: Section 4.3).

Different designs exist, for combining mixtures depending on the aim of the study. For binary chemicals, variation of mixture ratio can be used (Malich et al. 1998). For a multi-component mixture study, a fixed mixture ratio design is usually preferred. In this study, the aim was to establish the type of mixture effect, therefore a mixture ratio proportional to the potency of components was chosen. Fixed mixture-ratio design

182 Chapter 6

previously described by Altenburger et al., (2000), Payne et al., (2000) and Backhaus et al., (2000) lends itself particularly well to achieving the goals of this present study.

In this study, components were mixed in the ratio of their IC50 effect (Chapter 4: Equation 4-16). The fractional effects of the individual toxicants in the mixture were calculated as a proportion of their concentration to the total concentrations of all toxicants in the mixture. This is a typical experimental approach, leading to what is called equi-toxic mixtures. It has previously been used for assessing the predictability of the joint action of chemical mixtures and predictions are made assuming additive combination effects and then tested experimentally (Konemann 1981; Altenburger et al. 2000; Payne et al. 2000; Payne et al. 2001; Rajapakse et al. 2001; Kortenkamp 2004). This is a useful approach when using the concentration addition concept, because at mixture concentrations that equal the sum of 1/n of the IC50’s of the individual components, concentration addition predicts exactly 50% effect of the mixture (Backhaus et al. 2000a). In addition, equi-toxicity combination ensures no chemical contributes disproportionately to the overall combination effect. This allowed facilitation of comparative analysis of mixture toxicity, and indicated the possible interactions that may exist in the mixture and if they differed from the expected additive approach.

Evaluation of the combined effects of agents relies critically on the method used to estimate the expected effect of a mixture. Synergism and antagonism can be defined as deviations from the expected effects, where synergism is a lower observed IC50 and antagonism is a higher observed IC50 than expected (Berenbaum 1989; Rajapakse et al. 2001). Cytotoxicity studies on chemical mixtures were conducted as outlined in Chapter 4 (Section 4.3).

6.2.3 Assays

Based on results from studies from Chapter 5, two assays (MTS and NRU) were selected from the battery of cytotoxicity assays. A strong correlation was established between the assays (Section 5.3.2). The NRU assay had strong correlation with LDH

183 Chapter 6

and ATP assays, the weakest being with the MTS assay. The PCA modelling (Section 5.3.2.1) graphically presented the MTS assay as having the weakest correlation against the assays.

The protocol for the study of mixture toxicity using both cytotoxicity assays (MTS and NRU), was similar to that used for the individual chemicals, and as described in Chapter 4 (Sections 4.2.6.1 and 4.2.6.2).

6.2.4 Dose response curve

Dose response curves were plotted for the selected chemicals after correction by subtracting the background absorbance from the controls (as per Section 4.2.5). IC50 values (50% inhibitory concentration) for the mixtures were derived from the plotted absorbance data based on the statistical method detailed in Chapter 4 (Section 4.2.7.3.1).

6.3 Cytotoxicity Results

The in vitro cytotoxicity results (IC50) for the selected chemical mixtures on human skin fibroblast cell culture as determined with the MTS and NRU assays are presented in Table 6-1. Experiments for each chemical mixture were repeated on three separate occasions, to assess inter-test variations and reproducibility. All values (mg/L) were expressed as mean and ± standard deviation (m ± SD). In general, the results of the in vitro cytotoxicity tests indicated that the mixture SLS: Irgasan (IC50 46 mg/L) was the most cytotoxic as measured with the MTS assay, whereas the binary mixture

(HgCl2:CdCl2) (IC50 23 mg/L) had the highest cytotoxicity using the NRU assay. In contrast, the binary mixture (EtOH: LiSO4) (IC50 11 000 – 35 000 mg/L) was the least toxic among chemical mixtures tested using both assays.

184 Chapter 6

Table 6-1 Cytotoxicity data (IC50’s) for chemical mixtures using the MTS and NRU assays.

MTS assay NRU assay

IC50 experimental ±SD IC50 experimental ± SD SLS; NaF; Irgasan 0.38 0.07 0.31 0.01

NaF; SLS 0.79 0.04 0.50 0.20

NaF; Irgasan 2.04 0.24 0.69 0.22

SLS; Irgasan 0.08 0.01 0.14 0.01

CrVI; CuSO4; Phenol 0.90 0.07 4.20 2.07

CrVI; CuSO4 0.19 0.01 1.07 0.30

CuSO4; Phenol 4.20 1.18 4.10 0.12

CrVI; Phenol 2.66 0.25 2.41 0.18

EtOH; CdCl2; LiSO4 50.06 7.18 17.23 0.01

CdCl2; LiSO4 28.17 4.06 3.40 0.84

EtOH; LiSO4 >172.39 NA 46.46 5.75

EtOH; CdCl2 >175.39 NA 40.08 1.46

KCN; HgCl2; CdCl2 2.99 0.94 0.62 0.051

KCN; HgCl2 2.69 0.21 1.70 0.31

KCN; CdCl2 6.94 2.89 1.26 0.28

CdCl2; HgCl2 4.62 1.32 0.05 0.01

KCl; NaF; Phenol 36.55 0.00 9.39 1.10

KCl; NaF 2.59 19.07 12.21 2.49

KCl; Phenol >118.58 NA 16.33 4.56

NaF; Phenol 25.45 7.34 3.53 0.53

SD (mmol/L) represents standard deviation for at least 3 separate experiments. IC50exp (mmol/L) represents inhibitory concentration 50% for MTS assay as per Section 4.2.5.3 NA Not applicable

185 Chapter 6

6.4 Comparison between Assays

In general, for the total of 20 chemical mixtures measured using both the MTS and NRU assay (Figure 6-1):

1- IC50 MTS > IC50 NRU for 15 chemical mixtures (§ 75%);

2- IC50 NRU > IC50 MTS for 3 chemical mixtures (§ 15%);

3- IC50 MTS § IC50 NRU for 2 chemical mixtures (§ 10%).

These results showed that measurement of toxicity using the NRU assay was more sensitive than the MTS assay. The results were comparable to results obtained from the selected individual 21 chemicals described in Chapter 5 and measured using the MTS and NRU assay (Section 5.3.2).

6.4.1 Correlation in vitro – in vitro

Correlation between IC50 values for the chemical mixtures was performed to establish an in vitro – in vitro correlation between the MTS and NRU assays. A significant but weak correlation was initially observed between both assays (R2 = 0.45, p < 0.05) (Table 6-2). One chemical mixture affected the weak correlation, and was considered an outlier as can be seen from Figure 6-2. When the chemical mixture was removed the correlation improved significantly (R2 = 0.70, r = 0.80) (Table 6-2). The mixture was

CdCl2:HgCl2, and the NRU (IC50) < MTS (IC50) for the mixture (Table 6-1). The reason for the dissimilarity was in the CdCl2 found in the mixture, whose NRU IC50 < MTS

IC50 when tested individually. As described in Chapter 5 (Section 5.4.2), CdCl2 has been found to damage the intactness of the epithelial barrier in vitro. The NRU assay measures membrane integrity through the uptake of the neutral red dye in the lysosomes of viable cells; cadmium chloride toxicity could also have a localized effect upon the lysosomes resulting in lower cell viability than with other assays.

186 Chapter 6

IC50 (mg/L) (Log scale) 100000 10000 1000 00 10 10 1

CdCl2, HgCl2

SLS, Irgasan

SLS, NaF, Irgasan

NaF, Irgasan

KCyanide, HgCl2, CdCl2

Kcyanide, CdCl2 IC50 nru IC50

Chemical mixtures NaF, SLS

NaF, Phenol

KCyanide, HgCl2

Cr(VI), CuSO4 IC50 mts Cr(VI), Phenol

CdCl2, LiSO4

CuSO4, Phenol

KCl, NaF

KCl, NaF, Phenol

Cr(VI), CuSO4, Phenol

KCl, Phenol

EtOH, CdCl2, LiSO4

EtOH, LiSO4

EtOH, CdCl2

Figure 6-1 Bar plot of cytotoxicity values (IC50) for chemical mixtures measured with MTS and NRU assays. Error bars represent m ± SD for 3 separate experiments.

187 Chapter 6

Table 6-2 Regression analysis between experimental IC50 values for selected chemical mixtures

using MTS and NRU assays with the addition and removal of CdCl2:HgCl2 binary mixture

IC50 values NRU MTS r 0.67 p (2-tailed) 0.001 R2 0.45 N 20 MTS r 0.80 p (2-tailed) < 0.001 R2 0.70 N 19 r represents correlation coefficient as determined from the analysis of variance (ANOVA) p determines significance of relationship between MTS (IC50) values and NRU (IC50) values. Since the p-value is less than 0.01 then there is a statistically significant relationship at the 99% confidence level.

N represents the number of chemical mixtures correlated together. For N = 19, the mixture CdCl2:

HgCl2 was removed as an outlier. R2 represents the coefficient of determination that explains the percentage of results that can be explained by the regression model obtained between the two sets of data.

4 (mg/L)

50 3 NRU vs MTS Linear (NRU vs MTS) 2

CdCl2:HgCl2 1 Log NRU IC

0 135

Log MTS IC50 (mg/L)

Figure 6-2 Regression plot for selected chemical mixtures using MTS IC50 values against NRU IC50 values.

Values represent Log mean IC50 values from three independent experiments of chemical mixtures (20) using the MTS assay against the NRU assay. Correlation coefficient for MTS vs. NRU (R2 = 0.45) and 2 with the removal of outlier mixture CdCl2:HgCl2 (R = 0.65).

188 Chapter 6

6.4.2 Comparison of in vitro data with GHS categories

The harmonised criteria for the acute toxicity of substances for mixtures are as described in Chapter 4 (Table 4-1). The criteria for substances classify acute toxicity by use of lethal dose data (tested or derived), mostly derived from in vivo lethality tests (Chapter 2: Section 2.3.2). For mixtures, it is necessary to obtain or derive information that allows the criteria to be applied to the mixture for the purpose of classification. Development of an in vitro alternative to the in vivo lethality test, takes into account the in vitro tests’ ability to accurately classify chemicals in acute toxicity hazard categories according to current classification schemes (Gennari et al. 2004).

In this study, the MTS and NRU assays were evaluated for the predictive performance of the methods for correctly estimating all of the GHS hazard categories. Figure 6-3 and Figure 6-4; graphically present the chemical mixtures into the different categories based on the criteria for the GHS as the cut-off point for each category. The classifications of the selected chemical mixtures are presented in a tabular form (Appendix D3). The chemical mixtures for both the MTS and NRU assays were classified into the 5 categories of the GHS and unclassified for acute oral toxicity. Characteristic (danger label) for each category was adapted from (Gosselin et al., 1984).

Classification based on the NRU assay (Figure 6-2):

Category 1 chemicals: consisted of super-toxic chemicals (IC50 5 mg/L). Mercuric chloride was the only chemical in the category. No mixture fell into category 1.

Category 2 chemicals: consisted of extremely toxic chemicals (5 < IC50 50 mg/L). The category consisted of only one mixture: CdCl2:HgCl2. The IC50 of CdCl2 when tested individually would place the chemical in category 2.

Category 3 chemicals: consisted of very toxic chemicals (50 > IC50 300 mg/L). The category consisted of 4 mixtures: SLS:NaF:Irgasan and its respective binary combinations. When tested individually only NaF would be classified as very toxic

189 Chapter 6

based on its IC50 value. Based on the in vitro cytotoxicity of SLS measured with the NRU assay, it is classified as extremely toxic, whereas irgasan would be classified in category 4. However, the overall toxicity of the ternary mixture and its binary components were considered very toxic chemicals.

Category 4 chemicals: consisted of moderately toxic chemicals (300 < IC50 2000 mg/L). The majority of the tested chemical mixtures were classified in this category. The category consisted of 10 mixtures as can be seen from Figure 6.3.

Category 5 chemicals: consisted of slightly toxic chemical mixtures (200 > IC50 5000 mg/L). The category consisted of 2 chemical mixtures: Cr (VI):CuSO4:phenol and KCl: phenol. Both Cr (VI) and CuSO4 when tested independently were classified as very toxic based on their IC50 values, whereas phenol would be classified as moderately toxic.

Category 6 chemicals: or unclassified category consisted of practically non-toxic chemicals (IC50 > 5000g/L). The category consisted of 4 chemical mixtures: EtOH:

CdCl2:LiSO4 and its respective binary combinations. LiSO4 when tested independently was classified as moderately toxic and CdCl2 as extremely toxic, on the other hand EtOH was classified as practically non toxic, and therefore affecting the overall toxicity of the mixture.

Classification based on the MTS assay (Figure 6-4):

Category 1 chemicals: There were no chemicals in this category.

Category 2 chemicals: consisted of only one mixture: SLS: Irgasan. SLS when tested independently was classified in the same category; however, irgasan was classified in category 3.

Category 3 chemicals: consisted of 3 chemical mixtures: SLS:NaF:Irgasan; NaF:SLS and Cr(VI): CuSO4. Both Cr(VI) and irgasan when tested independently were classified

190 Chapter 6

in the same category as the mixtures. On the other hand both SLS and CuSO4 are classified in Category 2 when tested independently using the MTS assay.

Category 4 chemicals: consisted of 6 chemical mixtures, as can be seen from Figure 6-4.

Category 5 chemicals: consisted of 3 chemical mixtures: KCN:CdCl2; CdCl2:HgCl2 and

NaF:phenol. HgCl2 when tested independently was classified in category 2; the remaining chemicals (CdCl2, NaF and phenol) belonged to category 4 when tested independently.

Category 6 chemicals: consisted of the majority of chemical mixtures (7).

6.4.3 Reproducibility of the in vitro assays

Reproducibility of results was compared between the MTS and NRU assays in their measurement of cytotoxicity of the selected chemical mixtures. Reproducibility of results was calculated as in Chapter 5 (Section 5.4.3; Equation 5-1). The reproducibility of the NRU assay (CoV = 17.14%) was similar that of the MTS assay (CoV = 17.45%) for chemical mixtures. This is comparable to the reproducibility of the assays for single chemicals (21) where the NRU assay had a better reproducibility than the MTS assay.

191 Chapter 6

Et OH Et OH, LiSO4 Et OH, CdCl2, LiSO4 Category 5 KCl, Phenol Cr(VI), CuSO4, Phenol Category 4 KCl LiSO4 KCl, NaF, Phenol KCl, NaF CuSO4, Phenol CdCl2, LiSO4 Cr(VI), Phenol Cr(VI), CuSO4 KCN, HgCl2 NaF, Phenol NaF, SLS Phenol Irgasan KCN, CdCl2 KCN KCN, HgCl2, CdCl2 Test chem icals Category 3 NaF NaF, Irgasan Cr(VI) CuSO4 SLS, NaF, Irgasan SLS, Irgasan Category 2 SLS CdCl2, HgCl2 CdCl2 Category 1 Hg C l2 012345

Log NRU IC50 (mg/L)

Figure 6-3 Classification of chemical mixtures measured with NRU assay (IC50) The red bars represent cut off limits for categories 1-5 based on the GHS classification.

192 Chapter 6

Et OH, CdCl2 Et OH Et OH, LiSO4 EtOH, CdCl2, KCl, Phenol LiSO4 KCl CdCl2, LiSO4 KCl, NaF KCl, NaF, Category 5 NaF, Phenol CdCl2, HgCl2 KCN, CdCl2 Category 4 Phenol CdCl2 KCN, HgCl2, CuSO4, Phenol NaF Cr(V I), Phenol KCN, HgCl2 NaF, Irgasan

Test chem icals Cr(VI), CuSO4, Category 3 NaF, SLS Irgasan SLS, NaF, KCN Cr(VI) SLS Cr(VI), CuSO4 Category 2 SLS, Irgasan CuSO4 Hg C l 2 Category 1

012345

Log MTS IC50 (mg/L)

Figure 6-4 Classification of chemical mixtures measured with the MTS assay (IC50) The red bars represent cut off limits for categories 1-5 based on the GHS classification.

193 Chapter 6

6.5 Data Analysis

6.5.1 Prediction model

A prediction model used to generate predicted IC50 values based on an expected additivity outcome was applied. The prediction model is widely used in the literature (Konemann 1981; Ribo and Rogers 1990; Malich et al. 1998; Preston et al. 2000).

CM IC50 (predicted) [6-1] CCCABC C n ... IC50A IC 50B IC 50C IC 50n

CA,CB, CC… Cn represent the final concentrations of chemicals A; B; C…n found in the mixtures in (mg/L). It is calculated

from the ratio of the components in the solution (Xi) (Equation 4-23) and the chemical concentration in cell solution of each chemical (Equation 6-2).

Xi Represents the ratio which chemicals in the mixture are combined at. It is derived from Equation 4-23. Xi values for the selected chemical combinations are found in Chapter 4 (Tables 4-8 and 4-9).

CM represents the sum of concentrations of the components in the mixture (mg/L).

IC50A, IC50B, IC50C, IC50n represent 50% inhibitory concentrations of each component in the mixture (mg/L), when administered individually

IC50pred represents expected 50% inhibitory concentration of the relevant mixture (mg/L), if chemical interact in an additive manner.

194 Chapter 6

For 3 chemicals (A, B and C) Equation 6-1, can be modified to calculate predicted IC50 values as follows:

CXICAA u 50A

CXICBB u 50B [6-2] CXICCC u 50C

And knowing that accordingly

CCCCMABC [6-3]

As per Section 4.3.2.2 the sum of all ratio combinations equaled to a unit value of one such that:

XXX1ABC [6-4]

The calculation of IC50 predicted based on an assumption of additivity and for n number of chemicals was calculated using Equation 6-5 (modified Equation 6-1):

IC50 (predicted) XA u IC 50A X B u IC 50B X C u IC 50C .... X n u IC 50n [6-5]

Further, a 95% confidence interval (CI) was calculated to account for the variability around predicted IC50 value, and for setting upper and lower limits (Equation 6.6).

IC50(predicted) r t u s(IC50pred ) 3 [6-6]

S (IC50 predicted) represents the standard deviation around predicted IC50 value for 3 experiments. Method of derivation is found in Appendix D1. t represents student t test for a 95% CI for calculated degree of

freedom (df) [(nA-1)+(nB-1)+…]. Values are taken from the 2 sided t-Table for 95% CI in the handbook of Chemometrics and Qualimetrics (Massart et al. 2003).

195

Chapter 6

Cl, Cl,

H, CdCl2 H,

IC50 experimental IC50 predictedIC50 O EtO

O OH, LiS OH,

LiS

SO i

EtOH, CdCl2, CdCl2, EtOH, a

dl,L CdCl2,

l KCl, N KCl,

F, ph F,

a l

KCl, N KCl, ,

NaF

gCl

CdCl2, CdCl2,

KCN, CdCl KCN,

,HC2 CdCl HgCl2, N,

K O4, Pheno O4,

S mixtures Chemical

CrVI, Phenol CrVI, g

, H ,

KCN

, Irgasan ,

F ol

, SL ,

n Na

sa CrVI, CuSO4, Phen CuSO4, CrVI,

, Irga ,

CuSO

SLS, SLS, I,

CrV a , Irg ,

1 10 SLS 100

1000

10000

100000

(mg/L) IC MTS 50

Figure 6-5 Bar plot of experimental and predicted cytotoxicity values (MTS IC50) for chemical mixtures

Error bars represent m ± 95% CI for 3 separate experiments for experimental IC50, and calculated 95% CI for predicted IC50 values

196

Chapter 6

l2, l2,

LiSO4

IC50 experimental IC50 predictedIC50 4

EtOH

phenol

, l

EtOH, CdCl2, LiSO CdCl2, EtOH,

KCl

heno

CuSO4, CuSO4,

I, I,

,ph F, F

CrV

,Na

KCl l K

Pheno ,

SO4

L CuSO4

2 l

nol dC

Phe

VI, VI,

C uSO

CrVI, CrVI,

nol

KCN, HgCl2 KCN, mixtures Chemical e

ph ,

CdCl2

KCN Cl dC

gCl2 sa

,H N,

F, Irg F,

KC sa

a L

L,N SLS, Na

S 2

gCl Hg , 1 2

100 CdCl

1000

(mg/L) IC NRU 10000 50 100000

Figure 6-6 Bar plot of experimental and predicted cytotoxicity values (NRU IC50) for chemical mixtures

Error bars represent m ± 95% CI for 3 separate experiments for experimental IC50, and calculated 95% CI for predicted IC50 values.

197 Chapter 6

A bar plot of experimental and predicted cytotoxicity values for the MTS and NRU assays was formed for the chemical mixtures (Figure 6-5 and Figure 6-6). In addition the predicted IC50 values derived using Equation 6-5 were plotted against MTS and

NRU cytotoxicity data (IC50) for binary and ternary mixtures, respectively (Appendix

D2). Figures also displayed individual IC50 values for the 15 single chemicals. For these substances experimental and predicted data were identical. The Figures were designed considering a diagram that was used by Sprague (1970) and Malich et al., (1998). Accordingly the different cases of chemical interaction were identified (additivity, antagonism and synergism). Each figure displays a straight line that represents a 1:1 correlation between predicted and experimentally derived toxicity data (MTS and NRU). Hence, if mixture toxicity is represented by points above the line then the effects are antagonistic, meaning that the predicted effects overrate the experimentally determined effects. Conversely, if the experimental mixture values are represented by points below the line then the effects are considered to be synergistic, since the predicted effects would have underrated the experimentally determined effect. IC50 values could not be determined experimentally for two mixtures using the MTS assay ((EtOH: LiSO4 and phenol: KCl); a full dose response curve could not be derived, indicating antagonism. The mathematical model used, assumed the components have similar mechanism of action and did not account for lower or greater than additive effects. This must not be assumed for the test chemicals used in this study. However, linear regression analysis indicated the usefulness of this model even for the study of mixtures with disparate chemical structures. High coefficients of determination (R2) and highly significant correlations (r) were determined between experimental and predicted values. For binary mixtures tested with MTS and NRU (R2 = 0.80; r = 0.87; p<0.05) and (R2 = 0.76, r = 0.87, p<0.05), respectively. For MTS and NRU cytotoxicity studies on ternary mixtures (R2 = 0.88, r = 0.94, p<0.05) and (R2 = 0.55, r = 0.74, p<0.05), respectively (Table 6-3).

198 Chapter 6

Table 6-3 Correlation and regression analysis of IC50 predicted and experimental for binary and ternary chemical mixtures

IC50 experimental MTS mixtures NRU mixtures ternary binary ternary Binary

IC50 predicted r 0.94 0.87 0.74 0.87

p 0.01 <0.01 0.15 <0.01

R2 0.88 0.80 0.55 0.76

Binary and ternary mixtures showed lower correlation coefficients against predicted

IC50 values for the NRU assay than the MTS assay. This could be related to the ability of the NRU assay to be more sensitive to chemical interaction and therefore has a greater ability to detect chemicals that deviate from the additivity line. Lower coefficients of determination, observed for ternary mixtures can also indicate that simple similar joint action, and therefore the use of the model, seems to decrease with more complex chemical mixtures. It must also be remembered that the calculation of coefficient of determination values (R2) and correlation factor (r) did not include the mixtures for which cytotoxicity data could not be determined, and that did not seem to exert an additive effect. Thus R2 and r values given in the figures might overestimate the correlation between predicted and experimentally derived effects. All in all, when the performance of the mathematical model for the calculation of additivity was compared with observed results, a high correlation was established. This confirmed the general concept of additivity being the most common form of mixture toxicity (Ikeda 1995; Malich et al. 1998).

199 Chapter 6

6.5.2 Concentration Addition (CA) model

The concept of concentration addition is usually defined for a binary mixture of substances A and B according to Equation 6-7.

CCAB 1 [6-7] IC50A IC 50B

CA and CB represent individual concentrations of substances A and B which are present in a mixture and elicit a 50% inhibitory effect.

IC50A and IC50B represent equivalent 50% inhibitory effect concentrations of the chemicals A and B when administered individually.

Equation 6-7 can be extended for any number n of components and any effect x (Berenbaum 1985).

n Ci ¦ 1 [6-8] i1 ICxi

Ci represents the individual concentrations of the substances 1 to n which are present in a mixture that elicits the effect x (in this study = 50% inhibitory effect).

ICxi denotes the equivalent inhibitory concentrations of the single substances when administered alone and would also cause the same quantitative effect x as the mixture.

Based on Equation 6-8, if the sum of all individual toxicants is additive, then calculation of the ratio of ICx should approach unity plus or minus the relevant level of variability. An effect greater than unity, would be an antagonistic effect and synergistic interaction if less than unity.

200 Chapter 6

If the ratio of the mixture components is known, the concentration of each component can then be expressed as a fraction of the total concentration. Consequently for the calculation of the effect concentration predicted by concentration addition, Equation 6-8 can then be rewritten as

n §ICXi · ICx(mixture) ¦ ¨ ¸ [6-9] i1 ©Xi ¹

ICxi denotes the equivalent inhibitory concentrations of the single substances when administered alone and would also cause the same quantitative effect x as the mixture.

ICx (mixture) represents predicted 50% inhibitory concentration of the relevant mixture (mg/L), if chemical interact in an additive manner.

Xi denotes the fraction of individual chemical i in the mixture i individual chemical in the mixture

The concentration addition (CA) ratios calculated according to Equation 6-8 from experimental IC50 values for the selected chemical mixtures are listed in Table 6-4. An interaction effect is also listed in Table 6-4, deduced from Equation 6-8, such that if the sum of all individual toxicants is additive, then calculation of the ratio of ICx should approach unity plus or minus the relevant level of variability. An effect greater than unity, would be an antagonistic effect and synergistic interaction if less than unity.

The two models used in the assay: the mathematical model (Equation 6-1) and the concentration addition model (Equation 6-8) gave similar results. This is shown upon a comparison of Figures 6-5 and 6-6 with the CA ratio in Table 6-4. Two mixtures measured with the NRU assay were an exception and they were KCN: CdCl2 and

CdCl2: KCN: HgCl2. Upon comparison between the experimental IC50 of both mixtures and predicted value using mathematical model (Equation 6-1) the chemicals in the mixture tend to interact in an additive manner (IC50 (mixture) § IC50 (predicted)) and as can be seen visually from Figure 6-6. However, when the experimental IC50 value for

201 Chapter 6

the mixtures was used to calculate the CA ratio using Equation 6-8, the ratio for both mixtures was (2.03 and 2.85) > 1 therefore both mixtures tended to interact in an antagonistic manner (Table 6-4). The concentration addition model appeared to overestimate the toxicity of the mixture. The IC50 (CdCl2) < IC50 (KCN), therefore the fractions of the individual chemicals in the mixture X (CdCl2) was much smaller, almost negligible as compared to X (KCN) (Table 4-9). In this case, both the IC50i and Xi tend towards zero, therefore the relation between the two become indeterminate (a high error). If for a chemical the IC50i is very small as was the case for these mixtures, then experiments should have been done with a higher fraction of each chemical (e.g. Xi > 0.5) to make the error smaller.

202 Chapter 6

Table 6-4 Concentration addition ratio and interaction effect of chemical mixtures using MTS and NRU assays.

MTS assay NRU assay CA ratio Interaction CA ratio Interaction

SLS; NaF; Irgasan 0.54 Synergistic 0.80 Synergistic

NaF; SLS 0.49 Synergistic 0.95 Additive

NaF; Irgasan 1.05 Additive 0.68 Synergistic

SLS; Irgasan 0.30 Synergistic 0.39 Synergistic

CrVI; CuSO4; Phenol 0.90 Synergistic 9.60 Antagonistic

CrVI; CuSO4 1.05 Additive 2.81 Antagonistic

CuSO4; Phenol 1.62 Additive 4.49 Antagonistic

CrVI; Phenol 1.15 Additive 2.97 Antagonistic

EtOH; CdCl2; LiSO4 1.05 Additive 1.52 Additive

CdCl2; LiSO4 0.80 Additive 1.25 Additive

EtOH; LiSO4 2.00 Antagonistic 1.18 Additive

EtOH; CdCl2 2.00 Antagonistic 2.00 Antagonistic

KCN; HgCl2; CdCl2 2.70 Antagonistic 2.85 Additive

KCN; HgCl2 5.00 Antagonistic 3.18 Antagonistic

KCN; CdCl2 2.20 Antagonistic 2.03 Additive

CdCl2; HgCl2 3.00 Antagonistic 2.69 Antagonistic

KCl; NaF; Phenol 1.15 Additive 1.87 Additive

KCl; NaF 1.04 Additive 1.28 Additive

KCl; Phenol 2.00 Antagonistic 2.34 Antagonistic

NaF; Phenol 2.60 Antagonistic 1.34 Additive

CA ratio : value derived by the application of Equation 6-8 Interaction: represents interpretation of nature of toxicity interaction of selected mixture based on graphical interpretation, and CA ratio.

203 Chapter 6

6.5.3 Graphical representation for binary mixtures

The additivity approach is a prudent method for estimating the toxicity of a mixture where individual components in a mixture have a common mechanism of action. However, if this method is used to assess the toxicity of a mixture of chemicals with known toxicity values and known concentrations present, it can be used to indicate the possible interactions that might exist in the mixture. For example, once such a calculation is made, equations and graphical representations can then be used to indicate whether a measured interaction is: o Additive (interaction curve is between the upper and lower curve of interaction variability in Figure 6-8); o Antagonistic (interaction curve is above the variability curve in Figure 6-8); o Synergistic (interaction curve is below the variability curve in Figure 6-8).

Graphically, determining the interaction of two toxicants in a mixture is shown in Figures 6-9, 6-10, and 6-11. The remainder of the graphs for binary mixtures is located in Appendix D4. Table 6-5 and Table 6-6 list a summary of the parameters and calculations needed to represent and interpret mixture interaction graphically for NaF: phenol, using the MTS assay. The line in Figure 6-7 indicates the additive interaction between the IC50’s of NaF, as it changes with increasing presence of phenol (assuming a toxicant interaction). In such interactions, the estimated ratio of IC50’s would be equal to unity.

204 Chapter 6

Table 6-5 Summary of parameters for graphical representation of binary mixtures (NaF: Phenol) using the MTS assay.

2 2 Chemicals IC50i s Xi X s Ni NaF 1045 95.26 0.38 0.15 9074 3

Phenol 1655 200.0 0.61 0.40 40000 3

IC50i represents mean values in mg/L for the chemicals when tested independently using MTS assay. s represents standard deviation of each chemical for at least three separate experiments.

Xi ratio of IC50i used to form the mixture combination (ratio of component in the solution) (Table 4-9).

Ni number of experiments repeated for each chemical when tested independently. i individual chemical

Table 6-6 Parameters for the calculation of predicted IC50 value and 95% CI upper and lower limits

2 XNaF Xphenol s (pred) t*s IC50 (pred) IC50 (pred) IC50 (pred) (*104) (+t*s) (-t*s) 0 1 4.00 555.2 1655 2210 1099

0.25 0.75 2.31 421.6 1502 192 1080

0.39 0.61 1.64 355.4 1419 1774 1063

0.5 0.5 1.23 307.4 1350 1657 1042

0.75 0.25 0.76 242.0 1197 1439 955.5

1 0 0.91 264.4 1045 1309 780.7 s2 pred square of the standard deviation. Method of derivation can be found in Appendix D4.

IC50pred derived from Equation 6-5 t*s calculation of 95% confidence interval based on student t test (df = 4).

205 Chapter 6

2000

additive interactions

1000 (mg/L) 50 IC

0 0.00 0.50 1.00

100% NaF %NaF:Phenol 100% phenol

Figure 6-7 Theoretical additive interaction of a binary mixture of NaF and phenol

3000 antagonistic interactions

1500 (mg/L) additive interactions 50 IC

synergistic interactions 0 0.0 0.5 1.0 100% NaF %NaF:Phenol 100% phenol

Figure 6-8 Graphical representation of 95% CI around additive line for binary mixture of NaF and phenol

The following graphs are selected examples of the different types of interactions found between the tested mixtures: antagonistic (Figure 6-9); additive (Figure 6-10) and synergistic (Figure 6-11).

206 Chapter 6

4000 antagonistic interactions exp

2000 (mg/L) 50

IC additive interactions

synergistic interactions 0 0.00 0.50 1.00 100% NaF %NaF:Phenol 100% phenol

Figure 6-9 Example of antagonistic interaction (NaF: Phenol) (MTS)

Based on the discussion above the interaction between NaF and phenol is antagonistic. The experimental value lies above the area of additive interaction, this is confirmed when values are added to the concentration addition model (Equation 6-8) giving a ratio of 2.60 (Table 6-4). Error bars around experimental value represents ±95% confidence interval for three separate experiments. The dashed (---) line represents calculated 95% confidence interval criterion used to assess results statistically (Equation 6-6).

3000

antagonistic interaction

1500 exp. (mg/L)

50 additive interaction IC

synergistic interaction

0 00.51 100% CuSO4 %CuSO4:phenol 100% phenol

Figure 6-10 Example of an additive interaction betweenCuSO4 and phenol (MTS).

207 Chapter 6

Based on the discussion above the interaction between copper sulfate and phenol is found to be additive. The experimental value lies within the area of additive interaction, this is confirmed when values are added to the concentration addition model (Equation 6-8) giving a ratio of 1.4 close to unity (Table 6-4). Error bars around experimental value represents ± 95% confidence interval derived for three separate experiments. The dashed (---) line represents calculated 95% confidence interval (Equation 6-6).

1500

antagonistic interactions

1000 additive interactions (mg/L) 50

IC 500 Exp synergistic interactions 0 00.51

100% SLS %SLS:NaF 100% NaF

Figure 6-11 Synergistic interaction between SLS and NaF (MTS)

Based on the discussion above the interaction between SLS and NaF is synergistic. The experimental value lies below the area of additivity. The ratio based on CA model is 0.46. Error bars around experimental value represents ± 95% confidence interval derived for three separate experiments. The dashed (---) line represents calculated 95% confidence interval (Equation 6-6).

6.5.4 Graphical representation for ternary mixtures

Multidimensional representations were needed to represent ternary mixtures. Statistica (StatSoft, Inc., US) software was used to plot the ternary graphs. Ternary plots (trace plot) graph option was used for plotting the data. 95% CI was used for calculating variability for setting upper and lower limits around the additivity area (Equation 6-6).

208 Chapter 6

X-axis, Y-axis and Z-axis were used to plot the chemical interaction, forming a

triangular area for nature of effect. Table57H6 6-7 and 6-8 give an example of the parameters needed to plot a ternary mixture (e.g. CdCl2:LiSO4: EtOH), in a multidimensional graph. Figures 6-12 to 6-14 are the graphical plots for the mixture CdCl2:LiSO4: EtOH using the values from Table 6-7 and Table 6-8. Figures 6-15 and 6-16 represent selected examples of different ternary mixtures representing the different types of interactions (synergism, antagonism). Remaining tables and graphs for all tested mixture combinations are found in Appendix D5.

Table 6-7 Parameters for ternary mixture (CdCl2: LiSO4: EtOH)

2 2 IC50i s Xi X S Ni 4 CdCl2 1500 200 0.026 0.0001 4×10 3

4 LiSO4 25030 8800 0.437 0.191 7744×10 3

EtOH 30806 6000 0.537 0.288 3600×104 3

IC50i represents mean values for the chemicals when tested independently using MTS assay. s represents standard deviation of each chemical for at least three separate experiments.

Xi ratio of IC50i used to form the mixture combination (ratio of component in the solution)

Ni represents number of repeat experiments for each chemical. i individual chemical

209 Chapter 6

Table 6-8 Parameters for the calculation of predicted IC50 value ± 95% CI for ternary mixture

(CdCl2: LiSO4: EtOH)

2 x-axis y-axis z-axis s pred t*s IC50pred IC50pred IC50pred IC50exp (CdCl2) (LiSO4) (EtOH) (+ ts) (– ts) (*104) 0 0 1 3600 14682 30806 45488 16124 N/A 0 1 0 7744 21533 25000 46563 3496 N/A 1 0 0 4 489 1501 1990 1011 N/A 0.026 0.437 0.537 2516.9 12276 27519.9 3796 15243.4 23105

s2 pred square of the standard deviation. Method of derivation outline in Appendix C

IC50pred derived from equation 6-5

IC50exp extrapolated from concentration-response curve for chemical mixture using MTS assay as per Section 4.2.7.3.1.

In Figures 6-14 to 6-16 error bars around experimental value represent ± 95% confidence interval derived for three separate experiments for the chemical mixture. The purple and green lines represent calculated ± 95% confidence interval (Equation 6-6). In

Figure 6-14: The interaction between CdCl2: LiSO4: EtOH is additive. The experimental value lies within the area of additivity. The ratio based on CA model is 1.05 (Table 6-4). In Figure 6-15: The interaction between SLS: Irgasan: NaF is synergistic. The experimental value lies below the area of additivity. The ratio based on CA model is

0.54 (Table 6-4). In Figure 6-16: The interaction HgCl2: KCN: CdCl2 is antagonistic. The experimental value lies above the area of additivity. The ratio based on CA model is 2.70 (Table 6-4). The bars in Figure 6-16 represent the limits of each interaction region and can be visually seen.

210 Chapter 6

100% LiSO4

pred. 100% EtOH

additive interaction

)

(

C I

100% CdCl2

CdCl2:LiSO4:EtOH

Figure 6-12 Region of additive interaction in a ternary mixture (CdCl2:LiSO4: EtOH).

antagonistic interaction

)

L / n

g io ct m 100% LiSO 4 a 100% CdCl 2 ( er 0 t

5 in e C 100% EtOH I iv it d ad

synergistic interaction

CdCl2:LiSO4:EtOH

Figure 6-13 Plot of upper and lower limits for theoretically predicted additive interaction and interaction areas

211 Chapter 6

antagonistic interaction

100% LiSO4

pred n io 100% EtOH exp ct ra te in e (mg/L) tiv di 50 ad IC

synergistic interaction 100% CdCl2

CdCl2:LiSO4:EtOH

Figure 6-14 Toxicological interaction between CdCl2:LiSO4: EtOH (MTS)

antagonistic interaction

100% NaF pred n tio 100% Irgasan ) c ra L te / in g ive dit m d ( a

0 exp.

synergistic interaction

100% SLS SLS:Irgasan:NaF

Figure 6-15 Toxicological Interaction between SLS: Irgasan: NaF (MTS)

212 Chapter 6

antagonistic interaction exp

100% KCyanide pred ion (mg/L) 100% CdCl ct 2 tera 50 e in itiv IC add

synerg istic int eraction

100% HgCl 2 HgCl2:KCN:CdCl2

Figure 6-16 Toxicological interaction between HgCl2: KCN: CdCl2 (MTS)

6.6 General Discussion

The basic approach for the analysis conducted on cytotoxicity data, includes fitting an additivity model to the experimental data (Berenbaum 1985; Bae et al. 2001; Gennings et al. 2005). A number of models have been proposed to predict the toxicity of mixtures to an organism, most of which are based on the concept of additivity (Preston et al. 2000). The models were used in this study to investigate the nature of the toxicity of mixtures (i.e. additive, synergistic or antagonistic). Comparisons were then made between the observed responses and the predicted response under the additivity model. In this study deviation from the behaviour described by both the mathematical model (Equation 6-5) and a concentration addition ratio different to unity, plus or minus the relevant variability (Equation 6-7) was considered to indicate interactions between mixtures.

213 Chapter 6

In this study, all 3 types of responses to the chemical mixtures in both assays were observed (Figure 6-17). Even though, some chemicals, did not give the same nature of interaction with both assays. For example a mixture of NaF: SLS using MTS assay, seemed to interact in a synergistic manner (CA ratio = 0.49) whereas the same mixture using the NRU assay showed additivity (CA ratio = 0.9). Similarly both ternary and binary mixtures of Cr (VI), CuSO4 and phenol were synergistic in nature using MTS assay while showing antagonism with the NRU assay. And mixtures of NaF: phenol showed antagonism with the MTS assay as compared to the NRU assay which showed an additive interaction. The additivity model, tended to apply more to the MTS assay, whereas the NRU assay tended to deviate more from additivity expectation (Section 6.5.1). As chemical interaction is dependent on dose-response and ratio of combination therefore each assay might depict toxicity of chemical interaction differently depending on the toxicity of each chemical when tested individually with the assay. As a general overview, it seemed that for the MTS assay 45% of chemical mixtures tested interacted in an additive manner, 35% had lower than expected effect (antagonism) and 20% had greater than expected effect (synergism). On the other hand, 40% of chemical mixtures interacted in an additive manner, 40% in an antagonistic manner and the remaining 20%

in synergism, as measured with the NRU assay (Figure59H6 6- 17).

30 MTS assay 20 NRU assay % Interaction 10

0 synergism antagonism additivity Type of interaction

Figure 6-17 Interaction nature (%) of the number of chemicals tested for chemical mixtures (MTS and NRU)

214 Chapter 6

The mathematical model used in this study was found by Ribo and Rogers (1990) also to apply to the combined effect of a range of phenolic chemicals on the luminescence based Microtox assay. On the other hand Galli et al., (1994) found that the model did not test the effect of mixtures of pesticides well. There are a number of in vitro studies on combinations of chemicals that have shown that the general application of the additivity rule in risk assessment for exposure to a mixture of chemicals is unjustifiable (Marinovich et al. 1994; Marinovich et al. 1996). One draw back of the in vitro cytotoxicity assays used in this test is their inability to model toxicokinetic interactions between test compounds that might occur at higher physiological level.

Most supporting evidence for concentration addition concept comes from multicomponent mixture toxicity studies with different groups of toxicants and different types of aquatic organisms (Altenburger et al. 2000; Backhaus et al. 2000b; Faust et al. 2001; Faust et al. 2003). Several studies in the literature have tested the concentration addition approach for the calculation of mixture effects. Silva et al., (2002) found excellent agreement between prediction and observation in their study of 8 weak estrogenic chemicals using a recombinant yeast and estrogen screen. Multicomponent mixtures of 4 and 12 estrogenic chemicals also closely followed the responses predicted by concentration addition (Payne et al. 2001; Rajapakse et al. 2001). Given though that all chemicals tested in the mixture were estrogenic, (i.e., likely to be similarly acting), it was not surprising that concentration addition was the better predictor. Faust et al., (2003) found that concentration addition tended to over-estimate toxicity for dissimilarly acting substances when compared with another competing concept of Independent Action. But from a regulatory perspective, such overestimation can be defendable based on the precautionary principle.

In a study of toxicological interactions among 4 metals, combined at different ratios, and tested on human epidermal keratinocytes cells using the MTT assay, it was found that the interaction was dose-dependent. Such that, a synergistic interaction turned to an antagonistic effect at higher mixture exposure concentrations (Bae et al. 2001). It appeared that the cellular defence mechanisms of the cells were enhanced at the higher concentration. They generalised that the additivity concept holds true for a narrow dose range.

215 Chapter 6

Metal mixtures such as CdCl2:HgCl2 tested in this study using both MTS and NRU assay showed an antagonistic response. There are many types of metal-metal interactions that could be responsible for the antagonism observed; such as the activation by one of the metals in the mixture of detoxifying or metabolizing pathways. For example, the over-expression of members of the metallothionein (MT) gene family may be involved in protection against metal-induced toxicity (Bae et al. 2001). It has been shown by some studies that merely by alteration in expression of levels of detoxifying molecules such as MT, one metal may have substantial impact on the resulting toxicity of another when they are present together in a mixture (Bae et al. 2001). However in this study an exposure of 4 hours might not have been enough to cause an MT induction.

6.7 Summary of Findings

There is an immense number of potential chemical mixtures that a person can be exposed to, not forgetting the number of new chemicals and mixtures for which no data exists (Marinovich et al. 1996). Testing even the most potential mixtures that a person can be exposed to with classical toxicological protocols is unfeasible. Most Federal Agencies and international organisations such as ATSDR, US EPA, NIOSH use a default assumption of dose or response additivity for the assessment of mixture toxicity which human populations are exposed to (EPA 2000; ATSDR 2004). However this assumption does not allow for the factoring of chemical interactions into the toxicity assessments. Numerous approaches have been developed for the assessment of mixture toxicity, which could differ in theory, application, assumptions and concept (Pounds et al. 2004). There is also a lack in direct comparison between the available approaches, as most tend to differ in the number of data, kind of data, nature of the observations as reported in the literature (Pounds et al. 2004). And conclusions drawn are partially dependent upon the method of analysis.

The conclusions from this data show there are possible toxicological interactions (i.e. departures from additivity) that have clear implications for risk assessment. Results, in

216 Chapter 6

this study, highlight the limitations of assuming an additive interaction and the need to focus on chemical mixture studies. Whereas, for risk assessment purposes additivity can be considered a precautionary principle for chemicals exerting an antagonistic effects, but it would not cover for e.g. the 20% synergistic toxic interactions observed with certain chemical mixtures in this study.

In summary, the in vitro cytoxicity studies conducted are relevant and important to risk assessment of chemical mixtures in several ways. The results show that chemicals in a mixture do not necessarily act in an additive fashion and the possible inclusion of cytotoxicity assays can help in the regulatory decision making process. Assays such as MTS and NRU can help in harmonizing approaches to testing of chemical mixtures and increase consistency in data uses and interpretation. An advantage of using cytotoxicity assays is their ability to screen large number of chemicals and their interactions. These methods have been shown to be fast, inexpensive, reproducible and sensitive.

A harmonisation of methods and concepts is needed in the study of chemical mixtures. The nature of interaction between chemicals is also found to be dose-dependent (Bae et al. 2001). More development is needed for describing and testing additivity and departures from additivity for chemicals combined at their no observable toxic levels and low doses (Teuschler et al. 2002). Research should be continued in developing a battery of in vitro tests, measuring different endpoints to study the combinatory effects of mixtures, and the application of toxicokinetic factors. Collaboration of different disciplines: experimental toxicologists, biomathematicians, biologists, bioengineers and pharmacologists are needed in this field of mixture toxicology.

217 Chapter 7

Chapter 7. Conclusion

Today’s modern society is becoming increasingly more dependent on the use of chemicals (e.g. industrial products, pharmaceuticals, cosmetics, pesticides and household products). Potential hazards to health arise from exposure to the several hundred new synthetic chemicals introduced each year and the myriad of approximately four million mixtures, formulations and blends that are already in commercial use. Currently the toxic potential of chemicals and products is often assessed using standard animal models, which comprise the basic tests for risk assessments. However, since the 1970’s these tests have been widely criticised for scientific, animal welfare and regulatory reasons (Section 2.2.1.1). Due to the biotechnological advances of the last 20 years new options are becoming possible, and alternatives to animal testing such as in vitro toxicity methods are available. The outcome of this research will have a benefit in aiding the classification process of hazardous chemicals and setting standards and exposure limits for the chemical industry, scientific community and the consumer.

During method refinement and development in this thesis, a number of studies have been performed that could potentially refine, and ultimately replace the use of experimental animals for toxicity testing and hazardous classification of chemicals. These included: in vitro cytotoxicity testing using a battery of assays for twenty one chemicals using human cells and the application of in vitro methods to test for possible toxicological interactions among selected binary (15) and ternary (5) chemical mixtures. These methods enabled the generation of data that could be used for a comprehensive risk evaluation process for assessing the safety of single chemicals, chemical mixtures and possible implications to the conventional risk assessment process.

The validation of in vitro assays is necessary to demonstrate the relevance, reliability and predictability of new methodology, before gaining acceptance and usage as replacement and/or complimentary for traditional in vivo methods. In order to do that, a wide variety of chemicals and mixtures with different mechanisms of action were selected and whose toxicity was spread across the GHS categories (Section 4.1.1). The

218 Chapter 7

GHS is a joint international effort aimed at providing the infrastructure for a globalised, consistent approach for the classification of chemicals. Cytotoxicity was tested using a battery of in vitro tests comprising different endpoints (Section 4.2). The toxicity assessment of chemicals and mixtures involved data generation of human cell toxicity

(measured as NOEC, IC50 and TLC). The predictive accuracy of the experimental in vitro toxicity data for human acute toxicity was evaluated using different types of comparative data.

A battery of in vitro cytotoxicity tests including the MTS, NRU, LDH and ATP assays were used to assess the cytotoxicity of twenty one chemicals on human skin fibroblast cultures. Most chemicals tested showed IC50 values with statistical differences among the assays highlighting the importance of using a range of cytotoxicity indicators. The most toxic chemical tested was found to be mercuric chloride (GHS category 1) and the least toxic was glycerol (GHS unclassified). A good correlation was established among the assays, with the strongest between the NRU and ATP assay followed with the NRU and LDH assays. When PCA modelling was performed, the MTS assay was found to have a lower correlation with the other assays, and in general was less sensitive to chemical cytotoxicity. The MTS assay is a measure of the ability of the mitochondria to convert a tetrazolium salt to a coloured formazan product in viable cells, where the mitochondria may be less susceptible than the cells that contain them. The mitochondria may continue to reduce the tetrazolium salt even though injury has already occurred in the cell.

Chemicals that were present in the compounds as cationic electrophiles (e.g. Cu2+, Hg2+ and Co2+) appeared to form a covalent bond with the nicotinamide present in the assay substrate mixture of the LDH assay and subsequently deactivating the second step (diaphorase reaction) required for the conversion of a tetrazolium salt into a coloured formazan product. In those cases, data generated values (NOEC, IC50 and TLC) were taken from dose response curves of intracellular LDH released after cell lysis, and therefore medium containing the chemicals had already been removed. A comparison of the reproducibility of the assays found the ATP assay to have a slightly greater reproducibility than the LDH assay followed by the NRU and MTS assays. The ATP assay was also found to be the most sensitive in detecting the lowest cell number. The

219 Chapter 7

NRU assay provided a more sensitive result in expressing the cytotoxicity of chemicals especially chemicals whose in vivo classification according to the GHS placed them in category 4, indicating that damage to the lysosomes might be present as an early toxic marker.

The accuracy of the cytotoxicity assays was determined by predicting the performance of the assays against data from reference methods (e.g. rodent LD50). A good correlation between IC50 against rodent LD50 values and published human LDL0 values was achieved when an appropriate adjustment factor was implemented to account for differences between in vitro and in vivo systems. In general most chemicals having neurotoxic effects, that act on specific receptors or organs or following bioactivation did not express basal cytotoxicity (e.g. potassium cyanide, sodium selenate, colchicine and strychnine). The most positive significant correlation was with the NRU assay, while the weakest but significant correlation was with the LDH assay. A good correlation between IC50 values and human lethal plasma concentration was achieved, and was similar to correlation results between rodent LD50 values (in the form of HETC) and human lethal plasma concentrations. This indicated the ability of in vitro assays to predict human toxicity at least as accurately as data derived from rodent in vivo studies. In general, a better similarity in terms of the specificity of the methods was obtained when IC50 values from the assays were correlated with published data from main stream in vitro toxicity assays, using human cell lines. NOEC values were also extrapolated from the dose response curves, and a good correlation was obtained against published TLV. The results from this study indicated the potential use of in vitro assays for the setting of permissible uses and exposure limits.

Two in vitro cytotoxicity assays including the MTS and NRU assays were applied to the study of chemical mixtures. Binary and ternary mixtures composed of chemicals spread across the GHS categories of acute oral toxicity were tested at fixed mixture ratios. To ensure that no chemical contributed disproportionately to the overall combination effect, the mixtures were prepared at a mixture ratio proportional to the potency (IC50) of each individual component. The performance of two approaches (mathematical model and concentration addition) for the prediction of mixture effects assuming additive combination effects were tested. These approaches are commonly used by international

220 Chapter 7

agencies for assessing risks and classifying chemical mixtures based on an estimation of additive interaction among the individual components in a mixture (ATSDR 2004; US EPA 2000). The predictions were tested experimentally and agreement between prediction and observation was assessed statistically. Multidimensional modelling of ternary mixtures was developed to graphically represent areas of toxicological interactions. Experimental testing of the compounds revealed that whereas both the concentration addition and the mathematical model were valid methods for the calculation of additive mixture effects, a number of chemicals deviated from the expected additive interaction of their individual mixture components. All three types of effects (antagonism, additivity and synergism) were observed with both assays. For example the ternary mixture of NaF: SLS: irgasan was found to interact in a synergistic manner when measured with both the MTS and NRU assays. The conclusions from this study show that there are possible toxicological interactions (i.e. departures from additivity) that have clear implications for the risk assessment process. Moreover in vitro assays can aid in harmonising approaches to testing of chemical mixtures and increase consistency in data uses and interpretation.

Implementation of the GHS is currently progressing and any validation effort for in vitro tests for acute toxicity should evaluate the predictive performance of the methods for correctly estimating the acute toxicity hazard categories according to the harmonised classification scheme. The NRU assay was able to reliably distinguish between chemicals in adjacent classifications, predicting all of the GHS categories for classification and labelling of acute oral toxicity of chemicals and mixtures. The MTS and LDH assays did not classify chemicals in category 5 of the GHS and the ATP assay did not classify chemicals in both category 3 and category 5 of the GHS.

Future directions

Basal cytotoxicity studies have shown that most chemicals disrupt basic functions common to all cells however a number of questions still remain regarding the molecular toxicity of chemicals and mixtures. This can be explored by a number of means. For example, greater understanding of toxicity mechanisms has expanded the development of new tests that measure new mechanistically based endpoints such as apoptosis.

221 Chapter 7

As simple basal cytotoxicity assays cannot predict the complex interplay between cellular toxicity and chemical permeation through tissue barriers and elimination from the body, focus should be on the incorporation of these assays into a multi-step testing strategy, where other endpoints measuring target organ toxicity, and the relevant toxicokinetic parameters such as ADME are incorporated. In addition, the evaluation of the bioavailability of compounds by using in vitro tests for drug solubility and permeability should be taken into account. This data could then be used to predict the rat LD50 values or if possible, lethal concentrations in humans.

Currently there is a wide availability of databases containing in vivo data from different sources and endpoints. It would be useful if all available data could be integrated into a new extensive database on acute toxicity which could be made available to the scientific and regulatory community. This could help in analysing the outcomes of in vitro tests for validation studies worldwide. Hence, regulatory bodies in the international community such as in Australia can gain access to in vitro data thereby making the validation process of in vitro tests and ultimately the replacement process much quicker.

Numerous approaches have been developed for the assessment of mixture toxicity but many differ in theory, application and concepts. There is a need to be able to compare between available approaches and a harmonisation of methods and concepts for the study of chemical mixtures. More development is needed for testing chemical mixtures at realistic environmental exposures such as combining individual mixture components at their NOAEL and at low doses (IC10). Risk assessments need to take into account potential departures from additivity especially in cases where synergism is observed among the components in a mixture. It is proposed that a battery of in vitro tests can be used as a fast, inexpensive, reproducible and sensitive method to test for possible toxicological interactions. In vitro assays have the potential to screen large numbers of chemicals, and in cases of synergism, further testing is done before chemical mixtures are classified into appropriate hazard categories.

The outcome of this thesis and data generated will enable a more complete risk evaluation of single chemicals and binary and ternary mixtures which can be used to

222 Chapter 7

create more scientifically accurate and reliable safety levels for chemicals and products. It could also help in the availability and potential validation of in vitro methods that would comply with recent legislation calling for a ban of all animal testing in Europe by 2013, and provide an alternative to animal testing of the 30,000 chemicals currently required by the REACH system. It is expected that the studies and validation procedures of selected in vitro assays carried out in this research, even if on a small scale, will be a step forward to current and future validation efforts of alternative methods by industry and research organisations to the overall replacement of animal testing.

223 Bibliography

Bibliography

Abbott, A. (2005). "More Than a Cosmetic Change." Nature 438: 144-147.

Allen, M. and Rushton N. (1994). "Use of the Cytotox 96(Tm) Assay in Routine Biocompatibility Testing in Vitro." Promega notes magazine 45: 7-11.

Altenburger, R., Backhaus T., Boedeker W., Faut M., Scholze M. and Grimme L. (2000). "Predictability of the Toxicity of Multiple Chemical Mixtures to Vibrio Fischeri: Mixtures Composed of Similarily Acting Chemicals." Environmental Toxicology and Chemistry 19 : 2341-2347.

Ames, B. N., McCann J. and Yamasaki E. (1973). "Methods of Detecting Carcinogens and Mutagens with Salmonella/Mammalian-Microsome Mutagenicity Test." Mutation Research 31: 347-364.

Arechabala, B., Coiffard C., Rivalland P., Coiffard J. M. and de Roeck-Holtzhauer Y. (1999). "Comparison of Cytotoxicity of Various Surfactants Tested on Normal Human Fibroblast Cultures Using the Neutral Red Test, MTT Assay and LDH Release." Journal of Applied Toxicology. 19: 163-165.

Argese, E., Maromini, A., Miana, P., Bettiol, C. and Perin, G. (1994). "Submitochondrial Particle Response to Linear Alkylbenzene Sulfonates, nonylphenol polyeothoxylates and their Biodegradation Derivatives" Environmental Toxicology and Chemistry 13: 737-742.

ATSDR (2002). Agency for Toxic Substances and Disease Registry. Department of Health and Human Services. Available from: http://www.atsdr.cdc.gov/

ATSDR (2004). Guidance Manual for the Assessment of Joint Toxic Action of Chemical Mixtures., Agency for toxic substances and disease registry.

224 Bibliography

Atterwill, C. K. (1995). Alternative Method of Assessing Toxicity. In Vitro Toxicology Testing Protocols. S. O'Hare and C. K. Atterwill, Humana Press, New Jersey. 43.

Baba, C., Yanagida K., Kanzaki T. and Baba M. (2005). "Colorimetric Lactate Dehydrogenase Assay (Ldh) for Evaluation of Antiviral Activity against Bovine Viral Diarrhoea Virus (BVDV) in Vitro." Antiviral Chemistry and Chemotherapy. 16: 33-39.

Babich, H. and Babich J. P. (1997). "Sodium Lauryl Sulfate and Triclosan: In Vitro Cytotoxicity Studies with Gingival Cells." Toxicology Letters 91: 189-196.

Backhaus, T., Altenburger R., Boedeker W., Faust M., Scholze M. and Grimme L. H. (2000a). "Predictability of the Toxicity of a Multiple Mixture of Dissimlarly Acting Chemicals to Vibrio Fischeri." Environmental Toxicology and Chemistry 19(9): 2348- 2356.

Backhaus, T., Scholze M. and Grimme L. H. (2000b). "The Single Substance and Mixture Toxicity of Quinolones to the Bioluminiscent Bacteria Vibrio Fischeri." Aquatic Toxicology 49:49-61.

Bae, D. S., Gennings C., Carter W. H., Yang R. S. H. and Campain J. A. (2001). "Toxicological Interactions among Arsenic, Cadmium, Chromium, and Lead in Human Keratinocytes." Toxicological Sciences. 63: 132-142.

Balls, M. and Fentem J. H. (1992). "The Use of Basal Cytotoxicity and Target Organ Toxicity Tests in Hazard Identification and Risk Assessment." Alternatives to Laboratory Animals 20: 368-388.

Balls, M. and Fentem J. H. (1999). "The Validation and Acceptance of Alternatives to Animal Testing." Toxicology In Vitro 13: 837-846.

Balls, M. and Karcher W. (1995). "The Validation of Alternative Test Methods." Alternatives to Laboratory Animals 23: 884-886.

225 Bibliography

Barile, F. A. (1994). Introduction to in Vitro Cytotoxicology, Mechanisms and Methods, CRC press, Boca Raton.

Barile, F. A. (1997). Continuous Cell Lines as a Model for Drug Toxicology Assessment. In Vitro Methods in Pharmaceutical Research. J. Castell and M. Gomez- Lechon. San Diego, Academic press: 33-54.

Barile, F. A. and Cardona D. L. (1998). "Acute Cytotoxicity Testing with Cultured Human and Lung Dermal Cells." In Vitro Cellular and Developmental Biology. 34(8): 631-635.

Barile, F. J., Dierickx P. J. and Kristen U. (1994). "In Vitro Cytotoxicity Testing for Prediction of Acute Human Toxicity." Cell Biology and Toxicology 10: 155-162.

Barlow, S. M., Greig J. B., Bridges J. W., Carered A., Carpye A. J. M., Gallif C. L., Kleinerg J., Knudsenh I., Koeteri H. B. W. M., Levya L. S., Madseni C., Mayerj S., Narbonnek J. F., Pfannkuchl F., Prodanchuk M. G., Smith M. R. and Steinbergo P. (2002). "Hazard Identification by Methods of Animal-Based Toxicology." Food and Chemical Toxicology 40 (2-3): 143-191.

Barrouillet, M. P., Ohayon-Courtes C., Dubus I., L'azou B. and Nguyen Ba C. (2001). "Influence of Cadmium Speciation for the Evaluation of in Vitro Cadmium Toxicity on Llc-Pk1 Cells." Toxicology In Vitro 15: 525-529.

Baselt, R. C. and Cravey R. H. (1989). Disposition of Toxic Drugs and Chemicals in Man. Chicago, Yearbook Medical Publishers, Inc.

Berenbaum, M. C. (1985). "The Expected Effect of a Combination of Agents: The General Solution." Journal of Theoretical Biology 114: 413-431.

Berenbaum, M. C. (1989). "What Is Synergy?" Pharmacological Reviews 41: 93-141.

226 Bibliography

Blaauboer, B. J. (1996). "In Vitro and Other Animal Experiments in the Biomedical Sciences." ANZCCART News 9(4).

Blaauboer, B. J. (2001). "Toxicodynamic Modelling and the Interpretation of in Vitro Toxicity Data." Toxicology Letters 120: 111-123.

Blaauboer, B. J. (2002a). "The Necessity of Biokinetic Information in the Interpretation of in Vitro Toxicity Data." Alternatives to Laboratory Animals 30(2): 85-91.

Blaauboer, B. J. (2002b). "The Applicability of in Vitro-Derived Data in Hazard Identification and Characterisation of Chemicals." Environmental Toxicology and Pharmacology. 11: 213-225.

Blaauboer, B. J., Balls M., Barratt M., Casati S., Coecke S., Mohamed M. K., Moore J., Rall D., Smith K. R., Tennant R., Schwetz B. A., Stokes W. S. and Younes M. (1998). "13th Meeting of the Scientific Group on Methodologies for the Safety Evaluation of Chemicals (SGOMSEC): Alternative Testing Methodologies and Conceptual Issues." Environmental Health Perspectives. 106(2): 413-418.

Bondesson, I., Ekwall B., Hellberg S., Romert L., Stenberg K. and Walum E. (1989). "MEIC- a New International Multicentre Project to Evaluate the Relevance of Human Toxicity of in Vitro Cytotoxicity Tests." Cell Biology and Toxicology 5: 331-347.

Borenfreund, E. and Babich H. (1992). The Neutral Red Cytotoxicity Assay. INVITOX, ECVAM.

Borenfreund, E. and Puerner J. A. (1985). "Toxicity Determined in Vitro by Morphological Alterations and Neutral Red Absorption." Toxicology Letters 24: 119- 124.

Borgert, C. J., Price B., Wells C. S. and Simon G. S. (2001). "Evaluating Chemical Interaction Studies for Mixture Risk Assessment." Human and Ecological Risk Assessment 7(2): 259-306.

227 Bibliography

Botham, C., Holmes P., Levy L. S. and Harrison P. (2001). Prosepects for the Use of Alternative Test Methods. Leicester, Medical Research Council: Institute for Environment and Health.

Brickman, R., Jasanoff S. and Ilgen T. (1985). Controlling Chemicals: The Politics of Regulation in Europe and the United States. Ithaca, Cornell University Press.

BUAV (2001). Action to End Animal Testing: The Way Forward. London, UK, British Union for the Abolition of Vivisection: 19.

Budavari, S. (2000). Merck's Index.

Cairns, J. and Pratt J. R. (1990). The Scientific Basis of Bioassays. Environmental Bioassay Techniques and Their Application. M. Munawar, G. Dixon, C. Mayfield, T. B. Reynoldson and M. H. Sadar. Dodrecht, Kluwer Academic Publishers.

Calabrese, E. J. (2002). "Hormesis: Changing Views of the Dose Response." Mutation Research 511: 181-189.

Calabrese, E. J. (2005a). "Hormesis: How It Could Affect the Risk Assessment Process." Human and Experimental Toxicology 24: 265-270.

Calabrese, E. J. (2005b). "Toxicological Awakenings: The Rebirth of Hormesis as a Central Pillar of Toxicology." Toxicology and Applied Pharmacology 204: 1-8.

Calabrese, E. J. and Baldwin L. (2003). "Toxicology Rethinks Its Central Belief: Hormesis Demands a Reappraisal of the Ways Risks Are Assessed." Nature 421: 691- 692.

Calabrese, E. J., Baldwin L. and Holland C. D. (1999). "Hormesis: A Highly Generalizable and Reproducible Phenomenon with Important Implications for Risk Assessment." Risk Anal. 19: 261-281.

228 Bibliography

Carpy, S. A., Kobel W. and Doe J. (2000). "Health Risk of Low Dose Pesticide Mixtures: A Review of the 1985-1998 Literature on Combination Toxicology and Health Risk Assessment." Journal of toxicology and environmental health 3: 1-25.

Carson, B. L., Ellis H. V. and mCcann J. L. (1986). Toxicology and Biological Monitoring of Metals in Humans. Michigan, Lewis Publishers Inc.

Cassee, F. R., Groten J. P., van Bladeren P. J. and Feron V. J. (1998). "Toxicological Evaluation and Risk Assessment of Chemical Mixtures." Critical Reviews in Toxicology. 28(1): 73-101.

Cassee, F. R., Stenhuis W. H., Groten J. P. and Feron V. J. (1996). "Toxicity of Formaldehyde and Acrolein Mixtures: In Vitro Studies Using Nasal Epithelial Cells."

Castell, J. and Gomez-Lechon M. (1997). In Vitro Methods in Pharmaceutical Research. Spain, Academic press.

Chen, J. J., Chen Y. J., Rice G., Teuschler L., Hamernik K., Protzel A. and Kodell R. (2001). "Using Dose Addition to Estimate Cumulative Risks from Exposures to Multiple Chemicals." Regulatory Toxicology and Pharmacology. 34: 35-41.

Clayton, G. D. and Clayton F. E. (1994). Patty's Industrial Hygiene and Toxicology. NY, John Wiley and Sons.

Clemedson, C., Dierickx P. J. and Sjostrom M. (2003). "The Prediction of Human Acute Systemic Toxicity by the Edit/Meic in Vitro Test Battery: The Importance of Protein Binding and of Partitioning into Lipids." Alternatives to Laboratory Animals 31: 245- 256.

Clemedson, C. and Ekwall B. (1999). "Overview of the Final MEIC Results: I. The in Vitro-in Vivo Evaluation." Toxicology In Vitro 13: 657-663.

229 Bibliography

Clemedson, C., Nordin-Andersson M., Bjerregaard H. F., Clausen J., Forsby A., Gustafsson H., Hansson U., Isomaa B., Jorgensen C., Kolman A., Kotova N., Krause G., Kristen U., Kurppa K., Romert L. and Scheers E. (2002). "Development of an in Vitro Test Battery for the Estimation of Acute Human Systemic Toxicity: An Outline of the EDIT Project." Alternatives to Laboratory Animals 30: 313-321.

Cleuvers, M. (2004). "Mixture Toxicity of the Anti-Inflammatory Drugs Diclofenac, Ibuprofen, Naproxen, and Acetylsalicylic Acid." Ecotoxicology and Environmental Safety 59: 309-315.

Combes, R., Barratt M. and Balls M. (2003). "An Overall Strategy for the Testing of Chemicals for Human Hazard and Risk Assessment under the EU Reach System." Alternatives to Laboratory Animals 31: 7-19.

Cote, R., Davis H., Dimock C., Korpan M., Loewen K. and Segal L. M. (1998). "The Evaluation and Hazard Classification of Toxicological Information for Workplace Hazardous Materials Information System Material Safety Data Sheets." Regulatory Toxicology and Pharmacology. 27: 61-74.

Crishton, S. (2002). High-Flier Says Handyman Filler Destroyed His Life. Sydney Morning Herald. National News Friday April 26.

Cross, D. P., Ramachandran G. and Wattenberg E. V. (2001). "Mixtures of Nickel and Cobalt Chlorides Induce Synergistic Cytotoxic Effects: Implications for Inhalation Exposure Modeling." Annal of Occupational Hygiene 45(5): 409-418.

Crump, K. (1995). "Calculation of the Benchmark Doses from Continuous Data." Risk Anal. 15: 79-89.

Da Costa, A. O., De Assis M. C., De Andrade Marques E. and Plotkowski M. C. (1999). "Comparative Analysis of Three Methods to Assess Viability of Mammalian Cells in Culture." BioCell 23(1): 65-72.

230 Bibliography

Dalzell, D. J. B., Alte S., Aspichueta E., de la Sota A., Etxebarria J., Gutierrez M., Hoffmann C. C., Sales D., Obst U. and Christofi N. (2002). "A Comparison of Five Rapid Direct Toxicity Assessment Methods to Determine Toxicity of Pollutants to Activated Sludge." Chemosphere 47: 535-545.

Daston, G. P. and McNamee P. (2005). "Alternatives to Toxicity Testing in Animals: Challenges and Opportunities." Environmental Health Perspectives.: 113: 6-15.

De Rosa, C. T., Cibulas W., El-Masri H. E., Pohl H., Tylenda C. and Mumtaz M. (2001). Implications of Chemical Mixtures in Public Health Practice. Application of technology to chemical mixtures research, Colorado State University, Fort Collins, National Institute of Environmental Health Sciences.

Dierickx, P. J. (2000). "Cytotoxicity of the MEIC Reference Chemicals in Rat Hepatoma-Derived Fa32 Cells." Toxicology 150: 159-169.

Dierickx, P. J. (2003). "Evidence for Delayed Cytotoxicity Effects Following Exposure of Rat Hepatoma-Derived Fa32 Cells: Implications for Predicting Human Acute Toxicity." Toxicology In Vitro 17: 797-801.

Dierickx, P. J. and Scheers E. M. (2002). "Neutral Red Uptake Inhibition in Adhered and Adhering Rat Hepatoma-Derived Fa32 Cells to Predict Human Toxicity." Journal of Applied Toxicology. 22: 61-65.

Eaton, D. and Klaassen C. (2001). Principles of Toxicology. Casarett's and Doull's Toxicology: The Basic Science of Poisons. C. Klaassen. NY, McGraw Hill.

ECVAM (2002). "Alternative (Non-Animal) Methods for Chemicals Testing: Current Status and Future Prospects." Alternatives to Laboratory Animals 30(1): 1-125.

Eirheim, E. H., Bundgaard C. and Nielsen M. H. (2004). "Evaluation of Different Toxicity Assays Applied to Proliferating Cells and to Stratified Epithelium in Relation to Permeability Enhancement with Glycocholate." Toxicology In Vitro 18: 649-657.

231 Bibliography

Eisenbrand, G., Pool-Zobel B., Baker V., Balls M., B.J. B., Boobis A., Carere A., Kevekordes S., Lhuguenot J. C., Pieters R. and Kleiner J. (2002). "Methods of in Vitro Toxicology." Food and Chemical Toxicology 40: 193-236.

Ekwall, B. (1983). "Screening of Toxic Compounds in Mammalian Cell Cultures, in Cellular Systems for Toxicity Testing." Ann. New York Acad. Science 407: 64-77.

Ekwall, B. (1995). The Basal Cytotoxicity Concept. The world conference on alternatives and animal use in life sciences: education, research, testing., Mary Ann Liebert.

Ekwall, B. (1998). "MEIC Evaluation of Acute Systemic Toxicity. Part VI. The

Prediction of Human Toxicity by Rodent LD50 Values and Results from 61 in Vitro Methods." Alternatives to Laboratory Animals 26: 617-658.

Ekwall, B. (1999a). "Overview of the Final MEIC Results: II. The in Vitro--in Vivo Evaluation, Including the Selection of a Practical Battery of Cell Tests for Prediction of Acute Lethal Blood Concentrations in Humans." Toxicology In Vitro 13(4-5): 665-673.

Ekwall, B., Bondesson I., Castell J. V., Gomez-Lechon M., Hellberg S., Hogberg J., Jover R., Ponsoda X., Romert L., StenbErg K. and Walum M. (1989). "Cytotoxicity Evaluation of the First 10 MEIC Chemicals: Acute Lethal Toxicity in Man Predicted by

Cytotoxicity in 5 Cellular Assays and by Oral LD50 Tests in Rodent." Alternatives to Laboratory Animals 17: 83-100.

Ekwall, B., Bondesson I., Hellberg S., Hogberg J., Romert L., Stenberg K. and Walum E. (1991). "Validation of in Vitro Cytotoxicity Tests: Past and Present Strategies." Alternatives to Laboratory Animals 19(2): 226-233.

Ekwall, B., Clemedson C., Ekwall B., Ring P. and Romert L. (1999b). "EDIT: A New International Multicentre Programme to Develop and Evaluate Batteries of in Vitro Tests for Acute and Chronic Systemic Toxicity." Alternatives to Laboratory Animals 27: 339-349.

232 Bibliography

Ekwall, B., Ekwall B. and Sjostrom M. (2000). "MEIC Evaluation of Acute Systemic Toxicity. Part VIII. Multivariate Partial Least Squares Evaluation, Including the Selection of a Battery of Cell Line Tests with a Good Prediction of Human Acute Lethal Peak Blood Concentrations for 50 Chemicals." Alternatives to Laboratory Animals 28: 204-234.

EPA (1986). Guidelines for the Heal Risk Assessment of Chemical Mixtures. Washington, US environmental Protection Agency.

EPA (1992). Guidelines for Exposure Assessment. Washington DC, US environmental protection agency: 139.

EPA (1999). Proposed Guidance on Cumulative Risk Assessment of Pesticide Chemicals That Have a Common Mechanism of Toxicity. Washington, D.C., US environmental protection agency.

EPA (2000). Supplementary Guidance for Conducting Health Risk Assessment of Chemical Mixtures. Washington D.C, United States Environmental Protection Agency.

EPA (1984). Health Effects Assessment for Polycyclic Aromatic Hydrocarbons (PAH). Cincinnati, US Environmental Protection Agency.

EU directive 2003/15/EC of the European Parliament and the Council of 27 February 2003. "Amending Council Directive 76/768/EEC on the Approximation of the Laws of the Member States Relating to Cosmetic Products". Official Journal of the European Union. L 66 26-35 pages.

EVM (2003). Review of Cobalt. UK, Expert group on vitamins and animals.

Faust, M., Altenburger R, Backhaus T., Blanck H., Boedeker W., Gramatica P., Hamer V., Scholze M., Vighi M. and Grimme L. H. (2001). "Predicting the Joint Algal Toxicity of Multi-Component S-Triazine Mixtures at Low Effect Concentrations of Individual Toxicants." Aquatic Toxicology 56: 13-32.

233 Bibliography

Faust, M., Altenburger R., Backhaus T., Blanck H., Boedeker W., Gramatica P., Hamer V., Scholze M., Vighi M. and Grimme L. H. (2003). "Joint Algal Toxicity of 16 Dissimilarly Acting Chemicals Is Predictable by the Concept of Independent Action." Aquatic Toxicology 63: 43-63.

Faustman, E. M. and Omenn G. S. (2001). Risk Assessment. Casarett and Doull's Toxicology. C. Klaassen. New York, McGraw-Hill: 83-104.

Feron, V. J., Cassee F. R. and Groten J. P. (1998). "Toxicology of Chemical Mixtures: International Perspective." Environmental Health Perspectives. 106(6): 1281-1289.

Feron, V. J., van Bladeren P. J. and Hermus R. J. J. (1990). "A View Point on the Extrapolation of Toxicological Data from Animals to Man." Food and Chemical Toxicology 28: 783-788.

Frazier, J. M. (1992). In Vitro Toxicity Testing. New York, Marcel Dekker, Inc.

Frazier, J. M. and Goldberg A. M. (1990). "Alternatives to and Reduction of Animal Use in Biomedical Research, Education and Testing." The Cancer Bulletin. 42: 238-245.

Gad, S. C. (2000). In Vitro Toxicology. London, Taylor & Francis.

Galli, C. L., Viviani B. and Marinovich M. (1993). "Cell Cultures: A Tool for the Study of Mechanisms of Toxicity." Toxicology In Vitro 7: 559-568.

Garle, M. J., Fentem J. H. and Fry J. (1994). "In Vitro Cytotoxicity Tests for the Prediction of Acute Toxicity in Vivo." Toxicology In Vitro 8(6): 1303-1312.

Garthoff, B. (2004). Alternatives to Animal Experimentation: The Regulatory Background. Symposium S14. 10th International congress on toxicology (ICT-X), Tampere, Finland. Toxicology and Applied Pharmacology (Abstract number 69).

234 Bibliography

Gaylor, D. (1998). "Safety Assessment with Hormetic Effect." Human and Experimental Toxicology 17(5): 251-253.

Gennari, A., Cortese E., Boveri M., Casado J. and Prieto P. (2003). "Sensitive Endpoints for Evaluating Cadmium-Induced Acute Toxicity in Llc-Pk1 Cells." Toxicology 183: 211-220.

Gennari, A., van de Berghe C., Casati S., Castell J., Clemedon C., Coecke S., Colombo A., Curren R., Dal Negro G., Goldberg A. M., Gosmore C. and et al. (2004). "Strategies to Replace in Vivo Acute Systemic Toxicity Testing." Alternatives to Laboratory Animals 32: 437-459.

Gennings, C., Carter W. H., Carchman R. A., Teuschler L., Simmons J. E. and Carney E. (2005). "A Unifying Concept for Assessing Toxicological Interactions: Changes in Slope." Toxicological Sciences. 88(2): 287-297.

Gettings, S. D., Lardo R. A., Hintze K. L., Bagley D. M., Casterton P. L., Chudkowski M., Curren J. L., Demetrulias L. C., Dipasquale L. C., Earl L. K., Feder C. L., Galli S. M., Glaza S. M., Gordon V. C., Janus P. J., Kurtz K. D., Marenus J., Moral W. J., Pape K. J., Renskers K. J., Rheins L. A., Roddy M. T., Rozen M. G., Tedeschi J. and Zyracki J. (1996). "The CFTA Evaluation of Alternatives Program: An Evaluation of in Vitro Alternatives to the Draize Primary Eye Irritation Test. (Phase III) Surfactant-Based Formulations." Food and Chemical Toxicology 34: 79-117.

Goldstein, B. (2005). "Advances in Risk Assessment and Communication." Annual Review of Public Health 26: 141-163.

Gosselin, R.E., Smith, R.P. and Hodge, H.C. (1984). Clinical Commercial Toxicology of Commercial Products. Williams and Wilkins; Baltimore, USA.

Goyer, R. A. (2001). Toxic Effects of Metals. Casarett's and Doull'stoxicology: The Basic Science of Poisons. C. Klaassen. NY, McGraw-Hill.

235 Bibliography

Greco, W.R, Bravo G. and Parsons JC. (1995). "The Search for Synergy: a Critical Review from a Response Surface Perspective." Pharmacological Reviews 47: 331-385

Greenburg, M. I. and Phillips S. D. (2003). A Brief History of Occupational, Industrial and Environmental Toxicology. Occupational, Industrial and Environmental Toxicology. M. I. Greenberg, R. J. Hamilton and S. D. Phillips. Philadelphia, Mosby: 2- 5.

Gribaldo, L., Gennari A., Blackburn K., Clemedon C., Deguercy A., Meneguz A., Pfaller W. and Ruhdel I. (2005). "Acute Toxicity." Alternatives to Laboratory Animals 33(1): 27-34.

Groten, J. P. (2000). "Mixtures and Interactions." Food and Chemical Toxicology 38: S65-S71.

Groten, J. P., Feron V. J. and Suhnel J. (2001). "Toxicology of Simple and Complex Mixtures." Trends in Pharmacological Sciences. 22(6).

Harbell, J. W., Koontz S. W., Lewis D., Lovell D. and Acosta D. (1997). "IRAG Working Group 4. Cell Cytotoxicity Assays. Interagency Regulatory Alternatives Group." Food and Chemical Toxicology 35: 79-126.

Harris, R. D. and Gillett M. J. (1998). "Colchicine Poisoning - Overview and New Directions." Emergency medicine 10: 161-167.

Hayes, A. and Markovic B. (2000). "Alternatives to Animal Testing for Determining the Safety of Cosmetics." Cosmetics, Aerosols and Toiletries in Australia 12(5): 24-30.

Hayes, A. and Markovic B. (2002). "Toxicity of Australian Essential Oil Backhousia Citriodora (Lemon Myrtle). Part 1. Antimicrobial Activity and in Vitro Cytotoxicity." Food and Chemical Toxicology 40: 535-543.

236 Bibliography

Hodgson, E., Mailma R. B. and Chambers J. E. (1998). Dictionary of Toxicology: 2nd Edition. London, Macmillan Reference LTD.

Holme, J. A. and Dybing E. (2002). "The Use of in Vitro Methods for Hazard Characterisation of Chemicals." Toxicology Letters 127: 135-141.

Howell, G. (1998). Consumer Product Safety. Ashgate, Aldershot.

HSDB (2002). Hazardous Substances Data Bank (HSDB). March 2003 National Library of Medicine's. Toxicology Data Network (ToxNet), National Institutes of Health (NIH), Department of Health and Human Services.

Huggett, A. C., Schilter B., Roberfroid M., Antignac E. and Koeman J. H. (1996). "Comparative Methods of Toxicity Testing." Food and Chemical Toxicology 34(2): 183- 192.

IARC (2004). International Agency for Research on Cancer. Available at: wwww.iarc.fr

ICCVAM (2000). "Regulatory Acceptance of Toxicological Methods." http://iccvam.niehs.nih.gov.

ICCVAM (2001a). Guidance Document on Using in Vitro Data to Estimate in Vivo Starting Doses for Acute Toxicity., interagency coordinating committee on the validation of alternative methods (ICCVAM) and the National Toxicology Program (NTP) Interagency Centre for the Evaluation of Alternative Toxicological Methods (NICEATM).

ICCVAM (2001b). "Report of the International Workshop on in Vitro Methods for Assessing Acute Systemic Toxicity." Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM) and the National Toxicology Program (NTP) Interagency Centre for the Evaluation of Alternative Toxicological Methods (NICEATM).

237 Bibliography

IFCS (2003). Strategic Approach to International Chemicals Management (SAICM). Bangkok, Thailand, Intergovernmental forum on chemical safety: 46.

Ikeda, M. (1995). "Exposure to Complex Mixtures: Implications for Biological Monitoring." Toxicology Letters 77: 85-91.

ILO (1990). Safety in the Use of Chemicals at Work. Convention No 170. Geneva, International Labour Organisation.

INCHEM, I. (2005). Strychnine (Pim 507).

IOMC (1996). Strengthening National Capabilities for the Sound Management of Chemicals. Geneva, Inter-Organization Programme for the Sound Management of Chemicals.

IPCS (1999). Environmental Health Criteria 210: Principles for the Assessment of Risk to Human Health from Exposure to Chemicals. Geneva, World Health Organisation.

Issa, Y., Watts D., C.,, Brunton P. A. and Duxbury A. J. (2004). "Resin Composite Monomers Alter MTT and LDH Activity of Human Gingival Fibroblasts in Vitro." Dental Materials 20: 12-20.

Jirova, D., Kejlova K., Brabec M., Bendova H. and Kolarova H. (2003). "The Benefits of the 3T3 NRU Test in the Safety Assessment of Cosmetics: Long-Term Experience from Pre-Marketing Testing the Czech Republic." Toxicology In Vitro 17: 791-796.

Kalberlah, F., Schneider K. and Schuhmacher-Wolz U. (2003). "Uncertainty in Toxicological Risk Assessment for Non-Carcinogenic Health Effects." Regulatory Toxicology and Pharmacology. 37: 92-104.

Klaassen, C. (2001). Casarett and Doull's Toxicology; the Basic Science of Poisons, McGraw-Hill.

238 Bibliography

Kleinjans, J. (2003). "Principles in Toxicological Risk Analysis." Toxicology Letters 140-141: 311-315.

Konemann, W. H. (1981). "Fish Toxicity Tests with Mixtures of More Than Two Chemicals: A Proposal for a Quantitative Approach and Experimental Results." Toxicology 19: 229-238.

Koëter B.W.M (2002). "The Role of OECD in Achieving International Acceptance of Harmonized Test Guidelines". OECD Environment, Health and Safety Division, Paris, France. available online: www.oecd.org/dataoecd/6/16/2752903.ppt. Accession date: January 2002. 43 pages.

Kortenkamp, A. (2004). Mixture Effects and Their Assessment- Concepts, Experimental Design and Interpretation. Session CE2 in Continuing Education Course. International Congress of Toxicology (ICT-X), Tampere, Finland.

Langley, G. (2004). Chemical Safety and Animal Testing: A Regulatory Smokescreen? London, BUAV/European coalition to end animal experiments (ECEAE).

Langley, G. (2005). Acute Toxicity Testing without Animals "More Scientific and Less of a Gamble". London, European Coalition to End Animal Experiments (ECEAE).

Lewis, R. J. (2000). Sax's Dangerous Properties of Industrial Materials. New York, Wiley and Sons Inc.

Lobner, D. (2000). "Comparison of LDH and MTT Assays for Quantifying Cell Death: Validity for Neuronal Apoptosis?" Journal of Neuroscience Methods. 96: 147-152.

Louekari, K. (2004). "Status and Prospects of in Vitro Tests in Risk Assessment." Alternatives to Laboratory Animals 32(4): 431-435.

239 Bibliography

Lutz, W. K. (2002). "Differences in Individual Susceptibility to Toxic Effects of Chemicals Determine Dose-Response Relationship and Consequences of Setting Exposure Standards." Toxicology Letters 126: 155-158.

Mahmood, I. and Balian J. D. (1999). "The Pharmacokinetic Principles Behind Scaling from Preclinical Results to Phase I Protocols." Clinical Pharmacokinetics 36: 1-11.

Malich, G. (1998). Development of an in Vitro Toxicological Test Method for Contaminated Site Assessments. School of Safety Science. Sydney, The University of New South Wales: 261.

Malich, G., Markovic B. and Winder C. (1998). "Human Cell Line Toxicity of Binary and Ternary Chemical Mixtures in Comparison to Individual Toxic Effects of Their Components." Archives of Environmental Contamination and Toxicology. 35: 370-376. Malich, G., Markovic B. and Winder C. (1997). "The Sensitivity and Specificity of the MTS Tetrazolium Assay for Detecting the in Vitro Cytotoxicity of 20 Chemicals Using Human Cell Lines." Toxicology 124: 179-192.

Marinovich, M., Ghilardi F. and Galli C. L. (1996). "Effect of Pesticide Mixtures on in Vitro Nervous Cells: Comparison with Single Pesticides." Toxicology 108: 201-206.

Marinovich, M., Guizzetti M. and Galli C. L. (1994). "Mixtures of Benomyl, Pirimiphos-Methyl, Dimethoate, Diazinon and Azinphos-Methyl Affect Protein Synthesis in HL-60 Cells Differently." Toxicology 94: 173-185.

Massart, D. L., Vadeginste B. G. M., Buydens L. M. C., De Jong S., Lewi P. J. and Smeyers-Verbeke J. (2003). Data Handling in Science and Technology. Amesterdam, Elsevier.

McKarns, S. C., Hansch C., Caldwell W. S., Morgan W. T., Moore S. K. and Doolittle D. J. (1997). "Correlation between Hydrophobicity of Short-Chained Aliphatic Alcohols and Their Ability to Alter Plasma Membrane Integrity." Fundamental and Applied Toxicology 36: 62-70.

240 Bibliography

Meek, M. E., Renwick A., Ohanian A., Dourson M., Lake B., Naumann B. D. and Vu V. (2002). "Guidelines for the Application of Chemical-Specific Adjustment Factors in Dose/Concentration-Response Assessment." Toxicology 181-182: 115-120.

Merrill, R. A. (2001). Regulatory Toxicology. Casarett and Doull's Toxicology. The Basic Science of Poisons. C. Klaassen. New York, McGraw-Hill: 1141-1153.

Mileson, B., Chambers J., Chen W. L., Dettbarn W., Ehrich M., Eldefrawi A. T., Gaylor D. W., Hamernik K. and Karczmar A. G. (1998). "Common Mechanism of Toxicity: A Case Study of Organophosphorous Pesticides." Toxicological Sciences. 41: 8-20.

Mueller, H., Kassack M. U. and Wiese M. (2004). "Comparison of the Usefulness of the MTT, ATP, and Calcein Assays to Predict the Potency of Cytotoxic Agents in Various Human Cancer Cell Lines." Journal of Biomolecular Screening. 9(6): 506-515.

Mumtaz, M. (1995). "Risk Assessment of Chemical Mixtures from a Public Health Perspective." Toxicology Letters 82-83: 527-532.

Mumtaz, M. and Sipes G. (1993). "Risk Assessment of Chemical Mixtures: Biologic and Toxicologic Issues." Fundamental and Applied Toxicology 21: 258-269.

Mumtaz, M. M., DeRosa C. T. and Durkin P. R. (1994). Approaches and Challenges in Risk Assessment of Chemical Mixtures. Toxicology of Chemical Mixtures., Academic Press Inc.: 565-597.

Niederau, C., Fronhoffs K., Klonowski H. and HU. S. (1995). "Active Pancreatic Digestive Enzymes Show Striking Differences in Their Potential to Damage Isolated Rat Pancreatic Acinar Cells." The Journal of Lab and Clinical Medecine 125: 265-275.

NIH (1999). Evaluation of the Validation Status of Toxicological Methods: General Guidelines for Submission to ICCVAM. Research triangle park, NC, NIH publication 99-4496: 44.

241 Bibliography

NIOSH, U. S. (2002). Registry of Toxic Effects of Chemical Substances (RTECS).

NRC (1988). Science and Judgment in Risk Assessment. Washington DC, National Research Council.

NTP and NICEATM (2000). The Multicentre Evaluation of in Vitro Cytotoxicity (MEIC): Summary.

NTP and NICEATM. (2003). Validation Study to Evaluate the Usefulness of Two Cytotoxicity Methods for Estimating Acute Systemic Toxicity., National Institute of Environmental Health Sciences (NIEHS).,National Institutes of Health (NIH)., U.S. Public Health Service., Department of Health and Human Services.: 1-17.

NTP and NICEATM (2003b). Test Method Protocol for the NHK Neutral Red Uptake Cytotoxicity Test. A Test for Basal Cytotoxicity for an in Vitro Validation Study Phase III: 1-20.

Oakes, D. J. and Pollak J. K. (2000). "The in Vitro Evaluation of the Toxicities of Three Related Herbicide Formulations Containing Ester Derivatives of 2,4,5- T and 2,3- D Using Sub-Mitochondrial Particles." Toxicology 151: 1-9.

OECD (1996). Final Report of the Solna Workshop on Validation and Regulatory Acceptance Criteria for Alternative Tests. OECD report ENV/MC/CHEM/TG(96)9. Organisation for Economic Co-operation and Development, Paris, France.

OECD (2001a). Harmonised Integrated Classification System for Human Health and Environmental Hazards of Chemical Substances and Mixtures. Paris, France, Organisation for Economic Co-operation and Development, Paris, France. OECD Series on Testing and Assessment No 33: 247.

OECD (2001b). Test Guideline 420. Acute Oral Toxicity: Fixed Dose Method. Guidelines for the Testing of Chemicals Paris, France. Section 4: Health Effects (pink pages). Most recent update: 17 December, 2001.

242 Bibliography

OECD (2001c). Test Guideline 423. Acute Oral Toxicity- Acute Toxic Class Method. Guidelines for the Testing of Chemicals Paris, France. Section 4: Health Effects (pink pages). Most recent update: 17 December, 2001.

OECD (2001d). Test Guideline 425- Acute Oral Toxicity: Up- and -Down Procedure. Guidelines for the Testing of Chemicals. Paris, France. Section 4: Health Effects (pink pages). Most recent update: 23 March 2006.

OECD (2001e). Detailed Review Document on Hazard Classification Systems for Mixtures. OECD Series on Testing and Assessment Paris, France. No 30: 178.

OECD (2004a). Test Guideline 428 - Skin Absorption: In Vitro Method. Guidelines for the Testing of Chemicals Paris, France. Adopted: 13 April 2004.

OECD (2004b). Test Guideline 430 - In Vitro Skin Corrosion: Transcutaneous Electrical Resistance Test (TER). Guidelines for the Testing of Chemicals Paris, France. Section 4: Health Effects (pink pages). Adopted: 13 April 2004.

OECD (2004c). Test Guideline 431- In Vitro Skin Corrosion: Human Skin Model Test. Guidelines for the Testing of Chemicals Paris, France. Section 4: Health Effects (pink pages). Adopted: 13 April 2004.

OECD (2004d). Test Guideline 432- In Vitro 3TR NRU Phototoxicity Test. Guideline for the Testing of Chemicals Paris, France. Section 4: Health Effects. Adopted: 13 April 2004.

Omenn, G. S. (1995). "Assessing the Risk Assessment Paradigm." Toxicology 102: 23- 28.

O'Neill, A. J., Miller K., Oliva B. and Chopra I. (2004). "Comparison of Assays for Detection of Agents Causing Membrane Damage in Staphylococcus Aureus." Journal of Antimicrobial Chemotherapy. 54: 1127-1129.

243 Bibliography

Paine, A. J. (1996). "Validity and Reliability of in Vitro Systems in Safety Evaluation." Environmental Toxicology and Pharmacology. 2: 207-212.

Paustenbach, D. (2000). "The Practice of Exposure Assessment: A State-of-the-Art Review." Journal of toxicology and environmental health 3: 179-291.

Payne, J., Rajapakse N., Wilkins M. and Kortenkamp A. (2000). "Prediction and Assessment of the Effects of Mixtures of Four Xenoestrogens." Environmental Health Perspectives. 108(10): 983-987.

Payne, J., Scholze M. and Kortenkamp A. (2001). "Mixtures of Four Organochlorines Enhance Human Breast Cancer Cell Proliferation." Environmental Health Perspectives. 109(4): 391-397.

Petty, R. D., Sutherland L. A., Hunter E. M. and Cree I. A. (1995). "Comparison of MTT and ATP-Based Assays for the Measurement of Viable Cell Number." Journal of Bioluminiscence and Chemiluminiscence 10: 29-34.

Pohl, H., Hansen H., Selene C. H. and Chou J. (1997). "Public Health Guidance Values for Chemical Mixtures: Current Practice and Future Directions." Regulatory Toxicology and Pharmacology. 26: 322-329.

Ponsoda, X., Jover R., Nunez C., Royo M., Castell J. and Gomez-Lechon M. (1995). "Evaluation of the Cytotoxicity of 10 Chemicals in Human and Rat Hepatocytes and in Cell Lines: Correlation between in Vitro Data and Human Lethal Concentrations." Toxicology In Vitro 9: 959-966.

Posh, G. (1993). Combined Effects of Drugs and Toxic Agents. Modern Evaluation in Theory and Practice. NY, Springer Verlag.

Pounds, K. G., Haider J., Chen D. G. and Mumtaz M. (2004). "Interactive Toxicity of Simple Chemical Mixtures of Cadmium, Mercury, Methylmercury and Trimethyltin:

244 Bibliography

Model-Dependent Responses." Environmental Toxicology and Pharmacology. 18: 101- 113.

Pratt, I. S. (2002). "Global Harmonisation of Classification and Labelling of Hazardous Chemicals." Toxicology Letters 10: 5-15.

Preston, S., Coad N., Townend J., Killham K. and Paton G. (2000). "Biosensing the Acute Toxicity of Metal Interactions: Are They Additive, Synergistic, or Antagonistic?" Environmental Toxicology and Chemistry 19(3): 775-780.

Prieto, P., B.J. B., de Boer A. G., Boveri M., Cecchelli R., Clemedon C., Coecke S., Forsby A., Galla H. J., Garberg P., Greenwood J., Price A. and Tahti H. (2004). "Blood- Brain in Vitro Models and Their Application in Toxicology." Alternatives to Laboratory Animals 32: 37-50.

Promega (2001a). Celltiter 96r Aqueous Non-Radioactive Cell Proliferation Assay, Technical Bulletin No.169, Promega.

Promega (2001b). "Celltiter-Glotm Luminescent Cell Viability Assay." Promega Technical Bulletin 288.

Promega (2004). "Cytotox 96r Non-Radioactive Cytotoxicity Assay." Promega technical bulletin (TB 163).

Purchase, I., Botham P. A., Bruner L. H., Flint O. P., Frazier J. M. and Stokes W. S. (1998). "Workshop Overview: Scientific and Regulatory Challenges for the Reduction, Refinement, and Replacement of Animals in Toxicity Testing." Toxicological Sciences. 43: 86-101.

Putnam, K. P., Bombick D. W. and Doolittle D. J. (2002). "Evaluation of Eight in Vitro Assays for Assessing the Cytotoxicity of Cigarette Smoke Condensate." Toxicology In Vitro 16: 599-607.

245 Bibliography

Rajapakse, N., Ong D. and Kortenkamp A. (2001). "Defining the Impact of Weakly Estrogenic Chemicals on the Action of Steroidal Estrogens." Toxicological Sciences. 60: 296-304. Reinhardt, C., Pelli D. A. and Zbinden G. (1985). "Interpretation of Cell Toxicity Data for Estimation of Potential Irritation." Food and Chemical Toxicology: 231-247.

Ribo, J. M. and Rogers F. (1990). "Toxicity of Mixtures of Aquatic Contaminants Using the Luminescent Bacteria Bioassay." Toxicity Assessment 5: 135-152.

Riss, T., O'brien M. and Moravec R. (2003). "Choosing the Right Cell-Based Assay for Your Research." Cell Notes 6: 6-12.

Romert, L., Janson T. and Jenssen D. (1994). "The Cytotoxicity of 50 Chemicals from the MEIC Study Determined by Growth Inhibition of Ascites Sarcoma BP8 Cells: A Comparison with Acute Toxicity Data in Man and Rodents." Toxicology Letters 71: 39- 46.

Russell, W. M. S. and Burch R. L. (1959). The Principles of Humane Experimental Technique Methuan, London, England.

Safe, S. H. (1998). "Hazard and Risk Assessment of Chemical Mixtures Using the Toxic Equivalency." Environmental Health Perspectives. 106(4): 1051-1058.

Sauvant, M. P., Depin D., Bohatier J., Groliere C. A. and Guillot J. (1997). "Toxicity Assessment of 16 Inorganic Environmental Pollutants by Six Bioassays." Ecotoxicology and Environmental Safety 37: 131-140.

Schechtman, L. M. (2002). "Implementation of the 3rs (Refinement, Reduction, and Replacement): Validation and Regulatory Acceptance Considerations for Alternative Toxicological Test Methods." Institute for Laboratory animal Research (ILAR) journal 43.

246 Bibliography

Scheers, E. M., Ekwall B. and Dierickx P. J. (2001). "In Vitro Long Term Cytotoxicity Testing of 27 MEIC Chemicals on Hep G2 Cells and Comparison with Acute Human Toxicity Data." Toxicology In Vitro 15: 153-161.

Segner, H., Navas J. M., Schafers C. and Wenzel A. (2003). "Potencies of Estrogenic Compounds in in Vitro Screening Assays and in Life Cycle Tests with Zebrafish in Vivo." Ecotoxicology and Environmental Safety 54: 315-322.

Seibert, H., Balls M., Fentem J. H., Bianchi V., Clothier H., Dierickx P. J., Ekwall B., Garle M. J., Gomez-Lechon M., Gribaldo L., Gulden M., Liebsch M., Rasmussen E., Roguet R., Shrivastava R. and Walum E. (1996). "Acute Toxicity Testing in Vitro and the Classification and Labelling of Chemicals." Alternatives to Laboratory Animals 24: 499-510.

Sheghizzi, P., D'adda F., Barbic D. and Mosconi G. (1994). "Cobalt Cardiomyopathy: A Critical Review of the Literature." The Science of the Total Environment 150: 105-109.

Shrivastava, R., Delomenie C., Chevalier A., John G., Ekwall B., Walum E. and Massingham R. (1992). "Comparison of in Vivo Acute Lethal Potency and in Vitro Cytotoxicity of 48 Chemicals." Cell Biology and Toxicology 8(2): 157-170.

Sielken, R. and Stenvenson D. (1998). "Some Implications for Quantitative Risk Assessment If Hormesis Exists." Human and Experimental Toxicology 17: 259-262.

Silva, E., Rajapakse N. and Kortenkamp A. (2002). "Something from "Nothing" - Eight Weak Estrogenic Chemicals Combined at Concentrations below NOECS Produce Significant Mixture Effects." Environmental Science and Technology. 36: 1751-1756.

Smith, D. H. (1978). Alternatives to Animal Experiments. London, UK, Scholar Press.

Snyder, R. and Andrews L. S. (2001). Toxic Effects of Solvents and Vapours. Casarett's and Doull's Toxicology:The Basic Science of Poisons. C. Klaassen. NY, McGraw-Hill.

247 Bibliography

Society of Toxicology (2002).

Spielmann, H., Genschow E., Liebsch M. and Halle W. (1999). "Determination of the

Starting Dose for Acute Oral Toxicity (LD50) Testing in the up and Down Procedure (UDP) from Cytotoxicity Data." Alternatives to Laboratory Animals 27(957-966).

Spielmann, H. and Liebsch M. (2001). "Session 9: Education, Validation and Regulatory Issues. Lessons Learned from Validation of in Vitro Toxicity Test: From Failure to Acceptance into Regulatory Practice." Toxicology In Vitro 15: 585-590.

Stark, D. M., Shopsis C., Borenfreund E. and Babich H. (1986). "Progress and Problems in Evaluating and Validating Alternative Assays in Toxicology." Food and Chemical Toxicology 24(6/7): 449-455.

Stavric, B. (1988). "Methylxanthines: Toxicity to Humans. 2. Caffeine." Food and Chemical Toxicology 26(7): 645-662.

Taskforce, R. R. (2000). Reform of Public Health and Safety Regulation: An Information Paper Overview of Existing Regulatory Arrangements. Canberra, Commonwealth Department of Health and Ageing.

Teuschler, L., Klaunig J., Carney E., Chambers J., Conolly R., Gennings C., Giesy J., Hertzberg R. C., Klaassen C., Kodell R., Paustenbach D. and Yang R. (2002). "Support of Science-Based Decisions Concerning the Evaluation of the Toxicology of Mixtures: A New Beginning." Regulatory Toxicology and Pharmacology. 36: 34-39.

Thorne, P. S. (2001). Occupational Toxicology. Casarrett and Doull's Toxicology: The Basic Science of Poisons. C. Klaassen. New York, McGraw-Hill.

UN (2005). Globally Harmonized System of Classification and Labelling of Chemicals (GHS), United Nations Economic Commission for Europe (UNECE).

248 Bibliography

Weiss, M. T. and Sawyer T. W. (1993). "Cytotoxicity of the MEIC Test Chemicals in Primary Neuron Cultures." Toxicology In Vitro 7(5): 653-667.

Wilkinson, C. F., Christoph G. R., Julien E., Kelley M., Kronenberg J., McCarthy J. and Reiss R. (2000). "Assessing the Risks of Exposures to Multiple Chemicals with a Common Mechanism of Toxicity: How to Cumulate?" Regulatory Toxicology and Pharmacology. 31(30-43).

Williams, P. and Paustenbach D. (2002). "Risk Characterization: Principles and Practice." Toxicology and Environmental Health Part B 5(4): 337-406.

Wilson, A. P. (2000). Cytotoxicity and Viability Assays. Animal Cell Culture: A Practical Approach. J. R. W. Masters. Oxford, Oxford university press.

Winder, C., Azzi R. and Wagner D. (2005). "The Development of the Globally Harmonized System (GHS) of Classification and Labelling of Hazardous Chemicals." Journal of Hazardous Materials 125(1-3): 29-44.

Winder, C. and Barter J. (2004). Worplaces and Law. Occupational Toxicology 2nd Edition. C. Winder and N. H. Stacey. Boca Raton, CRC press: 91-521.

Worth, A. and Balls M. (2002). "Alternative (Non-Animal) Methods for Chemical Testing: Current Status and Future Prospects." ATLA 30(1): 1-125.

Yang, A., Cardona D. L. and Barile F. A. (2002). "In Vitro Cytotoxicity Testing with Fluorescence-Based Assays in Cultured Human Lung and Dermal Cells." Cell Biology and Toxicology 18: 97-108.

Yang, R. S. H. (1994). Toxicology and Chemical Mixtures: Case Studies, Mechanisms, and Novel Approaches. San Diego, Academic Press.

249 Bibliography

Yang, R. S. H. (1994b). Toxicology of Chemical Mixtures Derived from Hazardous Waste Sites or Application of Pesticides and Fertilisers. Toxicology of Chemical Mixtures., Academic Press Inc.: 99-114.

Yang, R. S. H. (1996). "Some Current Approaches for Studying Combination Toxicology in Chemical Mixtures." Food and Chemical Toxicology 34: 1037-1044.

Yang, R. S. H. (1998). "Some Critical Issues and Concerns Related to Research Advances on the Toxicology of Chemical Mixtures." Environmental Health Perspectives. 106(4): 1059-1063.

Yang, W. and Acosta D. (1994). "Cytotoxicity Potential of Surfactant Mixtures Evaluated by Primary Cultures of Rabbit Corneal Epithelial Cells." Toxiology Letters 70: 309-318.

Zarei, A. and Markovic B. (2000). "Refinement in the Use and Data Analysis of Promega Celltiter 96aqueous Non-Radioactive Cell Proliferation Assay."

Zhu, Q. X., Shen T., Ding R., Liang Z. Z. and Zhang X. J. (2005). "Cytotoxicity of Trichloroethylene and Perchloroethylene on Normal Human Epidermal Keratinocytes and Protective Role of Vitamin E." Toxicology 209: 55-67.

Zucco, F., De Angelis I., Testai E. and Stammati A. (2004). "Toxicological Investigations with Cell Culture Systems: 20 Years After." Toxicology In Vitro 18: 153- 163.

Zuckrebraun, H. L., Babich H., May R. J. and Sinensky M. C. (1998). "Triclosan: Cytotoxicity, Mode of Action, and Induction of Apoptosis in Human Gingival Cells in Vitro." European Journal of Oral Sciences. 106: 628-636.

Zurlo, J., Rudacille D. and Goldberg A. M. (1994). Animal and Alternatives in Testing: History, Science, and Ethics., Mary Ann Liebert, Inc.

250 Appendix A

Appendix A

A1 In vitro cytotoxicity assay reagents

The MTS assay

MTS detection reagent MTS Reagent Powder (Promega, G1111) was suspended in a container protected from light with Dulbecco’s Phosphate Buffered Saline (DPBS) (Sigma D-8537) at a concentration of 2 mg/ml. The MTS solution was dissolved and the pH measured and adjusted to pH 6.5 with 1 M HCl, if necessary. The solution was filter sterilised (0.22 ȝm) into a sterile, light protected container in 10 ml aliquots and stored at -20°C until required for use.

PMS detection reagent PMS reagent powder (Sigma P-5812) was resuspended in a light protected container with DPBS at a concentration of 0.92 mg/ml. The PMS solution was dissolved and filter sterilised (0.22 ȝm) into a sterile, light protected container in 1 ml aliquots and stored at -20°C until required for use.

MTS/PMS working solution Frozen containers of MTS and PMS reagents were thawed in a 37°C incubator for 30 mins prior to use. The MTS/PMS working solution was activated by mixing the MTS and PMS solution in a ratio of 20:1.

251 Appendix A

The Neutral Red Uptake (NRU) assay

Neutral Red medium (NR medium) The Neutral Red medium (NR medium) was prepared by transferring 600 ȝl of neutral red stock solution (Sigma, US) into 50 ml tubes containing culture medium to achieve a final concentration of 40 ȝg/ml. The NR medium was made 24 hours prior to use to allow for the precipitation of the undissolved dye. The NR medium was then centrifuged at 1500xg for 5 minutes and any precipitate discarded prior to use. The Supernatant was taken from the tubes, filter sterilised and used for the NRU assay.

Fixative solution The assay fixative was prepared from: 1gram of calcium chloride and in 1.3 ml 37% v/v formaldehyde solution dissolved into a total volume of 100 ml with distilled water.

Neutral red solubilisation solution The neutral red solubilisation solution was prepared from: 1 ml acetic acid and 50 ml ethanol dissolved into a total volume of 100 ml with distilled water.

The LDH assay

Lysis solution A lysis solution (10 x) is prepared from 9% (v/v) Triton X-100 in distilled water. 10 ȝl of lysis solution is added to all wells containing a 100 ȝl culture medium to lyse cells.

Reconstituted enzymatic substrate mixture Assay Buffer (12 ml) (Promega G1780) is added to substrate mix (Promega G1780), inverted and shaken gently to dissolve the substrate. 50 ȝl of the reconstituted substrate is then added to each well of the 96 well plates and covered with foil for 30 minutes prior to reading of absorbance. 12 ml of an assay buffer are usually enough for two 96 well plates.

252 Appendix A

A2 Optimal cell numbers

Skin fibroblast cell number against absorbance (MTS assay)

2.5

2.0 )

1.5

1.0

0.5 Absorbance (492nm Absorbance 0.0 0 50 100 150 200 250 -0.5 Cell number (x104 cells/ml)

Skin fibroblast cell number against absorbance (NRU assay)

1.5

0.5 Absorbance (at 540 nm) (at Absorbance 0 020406080 Cell number (104 cells/ml)

Skin fibroblast cell number against absorbance (LDH assay)

3.0

2.5

2.0

1.5

1.0

0.5 Absorbance (492 nm) Absorbance

0.0 0 50 100 150 200

Cell number (x 104 cells/ml)

253 Appendix A

A3 Optimisation of Serum Concentration

Effect of serum concentration on the LDH activity in culture supernatants

1 1% ABS 5% FCS Absorbance (492 nm)

0 Medium Cell Cell + Triton alone X-100

Culture supernatants of skin fibroblast and medium alone containing 1% ABS or 5% FBS were collected after 24h incubation and their LDH activity was measured. The activity is expressed as absorbance at 492 nm. All data represent the m ± SD of 4 replicates for each concentration

254 Appendix A

A4 Optimising of Exposure Time

Effect of 4h and 24h exposure for NaF on % LDH leakage in culture supernatants

120

24hrs exposure 60 4hrs exposure % LDH Leakage

0 0 1 1000 1000000

Concentration (mg/L)

Effect of 4h and 24h exposure for SDS on % LDH leakage in culture supernatant

100

24hrs exposure 50 4hrs exposure % LDH leakage

0 0.00 0.01 0.10 1.00 10.00 100.00 Concentration (mg/L)

255 Appendix B

Appendix B

B1 Statistical analysis

Calculation of standard error from MTS, NRU and ATP dose response curves

The variance of the mean around Equation 4-3 (Chapter 4) is calculated using the following equation:

2222 2 sY iiiisiibi sy se syn sen

ǔi represents the mean of the signals, as a measure of the number of absorbance

(nsi) wells containing cells and dilution series of chemicals;

Ɲi represents the mean of the signals as a measure of background absorbance of

chemicals obtained through nbi repeated measures. Yi represents the net mean of absorbance

And the standard error of the net mean is the square root of:

22 sY iisiibi s(y )n s(e)n

256 Appendix B

Sample concentrations and background control concentrations were added in quadruplets (Chapter 4: Figure 4-1) therefore (nsi = nbi = 4). Hence, the standard error of the net means is then calculated according to Equation 4-4 (Chapter 4).

Calculation of standard error for LDH dose response curve

Equation 4-6 (Chapter 4) was obtained using the following calculations: The variation around % LDH released is calculated:

22222 2 i iª ii iTiº sp p ¬ sYY sYY ¼

Yi represents net mean of absorbance for Extracellular LDH released (Chapter 4: Figure 4-5)

YTi represents total absorbance from: extracellular LDH+ intracellular LDH (YIi)

(Chapter 4: Figure 4-5) or YTi = Yi + YIi pi represents % LDH release calculated according to Equation 4-5 (Chapter 4).

Taking into account that sample concentrations and background control concentrations were added in quadruplets (Chapter 4: Figure 4-5) then Equation can be re-written as:

22 2 222 2 spiiiiiiiI1 pª s(y) s(e) Y s(y) s(e) s(y)iT Yiº 2 ¬ ¼

Calculation of TLC values for MTS, NRU and ATP assays

For the derivation of Equation 4-14 (Chapter 4) Cochran’s criterion was applied to prove the assumption of homogeneity of variance between the different absorbance values for the chemical concentrations in each experiment. Cochran’s criterion showed 2 2 that there was a significant difference between s (yi) and s (ei), but when they were taken separately, the variances were homogenous, independent of concentration.The

257 Appendix B

2 2 variations s (yi) and s (ei) were then replaced by their respective means to generate the following equations:

22 s(y) ¦ s(y)ki And

22 s(e) ¦ s(e)ki

2 s (yi) represents square of the variation around absorbance (Figure 4-1; column1) 2 s (ei) represents square of the variation around background absorbance (Figure 4-1; column 1) k being the total number of the different tested concentrations.

Knowing also that each concentration was tested in quadruplets for the sample and the background, the standard error of the means of the samples and background becomes ½*s(y) and ½*s (e), respectively. Taking into account type I error (Į) (the sample’s signal being different from the blank, even though it is confounded with the blank) the critical level is then written:

Yc 1 2 tD ,3 s(e)

Where tĮ, 3 is the Student criterion for a one sided test with an error risk Į and 3 degrees of freedom.

Taking into account type II error (ȕ) which consists in incorrectly accepting that the signal from the sample as confounded with the blank, while in reality it is not. The limit of detection becomes:

Y12ts(e)12ts(y)TLC D ,3 E,3

If we take a confidence interval of 95 % (Į = ȕ = 0.05, where tĮ, 3 = tȕ, 3 = 2.353), the limit of detection of the method is then given by:

YTLC 1 2 2.353 > s(e) s(y) @ 1.18 > s(e) s(y)@ And the TLC is then calculated according to Chapter 4 Equation 4-14

258 Appendix C

Appendix C

C1 In vitro cytotoxicity data

NOAEC values for test chemicals as determined by MTS, NRU, LDH and ATP assays.

NOAEC Chemicals MTS SD NRU SD LDH SD ATP SD mmol/L (±) mmol/L (±) mmol/L (±) mmol/L (±) Caffeine 0.33 0.03 0.50 0.009 0.70 0.07 1.15 0.17 Cadmium chloride 0.40 0.01 0.0002 0.0001 0.001 0.0003 0.15 0.02 Cobalt Chloride 0.002 0.001 0.20 0.02 0.38 0.04 0.13 0.04 Colchicine 0.07 0.01 0.15 0.04 0.40 0.06 1.36 0.15 Cupric sulfate 0.007 0.001 0.03 0.01 0.35 0.01 0.05 0.01 Ethanol 35.68 7.15 6.29 2.28 7.58 1.24 59.67 7.67 Glycerol 26.93 3.69 21.71 4.34 26.05 1.53 7.30 0.41 Irgasan 0.03 0.005 0.01 0.004 0.31 0.03 0.19 0.03 Lithium sulfate 4.44 0.88 0.29 0.06 5.10 1.03 4.89 0.28 Mercuric chloride 0.0003 0.0001 0.0001 8.24E-05 0.003 0.0004 7.37E-05 5.21E-05 Methanol 75.92 10.00 47.40 21.20 70.72 7.85 90.48 4.41 Phenol 0.58 0.14 0.02 0.02 2.67 0.12 0.07 0.02 Potassium chloride 8.49 1.53 1.13 0.10 18.69 3.67 12.00 1.80 Potassium cyanide 0.11 0.03 0.11 0.02 0.47 0.01 1.57 0.05 Saccharin 4.73 0.83 3.13 0.42 5.46 1.37 2.04 0.19 SLS 0.002 0.001 0.001 0.0001 0.005 0.001 0.002 0.0003 Sodium chloride 15.40 3.42 4.70 1.81 15.60 2.12 1.54 0.24 Sodium dichromate 0.001 0.0004 0.01 0.001 0.002 0.0002 0.03 0.004 Sodium fluoride 1.31 0.14 0.39 0.05 0.59 0.16 1.19 0.16 Sodium selenate 0.09 0.02 0.92 0.18 0.40 0.10 1.47 0.11 Strychnine 0.04 0.02 0.05 0.01 0.56 0.07 0.52 0.10

1 NOAEC data for skin fibroblasts represent the m ± SD for at least three separate experiments.

259 Appendix C

TLC values for test chemicals as determined by MTS, NRU, LDH and ATP assays

TLC Chemicals MTS SD NRU SD LDH SD ATP SD mmol/L (±) mmol/L (±) mmol/L (±) mmol/L (±) Caffeine 57.22 11.31 101.6 0.03 37.03 0.05 469.8 45.50 Cadmium chloride 35.72 4.78 0.37 0.02 0.02 0.002 5.03 0.92 Cobalt Chloride 0.29 0.04 27.71 0.87 4.62 0.59 91.74 10.85 Colchicine 30.03 0.01 22.01 0.14 11.62 0.50 85.11 14.16 Cupric sulfate 0.94 0.14 6.04 0.62 7.34 0.61 9.63 0.85 Ethanol 4031 436.5 1057 69.7 491.2 828.7 488.2 76.72 Glycerol 705.6 76.76 900.0 40 600.0 35.27 5428 1535 Irgasan 8.97 1.24 5.72 0.75 6.58 0.32 17.51 0.35 Lithium sulfate 741.3 46.82 82.31 5.88 132.7 6.84 345.1 2.96 Mercuric chloride 0.13 0.02 0.04 0.01 0.06 0.01 0.12 0.02 Methanol 13102 879.7 2408 166.1 747.2 15.68 2886 110.3 Phenol 102.1 6.52 32.95 0.91 26.92 6.50 35.94 3.97 Potassium chloride 1239 63.17 116.2 0 1683 1988 1418 108.4 Potassium cyanide 58.43 4.23 36.15 0.55 28.90 0.42 75.46 8.99 Saccharin 585.1 21.62 155.4 4.00 130.3 9.47 564.7 26.75 SLS 0.65 0.05 0.60 0.14 0.12 0.02 1.03 0.12 Sodium chloride 1528 66.19 263.4 9.60 396.6 21.03 860.0 27.62 Sodium dichromate 2.04 0.32 5.46 0.90 0.04 0.003 13.31 0.79 Sodium fluoride 159.6 10.45 48.09 1.01 41.98 3.13 41.67 8.41 Sodium selenate 233.5 6.55 43.50 0.40 5.15 0.17 179.9 29.94 Strychnine 118.5 1.53 30.47 0.03 7.81 3.52 240.1 7.26

1 TLC data for skin fibroblasts represent the m ± SD for at least three separate experiments.

260 Appendix C

C2 Cluster analysis of in vitro data

Data generated using SPSS for Windows version 10.0

Hierarchical clustering of selected chemicals using MTS assay

Dendrogram using Average Linkage (Between Groups)

Rescaled Distance Cluster Combine

C A S E 0 5 10 15 20 25 Label Num +------+------+------+------+------+

KCl 15 «± NaCl 16 «³ NaSe 14 «³ Saccharin 17 «³ LiSO4 18 ««± (1) Ethanol 19 «³ ²«««««««««««««««± Strychnine 13 «° ¬ ¬ Methanol 20 «´«° ¬ Glycerol 21 «° ²«««««««««««««««««««««««««««««± CdCl2 11 «± ¬ ¬ Phenol 12 ««± ¬ ¬ NaF 8 «³ ¬(2) ¬ ¬ Caffeine 9 «³ ²«««««««««««««««° ¬ Colchicine 10 «° ¬ ¬ Irgasan 6 «± ¬ ¬ KCN 7 ««° ¬ Cr(VI) 5 «° ¬ CuSO4 3 «± ¬ SLS 4 ««± ¬ (3) CoCl2 2 «° ²«««««««««««««««««««««««««««««««««««««««««««««° HgCl2 1 «««°

261 Appendix C

Grouping of chemicals measured with MTS assay (based on Dendrogram)

Category 1 Category 2 Category 3 5 > IC50 > 50 (mg/L) 50 > IC50 > 2000 (mg/L) IC50 > 5000 (mg/L) Mercuric chloride Cadmium chloride Potassium chloride Cobalt chloride Sodium dichromate Sodium chloride Copper sulfate Irgasan Sodium selenate SLS Potassium cyanide Lithium sulfate Colchicine Saccharin Caffeine Ethanol Sodium fluoride Strychnine Phenol Glycerol

Hierarchical clustering of selected chemicals using LDH assay

Dendrogram using Average Linkage (Between Groups)

Rescaled Distance Cluster Combine

C A S E 0 5 10 15 20 25 Label Num +------+------+------+------+------+

LiSO4 18 «± Saccharin 19 «³ KCl 16 «³ Ethanol 17 ««± NaCl 15 «³ ¬ (1) Methanol 20 «³ ²«««««± Glycerol 21 «° ¬ ¬ Caffeine 13 «´«° ¬ Colchicine 14 «° ¬ Strychnine 10 «± ²«««««««««««««««««««««««««««««««««««««««± Irgasan 11 «³ ¬ ¬ NaF 9 «³ ¬ ¬ KCN 8 «³ ¬ ¬ Phenol 12 ««±(2) ¬ ¬ CoCl2 6 «³ ²«««««° ¬ NaSe 7 «° ¬ ¬ CuSO4 5 «««° ¬ CdCl2 1 «± ¬ HgCl2 2 ««± (3) ¬ Cr(VI) 3 «° ²«««««««««««««««««««««««««««««««««««««««««««««° SLS 4 «««°

262 Appendix C

Grouping of chemicals measured with LDH assay (based on Dendrogram)

Category 1 Category 2 Category 3 1 > IC50 > 50 (mg/L) 50 > IC50 > 2000 (mg/L) IC50 > 2000 (mg/L)

Mercuric chloride Copper sulfate Colchicine Cadmium chloride Sodium selenate Caffeine Sodium dichromate Cobalt chloride Methanol SLS Phenol Glycerol Potassium cyanide Sodium chloride Sodium fluoride Potassium chloride Strychnine Ethanol Irgasan Saccharin Lithium sulfate

Hierarchical clustering of selected chemicals using ATP assay

Dendrogram using Average Linkage (Between Groups)

Rescaled Distance Cluster Combine

C A S E 0 5 10 15 20 25 Label Num +------+------+------+------+------+

Ethanol 11 «± CoCl2 12 «³ NaCl 13 «³ Colchicine 14 «³ Strychnine 15 ««««± LiSO4 16 «³ ¬ (1) Caffeine 17 «³ ²«««««««««± Saccharin 18 «³ ¬ ¬ KCl 19 «° ¬ ¬ Methanol 20 «´«««° ¬ Glycerol 21 «° ¬ NaF 4 «± ²«««««««««««««««««««««««««««««««««± CuSO4 5 «³ ¬ ¬ CdCl2 3 «³ ¬ ¬ Irgasan 7 «³ ¬ ¬ Cr(VI) 8 «³ ¬ ¬ Phenol 6 ««««««± ¬ ¬ KCN 9 «³ ²«««««««°(2) ¬ NaSe 10 «° ¬¬ SLS 2 «««««««° ¬ (3) HgCl2 1 «««««««««««««««««««««««««««««««««««««««««««««««««°

263 Appendix C

Grouping of chemicals measured with ATP assay (based on Dendrogram)

Category 1 Category 2 Category 3 5 > IC50 (mg/L) 5 > IC50 > 2000 (mg/L) IC50 > 2000 (mg/L) Mercury chloride SLS Cobalt chloride Sodium selenate Ethanol Potassium cyanide Sodium chloride Phenol Colchicine Sodium dichromate Strychnine Irgasan Lithium sulfate Cadmium chloride Caffeine Sodium fluoride Saccharin Copper sulfate Potassium chloride Glycerol Methanol

264 Appendix C

C3 Comparison in vitro – GHS categories

Grouping of chemicals measured with MTS assay based on the GHS cut-off limits

Category 1 Category 2 Category 3 Category 4 Category 5 Unclassified

IC50 < 5 IC50 >5 - 50 IC50 >50 - 300 IC50 >300 - 2000 IC50 >2000 - 5000 IC50 > 5000 (mg/L) (mg/L) (mg/L) (mg/L) (mg/L) (mg/L)

N/A HgCl2 CrVI Potassium cyanide N/A Strychnine

CoCl2 Irgasan Sodium fluoride Sodium selenate SLS Caffeine Potassium chloride

CuSO4 Colchicine Sodium chloride Cadmium chloride Saccharin Phenol Lithium sulfate Ethanol Methanol Glycerol

Grouping of chemicals measured with NRU assay based on the GHS cut-off limits

Category 1 Category 2 Category 3 Category 4 Category 5 Unclassified

IC50 < 5 IC50 >5 - 50 IC50 >50 - 300 IC50 >300 - 2000 IC50 >2000 - 5000 IC50 > 5000 (mg/L) (mg/L) (mg/L) (mg/L) (mg/L) (mg/L)

HgCl2 CdCl2 CrVI Potassium cyanide Sodium chloride Saccharin SLS Copper sulfate Phenol Caffeine EtOH Sodium fluoride Irgasan Methanol Sodium selenate Glycerol Cobalt dichloride Colchicine Strychnine Lithium sulfate Potassium chloride

265 Appendix C

Grouping of chemicals measured with LDH assay based on the GHS cut off limits

Category 1 Category 2 Category 3 Category 4 Category 5 Unclassified IC50 < 5 IC50 >5 - 50 IC50 >50 - 300 IC50 >300 - 2000 IC50 >2000 - 5000 IC50 > 5000 (mg/L) (mg/L) (mg/L) (mg/L) (mg/L) (mg/L)

CdCl2 Cr (VI) Copper sulfate Sodium selenate Colchicine Sodium chloride HgCl2 SLS Cobalt dichloride Potassium cyanide Potassium chloride Sodium fluoride Ethanol Strychnine Lithium sulfate Irgasan Saccharin Phenol Methanol Caffeine Glycerol

Grouping of chemicals measured with ATP assay based on the GHS cut off limits

Category 1 Category 2 Category 3 Category 4 Category 5 Unclassified IC50 < 5 IC50 >5 - 50 IC50 >50 - 300 IC50 >300 - 2000 IC50 >2000 - 5000 IC50 > 5000 (mg/L) (mg/L) (mg/L) (mg/L) (mg/L) (mg/L)

HgCl2 SLS N/A Cadmium chloride N/A Ethanol Sodium fluoride Cobalt dichloride Cupric sulfate Sodium chloride Phenol Colchicine Irgasan Strychnine Sodium dichromate Lithium sulfate Potassium cyanide Caffeine Sodium selenate Saccharin Potassium chloride Methanol Glycerol

266 Appendix D

Appendix D

D1 Statistical analysis for chemical mixtures

Calculation of 95% CI around IC50 predicted for chemical mixtures

95% confidence interval (CI) was calculated to account for the variability around predicted IC50 value, and setting upper and lower limits. Deriving from Equation 6-5 (Chapter 6) then:

2222222 s((IC50 predicted) X A)s( (IC50 A ) X B)s. (IC50B ) ..( Xn)s (IC50n )

Hence,

22 22 22 s((IC50 predicted) X A)s( (IC50 A ) X B)s. (IC50B ) ..( Xn)s (IC50n )

Xi represents the ratio used to combine chemical mixtures. It is derived from Equation 4-23 (Chapter 4).

IC50A, IC50B, IC50n represent 50% inhibitory concentrations of each component in the mixture (mg/L), when administered individually.

267 Appendix D

D2 Correlation of experimental and predicted IC50

Correlation of experimental and predicted IC50 of binary mixtures (MTS assay)

Antagonism

2.5 exp (mg/L) exp 50

Synergism Log IC 0 02.55

Log IC50 pred (mg/L)

The Line represents a 1:1 correlation: marginally lower and higher data are obtained experimentally indicating effects greater than additive (synergism) and lower than additive (antagonism), respectively.

Correlation of experimental and predicted IC50 for ternary mixtures (MTS assay)

Antagonism

2.5 exp (mg/L) exp 50 Synergism Log IC 0 012345

Log IC50 pred (mg/L)

The line represents a 1:1 correlation: marginally lower and higher data are obtained experimentally indicating effects greater than additive (synergism) and lower than additive (antagonism), respectively.

268 Appendix D

Correlation of experimental and predicted IC50 of binary mixtures (NRU assay)

Antagonism

2.5 exp (mg/L) exp 50

Synergism Log IC 0 02.55

Log IC50 pred (mg/L)

Correlation of experimental and predicted IC50 of ternary mixtures (NRU assay)

Antagonism

2.5 exp (mg/L) 50

Synergism Log IC

0 02.55

Log IC50 pred (mg/L)

269 Appendix D

D3 Comparison in vitro – GHS categories of chemical mixtures

Grouping of chemicals measured with MTS assay Category Category 2 Category 3 Category 4 Category 5 Unclassified 1 IC50 >5 - IC50 >50 - 300 IC50 >300 - 2000 IC50 >2000 - 5000 IC50 > 5000 IC50 < 5 50 (mg/L) (mg/L) (mg/L) (mg/L) (mg/L) (mg/L)

N/A SLS: Irgasan SLS: NaF: CrVI: CuSO4: KCN: CdCl2 EtOH: CdCl2: HgCl2 Irgasan Phenol CdCl2: HgCl2 LiSO4 SLS NaF: SLS NaF: Irgasan NaF: Phenol CdCl2: LiSO4 CuSO4 CrVI: CuSO4 CuSO4: Phenol EtOH: LiSO4 CrVI CrVI: Phenol EtOH: CdCl2 Irgasan KCN: HgCl2: KCl: NaF: Phenol CdCl2 KCl: NaF KCN: HgCl2 KCl: Phenol KCN KCl NaF LiSO4 CdCl2 EtOH Phenol

Grouping of chemicals measured with NRU assay Category 1 Category 2 Category 3 Category 4 Category 5 Unclassified

IC50 < 5 IC50 >5 - 50 IC50 >50 - 300 IC50 >300 - 2000 IC50 >2000 - 5000 IC50 > 5000 (mg/L) (mg/L) (mg/L) (mg/L) (mg/L) (mg/L)

HgCl2 CdCl2: SLS: NaF: Irgasan KCN: HgCl2: CdCl2 CrVI: CuSO4: EtOH: CdCl2: LiSO4 HgCl2 SLS: Irgasan KCl: NaF: Phenol Phenol EtOH: LiSO4 CdCl2 NaF: SLS KCl: NaF KCl: Phenol EtOH: CdCl2 SLS NaF: Irgasan CrVI: CuSO4 EtOH CrVI CuSO4: Phenol CuSO4 CrVI: Phenol NaF NaF: Phenol CdCl2: LiSO4 KCN: HgCl2 KCN: CdCl2 KCN Phenol Irgasan LiSO4 KCl

270 Appendix D

D4 Parameters and graphical representation of binary chemical mixtures

Parameters and binary graphical representation for the MTS assay

Definition of parameters for the following Tables: s2 pred square of the standard deviation. Method of derivation can be found in Appendix D1.

IC50pred derived from Equation 6-5 (Chapter 6) t*s calculation of 95% confidence interval based on student t test (df = 4).

X(chemical) ratio of IC50 of component chemical in the mixture

Graphical representations of the following binary mixtures are present in the text Section 6.5.3: NaF: SLS p.207

CuSO4: phenol p.206 NaF: phenol p.205

271 Appendix D

Binary mixture: NaF: irgasan showing an additive interaction

2 X(irgasan) X(NaF) S (pred) t*s IC50 (pred) (*104) 0 1 9.07 264.4 1045 0.187 0.813 6.09 216.7 894.6 0.25 0.75 5.27 201.5 843.9 0.5 0.5 2.94 150.6 642.6 0.75 0.25 2.08 126.7 441.3 1 0 2.69 144.2 240.0

1400 1200 antagonistic interaction 1000 800

(mg/L) 600 additive interaction 50 exp

IC 400 200 synergistic interaction 0 00.51 % Irgasan:NaF

Graphical representation of an additive interaction between binary mixture (NaF and irgasan)

272 Appendix D

Binary mixture: SLS: irgasan showing a synergistic interaction

2 X(SLS) X(irgasan) S (pred) t*s IC50 (pred) 0 1 2699 144.2 240.0 0.104 0.896 2167 129.2 217.9 0.25 0.75 1520 108.2 186.9 0.5 0.5 680.6 72.42 133.9 0.75 0.25 181.5 37.40 80.85 1 0 22.75 13.24 27.81

450 400 antagonistic 350 300 250

(mg/L) 200 50

IC 150 additive interaction 100 50 synergistic Exp 0 00.51 % SLS:Irgasan

Graphical representation of a synergistic interaction between binary mixture (SLS and irgasan)

273 Appendix D

Binary mixture: Cr (VI): CuSO4 showing an additive interaction

Parameters for the calculation of predicted IC50 value and 95% CI upper and lower limits

2 X(CuSO4) X(CrVI) s (pred) t*s IC50 (pred) 0 1 400.0 63.64 168.3 0.18 0.817 269.9 52.28 144.3 0.25 0.75 230.5 48.31 135.6 0.5 0.5 122.3 35.19 102.9 0.75 0.25 75.34 27.61 70.30 1 0 89.49 30.10 37.64

250

Antagonistic interaction 200

150

(mg/L) Additive interaction exp 50 100 IC

Synergistic interaction 0 0 0.2 0.4 0.6 0.8 1 % CuSO4:CrVI

274 Appendix D

Binary mixture: Cr (VI): phenol showing an additive interaction

Parameters for the calculation of predicted IC50 value and 95% CI upper and lower limits

2 X(CrVI) X(phenol) S (pred) t*s IC50 (pred) (*104) 0 1 4 555.2 1655 0.092 0.908 3.39 504.1 1518 0.25 0.75 2.22 416.6 1283 0.5 0.5 1.01 278.9 911.6 0.75 0.25 0.27 144.9 539.9 1 0 0.04 55.52 168.3

2500

antagonistic interaction 2000

1500

(mg/L) additive interaction 50 1000 IC Exp

500 synergistic interaction

0 0 0.2 0.4 0.6 0.8 1 %CrVI:phenol

Graphical representation of an additive interaction between binary mixture (Cr (VI) and phenol)

275 Appendix D

Binary mixture: CdCl2: LiSO4 showing an additive interaction

Parameters for the calculation of predicted IC50 value and 95% CI upper and lower limits

2 XCdCl2 XLiSO4 S (pred) t*s IC50 (pred) (*104) 0 1 7744 22428 25030

0.25 0.75 4356 18322 19147

0.06 0.094 6886 23036 23688

0.5 0.5 1937 12217 13265

0.75 0.25 4863 6121 7383

1 0 4 555.2 1501

60000

antagonistic it ti

30000 (mg/L)

50 additive interaction IC

exp

0 0 0.2 0.4 0.6 0.8 1

%CdCl2:LiSO4

Graphical representation of an additive interaction between binary mixture

(CdCl2 and LiSO4). N.B. in this case the CI for the graph is very wide making it difficult to detect a possible synergistic interaction

276 Appendix D

Binary mixture: HgCl2: CdCl2 showing an antagonistic interaction

2 X(HgCl2) X(CdCl2) S (pred) t*s IC50 (pred) (*104) 0 1 4.00 555.2 1501 0.003 0.997 3.97 553.5 1496 0.25 0.75 2.25 416.4 1127 0.5 0.5 1.00 277.6 753.1 0.75 0.25 0.25 138.8 379.1 1 0 0.00006 2.14 5.25

2500 exp. 2000 antagonistic interaction

1500

(mg/L) additive interaction 50 1000 IC

500 synergistic interaction 0 00.20.40.60.81

%HgCl2:CdCl2

Graphical representation of an antagonistic interaction between binary mixture

(HgCl2 and CdCl2)

277 Appendix D

Binary mixture: HgCl2: KCN showing an antagonistic interaction

2 X(HgCl2) X(KCN) S (pred) t*s IC50 (pred) 0 1 1225 97.16 355.5 0.015 0.985 1188 95.70 350.2 0.25 0.75 689.1 72.87 267.9 0.5 0.5 306.4 48.59 180.3 0.75 0.25 76.90 24.34 92.81 1 0 0.59 2.13 5.25

1500

antagonistic interaction 1000 exp. (mg/L) 50

IC 500

additive interaction 0 00.51synergistic interaction

% HgCl2:KCN

Graphical representation of an antagonistic interaction between binary mixture

(HgCl2 and KCN)

278 Appendix D

Binary mixture: CdCl2: KCN showing an antagonistic interaction

2 X(KCN) X(CdCl2) S (pred) t*s IC50 (pred) 0 1 40000 555.2 1501 0.191 0.8009 25702 445.0 1270 0.25 0.75 22576 417.1 1214 0.5 0.5 10306 281.8 928.3 0.75 0.25 3189 156.7 641.8 1 0 1225 97.16 355.5

2500

2000

1500 (mg/L)

50 1000 IC 500

0 00.51

%KCN:CdCl2

Graphical representation of an antagonistic interaction between binary mixture

(CdCl2 and KCN)

279 Appendix D

Binary mixture: NaF: KCl showing an additive interaction

2 X(NaF) X(KCl) S (pred) t*s IC50 (pred) (*104) 0 1 2500 13880 17328 0.057 0.943 2223 13088 16399 0.25 0.75 1406 10410 13257 0.5 0.5 625.2 6941 9186 0.75 0.25 156.7 3475 5115 1 0 9074 264.4 1045

35000 30000 antagonistic interaction 25000 20000

(mg/L) 15000 50

IC additive 10000 Exp. 5000 0 00.51 %NaF:KCl

Graphical representation of an additive interaction between binary mixture (NaF and KCl) N.B. in this case the CI for the graph is very wide making it difficult to detect a possible synergistic interaction

280 Appendix D

Parameters and binary graphical representation for the NRU assay

Binary mixture: NaF: irgasan showing a synergistic interaction

2 NaF Irgasan S (pred) t*s IC50 (pred) 0 1 1000 145. 374 0.25 0.75 683.5 72.57 354.2 0.441 0.559 689.0 72.86 339.1 0.5 0.5 734.0 75.20 334.5 0.75 0.25 115 94.20 314.7 1 0 1936 122.1 295.0

600

antagonistic interactions

additive interactions 300 (mg/L)

50 exp IC

synergistic interactions 0 00.51 % NaF:irgasan

Graphical representation of a synergistic interaction between binary mixture (NaF and irgasan)

281 Appendix D

Binary mixture: NaF: SLS showing an additive interaction

2 X (SLS) X (NaF) S (pred) t*s IC50 (pred) 0 1 1936 140.0 295.0 0.124 0.876 1486 122.6 263.6 0.25 0.75 1092 105.1 231.7 0.5 0.5 498.8 71.07 168.4 0.75 0.25 154.4 39.54 105.1 1 0 59.44 24.53 41.9

500

antagonistic interaction

250 (mg/L) 50 IC additive interaction

synergistic interaction 0 00.51 % NaF: SLS

Graphical representation of an additive interaction between binary mixture (SLS and NaF)

282 Appendix D

Binary mixture: SLS: irgasan showing a synergistic interaction

2 X (SLS) X (irgasan) S (pred) t*s IC50 (pred) 0 1 2500 138.8 374.1 0.101 0.899 2021 124.7 340.6 0.25 0.75 1409 104.2 291.1 0.5 0.5 639.8 70.22 208.0 0.75 0.25 189.6 38.23 124.9 1 0 59.44 21.40 41.9

600 antagonisti c

300 additive (mg/L) interactions 50 IC

synergistic interactions exp 0 00.51 % SLS:irgasan

Graphical representation of a synergistic interaction between binary mixture (SLS and irgasan)

283 Appendix D

Binary mixture: Cr (VI): CuSO4 showing an antagonistic interaction

2 X (CuSO4) X (CrVI) S (pred) t*s IC50 (pred) (*104) 0 1 9.20 266.2 214.1 0.487 0.513 3.15 156.0 208.9 0.25 0.75 5.36 203.4 211.4 0.5 0.5 3.07 153.9 208.7 0.75 0.25 2.32 133.8 206.0 1 0 0.31 154.7 203.4

800 700 antagonistic exp 600 interaction 500 400

(mg/L) 300 50 additive interaction

IC 200 100 0 synergistic interaction -100 00.51

% CrVI:CuSO4

Graphical representation of an antagonistic interaction between binary mixture

(Cr (VI) and CuSO4)

284 Appendix D

Binary mixture: Cr (VI): phenol showing an antagonistic interaction

2 X (CrVI) X (phenol) S (pred) t*s IC50 (pred) 0 1 216.6 40.86 424.7 0.335 0.665 1128 93.24 354.1 0.25 0.75 696.8 73.28 372.1 0.5 0.5 2354 134.6 319.4 0.75 0.25 5188 199.9 266.7 1 0 9200 266.2 214.1

1200 1000 antagonistic interaction exp. 800 600

(mg/L) 400 50 additive interaction IC 200 0 synergistic interaction -200 00.51 % CrVI: phenol

Graphical representation of an antagonistic interaction between binary mixture (Cr (VI) and phenol)

285 Appendix D

Binary mixture: CuSO4: phenol showing an antagonistic interaction

2 X (CuSO4) X (phenol) S (pred) t*s IC50 (pred) 0 1 216.6 40.86 424.7 0.25 0.75 316.1 49.35 369.3 0.324 0.676 425.2 57.24 353.0 0.5 0.5 831.1 80.03 314.0 0.75 0.25 1761 116.5 258.7 1 0 3108 154.7 203.4

1600 exp antagonistic

800 (mg/L) 50 IC

additive interactions synergistic interactions 0 00.51 % CuSO4: phenol

Graphical representation of an antagonistic interaction between binary mixture

(CuSO4 and phenol)

286 Appendix D

Binary mixture: CdCl2: LiSO4 showing an additive interaction

2 X(CdCl2) X(LiSO4) S (pred) t*s IC50 (pred) (*104) 0 1 9.00 832.8 1925 0.0029 0.997 8.94 830.3 1919 0.25 0.75 5.06 624.6 1445 0.5 0.5 2.25 416.4 965.4 0.75 0.25 0.56 208.2 485.5 1 0 0.0002 4.02 5.63

3000

antagonistic interaction

1500 (mg/L)

50 additive interaction Exp IC

synergistic interaction 0 00.51

% CdCl2: LiSO4

Graphical representation of an additive interaction between binary mixture

(CdCl2 and LiSO4)

287 Appendix D

Binary mixture: LiSO4: EtOH showing an antagonistic interaction

2 X (LiSO4) X (EtOH) S (pred) t*s IC50 (pred) (*104) 0 1 361.0 5274 11772 0.141 0.859 266.5 4532 10383 0.25 0.75 203.6 3961 9310 0.5 0.5 92.5 2669 6848 0.75 0.25 27.6 1459 4386 1 0 9.00 832.8 1925

24000

antagonistic interactions

16000 (mg/L) 50

IC 8000 additive interactions

synergistic interactions 0 00.51

% LiSO4:EtOH

Graphical representation of an antagonistic interaction between binary mixture

(LiSO4 and ethanol)

288 Appendix D

Binary mixture: HgCl2: KCN showing an antagonistic interaction

2 X (HgCl2) X (KCN) S (pred) t*s IC50 (pred) 0 1 2304 133.2 357.1 0.009 0.991 2262 132. 353.9 0.25 0.75 1296 99.93 268.6 0.5 0.5 576.0 66.62 180.1 0.75 0.25 144.1 33.32 91.60 1 0 0.24 1.36 3.08

1050 antagonistic interaction

700 exp (mg/L) 50

IC 350 additive interaction

0 00.51

% HgCl2: KCN

Graphical representation of an antagonistic interaction between binary mixture

(HgCl2 and KCN)

289 Appendix D

Binary mixture: CdCl2: KCN showing an additive interaction

2 X (CdCl2) X (KCN) S (pred) t*s IC50 (pred) 0 1 2304 133.2 357.1 0.015 0.985 2235 131.2 351.8 0.25 0.75 1296 99.94 269.2 0.5 0.5 576.5 66.65 181.3 0.75 0.25 145.1 33.44 93.49 1 0 2.10 4.02 5.6

600 antagonistic interaction

exp 300

(mg/L) additive 50 interaction IC

synergistic interaction 0 00.51

% CdCl2:KCN

Graphical representation of an additive interaction between binary mixture

(CdCl2 and KCN)

290 Appendix D

Binary mixture: HgCl2: CdCl2 showing an antagonistic interaction

2 X (HgCl2) X (CdCl2) S (pred) t*s IC50 (pred) 0 1 2.10 4.02 5.6 0.355 0.645 0.90 2.64 4.70 0.25 0.75 1.20 3.03 4.97 0.5 0.5 0.59 2.12 4.34 0.75 0.25 0.27 1.43 3.71 1 0 0.24 1.36 3.08

35 30 antagonistic exp 25 20

(mg/L) 15 50

IC 10 5 additive interactions 0 00.51

% HgCl2: CdCl2

Graphical representation of an antagonistic interaction between binary mixture

(HgCl2 and CdCl2)

291 Appendix D

Binary mixture: NaF: KCl showing an additive interaction

2 X (NaF) X (KCl) S (pred) t*s IC50 (pred) (*104) 0 1 9.00 832.8 1925 0.13 0.87 6.81 724.7 1712 0.25 0.75 5.07 625.3 1516 0.5 0.5 2.29 420.8 1107 0.75 0.25 0.67 227.4 697.9 1 0 0.19 122.1 288.8

3000

antagonistic interactions

1500 (mg/L) additive interactions exp 50 IC

synergistic 0 00.51 % NaF: KCl

Graphical representation of an additive interaction between binary mixture (NaF and KCl)

292 Appendix D

Binary mixture: phenol: KCl showing an antagonistic interaction

2 X (phenol) X (KCl) S (pred) t*s IC50 (pred) (*104) 0 1 9.00 832.8 1925 0.181 0.819 6.03 682.1 1653 0.25 0.75 5.06 624.6 1550 0.5 0.5 2.25 416.9 1175 0.75 0.25 0.57 210.4 799.8 1 0 0.021 40.89 424.7

4500 antagonistic interactions

3000 exp (mg/L)

50 additive

IC 1500 interactions

synergistic interactions 0 00.51 % phenol:KCl

Graphical representation of an additive interaction between binary mixture (phenol and KCl)

293 Appendix D

Binary mixture: NaF: phenol showing an antagonistic interaction

2 X (NaF) X (phenol) S (pred) t*s IC50 (pred) 0 1 216.6 40.86 424.7 0.25 0.75 242.8 43.26 390.7 0.405 0.595 394.2 55.12 369.6 0.5 0.5 538.1 64.39 356.7 0.75 0.25 1102 92.17 322.7 1 0 1936 122.1 288.8

600 antagonistic exp interactions

additive 300

(mg/L) interactions 50 IC synergistic interactions

0 00.51 % NaF:phenol

Graphical representation of an antagonistic interaction between binary mixture (NaF and phenol)

294 Appendix D

D5 Parameters and graphical representation of ternary chemical mixtures

Definition of parameters for the following Tables:

S2 pred square of the standard deviation. Method of derivation can be found in Appendix D1.

IC50pred derived from Equation 6-5 (Chapter 6) t*s calculation of 95% confidence interval based on student t test (df = 6).

X (chemical) ratio of IC50 of component chemical in the mixture

Graphical representations of the following ternary mixtures (MTS assay) are present in the text Section 6.5.4: SLS: NaF:Irgasan p.211

EtOH: CdCl2: LiSO4 p.208

KCN:HgCl2:CdCl2 p.212

295 Appendix D

Parameters and ternary graphical representation for the MTS assay

Ternary mixture: Cr (VI): CuSO4: phenol showing a synergistic interaction

2 X (CuSO4) X (CrVI) X (phenol) S (pred) t*s IC50 (pred) (*104) 0 0 1 4.00 489.4 1655 0 1 0 0.04 48.94 168.3 1 0 0 0.001 23.14 37.64 0.02 0.09 0.889 3.16 435.1 1487

antagonistic interaction (mg/L)

50 additive interaction IC exp

synergistic interaction

CrVI:CuSO4:phenol

Graphical representation of a synergistic interaction between ternary mixture

(CrVI: CuSO4: phenol)

296 Appendix D

Ternary mixture: Cr (VI): CuSO4: phenol showing an additive interaction

2 X (NaF) X (phenol) X (KCl) S (pred) t*s IC50 (pred) (*104) 0 0 1 2500 12235 17328 0 1 0 4.00 489.4 1655 1 0 0 0.90 233.1 1045 0.052 0.083 0.865 1870 10583 15180

antagonistic interaction

)

m n

( io ct 0 ra 5 e nt C i I exp e tiv di ad

synergistic interaction

NaF:phenol:KCl

Graphical representation of an additive interaction between ternary mixture (NaF: KCl: phenol)

297 Appendix D

Parameters and ternary graphical representation for the NRU assay

Ternary mixture: NaF: SLS: irgasan showing a synergistic interaction

2 X (SLS) X (NaF) X (irgasan) S (pred) t*s IC50 (pred) 0 0 1 1600 97.88 374.0 0 1 0 1936 107.6 295.0 1 0 0 59.00 18.79 41.9 0.059 0.415 0.526 776.3 68.17 321.6

antagonistic interaction

n tio ac ) r te

L in / e iv

g t di

5 C

I exp

synergistic interaction

SLS:NaF:Irgasan

Graphical representation of a synergistic interaction between ternary mixture (SLS: NaF: irgasan)

298 Appendix D

Ternary mixture: Cr (VI): CuSO4: phenol showing an antagonistic interaction

2 X (CuSO4) X (CrVI) X (phenol) S (pred) t*s IC50 (pred) 0 0 1 216.7 36.01 424.7 0 1 0 9216 234.9 214.0 1 0 0 3025 134.6 203.6

0.242 0.254 0.504 826.8 70.36 317.7

)

/ g

exp

(

C I antagonistic interaction

additive interaction

CuSO4:CrVI:phenol

Graphical representation of an antagonistic interaction between ternary mixture

(CuSO4: CrVI: phenol)

299 Appendix D

Ternary mixture: CdCl2: LiSO4: EtOH showing an additive interaction

2 X (CdCl2) X (LiSO4) X (EtOH) S (pred) t*s IC50 (pred) (*104) 0 0 1 361.0 4649 11772 0 1 0 9.00 734.1 1925 1 0 0 0.0002 3.54 5.65

0.0004 0.14 0.859 266.5 3995 10382

) L

/ antagonistic interaction

(

C I

n o i t c exp a r e t n i

e v i t i d d a

interaction synergistic

CdCl2:LiSO4:EtOH

Graphical representation of an additive interaction between ternary mixture

(LiSO4: CdCl2: EtOH)

300 Appendix D

Ternary mixture: HgCl2: KCN: CdCl2 showing an additive interaction

2 X (HgCl2) X (CdCl2) X (KCN) S (pred) t*s IC50 (pred) 0 0 1 2304 117.4 357.0 0 1 0 2.10 3.54 5.6 1 0 0 0.24 1.19 3.08 0.008 0.015 0.976 2194 114.6 348.5

antagonistic interaction

n o ti c exp ra te in e iv it d d

(mg/L) a 50 n IC tio rac inte istic erg syn

HgCl2:CdCl2:KCN

Graphical representation of an additive interaction between ternary mixture

(HgCl2: CdCl2: KCN)

301 Appendix D

Ternary mixture: KCl: NaF: phenol showing an additive interaction

2 X (NaF) X (phenol) X (KCl) S (pred) t*s IC50 (pred) (*104) 0 0 1 9.00 734.1 1925 0 1 0 0.02 36.01 424.7 1 0 0 0.19 107.7 288.8 0.109 0.161 0.73 4.79 536.1 1505

antagonistic interaction

exp (mg/L) 50

IC n tio ac er int ive dit ad

syn ergis tic i ntera ction

NaF:phenol:KCl

Graphical representation of an additive interaction between ternary mixture (KCl: NaF: phenol)

302