Toxicity of chemicals (Part 1), principles and methods for evaluating the (EHC 6, 1978)

INTERNATIONAL PROGRAMME ON CHEMICAL SAFETY

ENVIRONMENTAL HEALTH CRITERIA 6

PRINCIPLES AND METHODS FOR EVALUATING THE
TOXICITY OF CHEMICALS
PART I

This report contains the collective views of an international group of
experts and does not necessarily represent the decisions or the stated
policy of the United Nations Environment Programme, the International
Labour Organisation, or the World Health Organization.

Published under the joint sponsorship of the United Nations
Environment Programme, the International Labour Organisation, and the
World Health Organization

World Health Organization
Geneva, 1978

ISBN 92 4 154066 4

^(c) World Health Organization 1978

Publications of the World Health Organization enjoy copyright
protection in accordance with the provisions of Protocol 2 of the
Universal Copyright Convention. All rights reserved. The designations
employed and the presentation of the material in this publication do
not imply the expression of any opinion whatsoever on the part of the
Secretariat of the World Health Organization concerning the legal
status of any country, territory, city or area or of its authorities,
or concerning the delimitation of its frontiers or boundaries. The
mention of specific companies or of certain manufacturers' products
does not imply that they are endorsed or recommended by the World
Health Organization in preference to others of a similar nature that
are not mentioned. Errors and omissions excepted, the names of
proprietary products are distinguished by initial capital letters.

CONTENTS

PREFACE

REFERENCES

1. SOME GENERAL ASPECTS OF TOXICITY EVALUATION

1.1. Introduction
1.1.1. Defining toxicity, hazard, risk, and related terms
1.1.2. Laboratory testing
1.1.3. Toxicological field studies
1.1.4. Ecotoxicology
1.1.5. Priorities in the selection of chemicals for
testing
1.1.6. The extent of toxicity testing required
1.2. Dose-effect and dose-response relationship
1.2.1. Dose
1.2.2. Effect and response
1.2.3. Dose-effect and dose-response curves
1.2.4. Toxic effects due to a combination of chemicals
1.3. Interpretation of laboratory data
1.3.1. Distinction between adverse and nonadverse effects
1.3.2. Threshold: practical and theoretical considerations
1.3.3. Extrapolation of animal data to man
1.3.3.1 Species differences and related factors
1.3.3.2 Safety factors
1.3.3.3 Low-dose extrapolation
1.3.3.4 Other methods of extrapolation
1.4. Human data
1.4.1. Ethical considerations
1.4.2. Need for human investigations
1.5. The use of toxicological data in establishing environmental
health standards
1.5.1. Environmental health standards
1.5.2. Assessment of health risk and evaluation of
benefits
1.5.3. An example of toxicological information used in
standard setting
1.6. Limitations of safety evaluation
ANNEX
REFERENCES

2. FACTORS INFLUENCING THE DESIGN OF TOXICITY STUDIES
2.1. Introduction
2.2. Chemical and physical properties
2.2.1. General considerations
2.2.2. Physicochemical properties and the design of
toxicity studies
2.2.3. Impurities

2.3. Probable routes of exposure
2.3.1. General considerations
2.3.2. Specific variables related to route of exposure
2.3.2.1 Rate of absorption
2.3.2.2 Site of action
2.3.2.3 Biotransformation
2.3.2.4 Species
2.3.2.5 Unintended route
2.3.3. Special tests related to route
2.4. Selection and care of animals
2.4.1. General considerations
2.4.2. Animal variables
2.4.2.1 Selection of species
2.4.2.2 Animal models representing special
populations at risk
2.4.3. Cyclic variations in function or response
2.4.4. Environmental variables
2.4.4.1 Temperature
2.4.4.2 Caging
2.4.4.3 Diet and nutritional status
2.5. Statistical considerations
2.6. Nature of effects
2.6.1. Reversible and irreversible effects
2.6.2. Functional versus morphological changes
2.7. Dynamic aspects of predictive toxicology
2.7.1. Traditional versus new techniques
2.7.2. Toxicity of chemical analogues
2.7.3. Relation between site of metabolism and site of
injury
2.7.4. In vitro test systems
REFERENCES

3. ACUTE, SUBACUTE, AND CHRONIC TOXICITY TESTS
3.1. Introduction
3.2. General nature of test procedures
3.2.1. Housing, diet, and clinical examination of test
animals
3.3. Acute toxicity tests
3.3.1. Underlying principles
3.3.2. Experimental design
3.3.2.1 Selection of species
3.3.2.2 Selection of doses
3.3.2.3 Method of administration
3.3.2.4 Postmortem examination
3.3.3. Repeated high-dose studies

3.4. Subacute and chronic toxicity tests
3.4.1. Underlying principles
3.4.2. Experimental design
3.4.2.1 Selection of species and duration of
studies
3.4.2.2 Selection of doses
3.4.2.3 Method of administration
3.4.2.4 Biochemical organ function tests
3.4.2.5 Physiological measurements
3.4.2.6 Metabolic studies
3.4.2.7 Haematological information
3.4.2.8 Postmortem examination
3.4.2.9 Controls
3.4.3. Alternative approaches in chronic toxicity
3.4.3.1 Perinatal exposure
3.4.3.2 Use of nonrodent species
3.5. Evaluation and interpretation of the results of toxicity
tests
REFERENCES

4. CHEMOBIOKINETICS AND METABOLISM
4.1. Introduction
4.2. Absorption
4.2.1. General principles
4.2.2. Absorption from the lungs
4.2.3. Absorption from the skin
4.2.4. Gastrointestinal absorption
4.3. Distribution
4.4. Binding
4.4.1. Plasma-protein binding
4.4.2. Tissue binding
4.5. Excretion
4.5.1. Renal excretion
4.5.2. Biliary excretion
4.5.3. Enterohepatic circulation
4.5.4. Other routes of excretion
4.6. Metabolic transformation
4.6.1. Mechanism of metabolic transformation
4.6.1.1 Microsomal, mixed-function oxidations
4.6.1.2 Conjugation reactions
4.6.1.3 Extramicrosomal metabolic transformations
4.6.1.4 Nonenzymatic reactions
4.6.2. Species variability
4.6.3. Enzyme induction and inhibition
4.6.4. Metabolic saturation
4.7. Experimental design
4.8. Chemobiokinetics
4.8.1. One-compartment open model
4.8.2. Two compartment/multicompartment open systems
4.8.3. Repeated administration or repeated exposure
4.8.4. Kinetics of nonlinear or saturable systems

4.9. Linear and nonlinear one compartment open-model kinetics of
2,4,5-trichloro-phenoxyacetic acid (2,4,5-T)
4.10. Linear chemobiokinetics used to assess potential for
bioaccumulation of 2,3,6,7-tetrachlorodibenzo-p-dioxin
(TCDD)
ANNEX
REFERENCES

5. MORPHOLOGICAL STUDIES
5.1. Introduction
5.2. General recommendations
5.3. Gross observations
5.3.1. Autopsy techniques
5.3.2. Rat, mouse, guineapig, rabbit, monkey
5.3.3. Carnivores, swine
5.4. Selection, preservation, preparation, and storage of
tissues
5.4.1. Selection of tissues
5.4.2. Oral toxicity tests
5.4.3. Inhalation toxicity studies
5.4.4. Dermal toxicity studies
5.4.5. Special studies
5.5. Preservation of tissues
5.5.1. Immersion
5.5.2. Inflation
5.5.3. Perfusion
5.6. Trimming
5.7. Storage
5.8. Histological techniques
5.9. Special techniques
5.9.1. Enzyme histochemistry
5.9.2. Autoradiography
5.9.3. Immunofluorescence and immunoenzyme techniques
5.9.4. Electron microscopy
5.10. Microscopic examination
5.10.1. Number of animals and number of organs and tissues
studied microscopically
5.10.2. Description of the lesions
5.11. Presentation, evaluation, and interpretation of
pathological data
REFERENCES

6. INHALATION EXPOSURE
6.1. Introduction
6.2. Need for inhalation studies
6.3. Fate of inhaled materials
6.3.1. Nature of aerosols
6.3.2. Deposition
6.3.3. Clearance
6.4. Dose in inhalation studies

6.5. Choice of species
6.5.1. Anatomical differences
6.5.2. Physiological considerations
6.5.3. Disease and susceptibility states
6.6. Duration of exposure
6.6.1. Intermittent versus continuous exposure
6.7. Inhalation systems
6.7.1. Facilities required
6.7.2. Static systems
6.7.3. Dynamic systems
6.7.4. Typical whole-body systems
6.7.5. Construction materials
6.7.6. Engineering requirements
6.7.7. Special systems
6.7.7.1 Isolation units
6.7.7.2 Head and nose exposures
6.7.7.3 Instantaneous exposure systems
6.7.8. Variables to monitor
6.7.9. Human exposure facilities
6.8. Contaminant generation and characterization
6.8.1. Generation of vapours
6.8.2. Particle generators
6.8.2.1 Heterogeneous aerosols
6.8.3. Monitoring contaminant concentrations
6.8.3.1 Vapour sampling
6.8.3.2 Particulate sampling
6.9. Other methods of respiratory tract exposure
6.9.1. In vivo exposures
6.9.2. In vitro exposures
6.10. Biological end-points and interpretation of changes in
these end-points
6.10.1. Morphological changes
6.10.2. Functional changes
6.10.2.1 Measurement of respiratory frequency
6.10.2.2 Measurement of mechanics of respiration
6.10.3. Biochemical end-points
6.10.4. Other end-points in inhalation studies
REFERENCES

7. CARCINOGENICITY AND MUTAGENICITY
7.1. Introduction
7.2. Carcinogenicity
7.2.1. Long-term bioassays
7.2.1.1 Species, strain, and sex selection, and
size of groups
7.2.1.2 Route of administration
7.2.1.3 Inception and duration of tests
7.2.1.4 Dose-level and frequency of exposure

7.2.1.5 Combined treatment and cocarcinogenesis
7.2.1.6 Positive and untreated controls
7.2.1.7 Test material
7.2.1.8 Survey of animals, necropsy, and
histological examination
7.2.2. Short-term tests (rapid screening tests)
7.2.2.1 Metabolic activation, reaction with DNA,
and DNA repair
7.2.2.2 In vitro neoplastic transformation of
mammalian cells
7.2.2.3 Mutagenicity tests
7.2.2.4 Submammalian assay systems
7.2.2.5 Mammalian somatic cells
7.2.2.6 Host and tissue-(microsome) mediated
assays
7.2.3. Correlation between short- and long-term bioassays
for carcinogenicity
7.2.4. Significance of experimental testing for assessing
the possible carcinogenic risk of chemicals to man
7.3. Heritable mutations
7.3.1. Whole-animal tests
7.3.2. Monitoring of human populations
7.3.3. Significance of tests for heritable mutations
REFERENCES

NOTE TO READERS OF THE CRITERIA DOCUMENTS

While every effort has been made to present information in the
criteria documents as accurately as possible without unduly delaying
their publication, mistakes might have occurred and are likely to
occur in the future. In the interest of all users of the environmental
health criteria documents, readers are kindly requested to communicate
any errors found to the Division of Environmental Health, World Health
Organization, Geneva, Switzerland, in order that they may be included
in corrigenda which will appear in subsequent volumes.

In addition, experts in any particular field dealt with in the
criteria documents are kindly requested to make available to the WHO
Secretariat any important published information that may have
inadvertently been omitted and which may change the evaluation of
health risks from exposure to the environmental agent under
examination, so that the information may be considered in the event of
updating and re-evaluation of the conclusions contained in the
criteria documents.

PREFACE

The use of chemicals in practically every aspect of life has
grown very rapidly over the last few decades and international trade
in bulk chemicals, specialty chemicals, and consumer products has
increased proportionately, making imperative the need for continuous
review and reappraisal of procedures for evaluating their safety.
Concern about the possible health hazards that may arise from exposure
to chemicals has increased throughout the world, especially in the
industrialized countries. In many WHO Member States, this has resulted
in new laws and regulations which, in turn, have created a need to
assemble, analyse, and evaluate all available toxicological
information with a view to assessing hazard. Toxicologists have
responded by developing techniques for safety evaluation but these
often differ from one country to another. The differences are
sometimes slight, sometimes considerable; on occasion they have led to
unfortunate misunderstandings, and often to needless duplication of
work.

Ever since the World Health Organization started programmes on
food safety and drug evaluation, the need for some degree of
uniformity and for generally accepted principles and requirements for
toxicological testing and evaluation has been recognized. This has
resulted, in the last 20 years, in a number of technical reports and
guidelines on such topics as the general principles and methods for
the testing and evaluation of intentional and unintentional food
additives (WHO, 1957, 1958, 1967a, 1974a) and drugs (WHO, 1966, 1968,
1975a), on the evaluation of teratogenicity (WHO, 1967b),
mutagenicity, and carcinogenicity (WHO, 1961, 1969, 1971, 1974b,
1976a) and, more recently, on environmental and health monitoring and
the early detection of health impairment in occupational health (WHO,
1973, 1975b), on chemical and biochemical methodology for assessing
the hazards of pesticides to man (WHO, 1975c), and on the methods used
in establishing permissible levels of occupational exposure to harmful
agents (WHO, 1977). Several symposia have also been organized, to
discuss, for example, the methods used in the USSR for establishing
biologically safe levels of toxic substances (WHO, 1975d, 1975e), and
screening tests in chemical carcinogenesis (IARC, 1976). All these
publications remain a most useful source of information on selected
aspects of toxicological evaluation.

The need for more uniformity in methods of environmental health
risk evaluation was again raised at the 1973 World Health Assembly, in
resolution WHA26.58 on human health and the environment which inter
alia requested the Director-General to develop protocols for
experimental and epidemiological studies, uniform terminology, and
agreed definitions. Harmonization of toxicological and epidemiological
methods is also one of the objectives of the WHO Environmental Health
Criteria Programme (WHO, 1976b), initiated in 1973 in collaboration

with Member States and the United Nations Environment Programme
(UNEP), while a very recent (1977) World Health Assembly resolution
WHA30.47 requested the Director-General "to examine the possible
options for international cooperation with a view to accelerating and
making more effective the evaluation of health risks from exposure to
chemicals, and promoting the use of experimental and epidemiological
methods that will produce internationally comparable results".

Current concern about the health effects of chemicals is more
intense in some countries than in others, with the consequent
unevenness in political response reflected in variations in national
safety regulations. This situation is likely to continue for some
time. It is unrealistic and perhaps not really desirable at present,
to seek international standardization in safety testing and evaluation
as this might hinder the input of new ideas and the development of
improved methods and might lead either to the application of needless
tests or to failure to ask the essential questions. However, it is not
too early for scientists and decision-makers to try to understand the
similarities and differences in the safety evaluations made in
different countries. The underlying objectives are the same
everywhere, namely, to minimize harm and maximize safety and yet not
impede the beneficial use of chemicals. Similarly, the basic
scientific principles are globally accepted, so there is no reason why
there should not be a gradual harmonization of methods and procedures
for toxicological testing and evaluation.

With these views in mind and taking into account past work of
WHO, an attempt was made to set forth, comprehensively and on an
international basis, the principles and procedures for the safety
evaluation of all types of chemicals. More than 50 distinguished
experts from some 11 countries collaborated with the Organization and,
in a series of meetings and individual consultations, planned,
drafted, and revised this compilation of toxicological procedures,
providing at the same time an excellent example of international
cooperation. In addition, there was valuable support for the project
from the WHO collaborating centres at: the Institute of Hygiene and
Occupational Health, Sofia, Bulgaria; the National Institute of Public
Health, Bilthoven, Netherlands; the Department of Environmental
Hygiene, The Karolinska Institute, Stockholm, Sweden; the National
Institute of Environmental Health Sciences, Research Triangle Park,
North Carolina, USA; and the Sysin Institute of General and Communal
Hygiene, Moscow, USSR.

The general approach in preparing this publication has been to
present the underlying scientific principles, to evaluate the utility,
strengths and weaknesses of various methods and procedures, to help
the reader select the most suitable technique for a specific purpose
(bearing in mind that circumstances will often dictate the most
appropriate procedure) but not, as already mentioned, to prescribe

standard tests. While aiming at agreement on purely scientific issues,
it has not been sought on details of procedure, on the interpretation
of results, or on methods for setting environmental health standards.
Indeed, because there were often differences of opinion on these
matters, the solution adopted has been to present the different
viewpoints and interpretations. This explains a certain unevenness in
the text, particularly in those chapters prepared jointly by many
scientists from different countries.

Although an effort has been made to avoid inconsistency in
terminology, uniformity has not been possible; indeed, this is
something beyond the scope of the present monograph. However, WHO and
UNEP recently initiated another project that aims at internationally
agreed definitions for those terms most frequently used in
toxicological evaluation. Until this project is completed, it is
important to understand that some terms may have various meanings and
implications in different countries or in different scientific circles
and that it may be highly misleading to employ them outside the
national pattern of use or outside the context of a specialized field,
without precise definition. The reader is therefore warned to be wary
of the uncritical transfer of technical terms from one set of
circumstances to another.

Toxicology is a rapidly developing field, especially at this
time; it is hoped, nevertheless, that this monograph provides a valid
account of the present state of knowledge on the toxicity testing and
evaluation of chemicals as practiced by some of the leading experts in
the field. If it should also stimulate the exchange of knowledge and
experience and so contribute to greater efficiency and reliability in
toxicity testing and evaluation, it will have more than fulfilled its
purpose.

The work has been divided into two separate publications. The
first part contains the broad principles and more general aspects of
toxicity testing, the planning and evaluation of acute, subacute, and
chronic toxicity tests, chemobiokinetics and metabolism, morphological
tests, inhalation studies, and tests for carcinogenicity and
mutagenicity. Part 2 systematically covers some more specialized
procedures for safety evaluation, i.e. functional studies of organs
and systems, effects on reproduction, neurological and behavioural
studies, effects on the skin and the eye, cumulation and adaptation,
and finally discusses factors that could modify the outcome of
toxicity testing and evaluation.

The main authors mutually reviewed the chapters of the treatise,
which can therefore be considered to be a synthesis of various views
and opinions, but this does not detract from the merit of their own
contributions which are gratefully acknowledged. The WHO Secretariat
at the Meeting of the Main Authors in Geneva (28 July to 1 August
1975)^a and at the Scientific Group in Lyons (1 to 5 December 1975)^b
comprised: Dr M. El Batawi, Chief, Occupational Health^a;
Dr H. Bartsch, Unit of Chemical Carcinogenesis, IARC, Lyons^b:
Dr J. F. Copplestone, Vector Biology and Control^b; Dr F. C. Lu,
Chief, Food Additives^b; Dr R. Montesano, Unit of Chemical
Carcinogenesis, IARC, Lyons^a,b; Dr H. Nakajima, Drug Evaluation and
Monitoring^a; Dr M. Vandekar, Vector Biology and Control^a; and
Dr G. Vettorazzi, Food Additives^a. Dr V. B. Vouk, Chief, Control of
Environmental Pollution and Hazards was the Secretary of the Geneva
meeting, while Dr L. Tomatis, Chief, Unit of Chemical Carcinogenesis,
IARC, Lyons, and Dr Vouk were the Joint Secretaries of the Scientific
Group at Lyons. Representatives of other organizations who were
present at the meetings include: Dr M. Marcus (US Environmental
Protection Agency)^a; Dr W. J. Hunter (Commission of the European
Communities)^b; Mr C. Prior (Organization for Economic Cooperation
and Development)^b; Dr V. Smirnyagin (International Council of
Scientific Unions)^b. Miss S. Braman, Technical Assistant, Control of
Environmental Pollution and Hazards, serviced the two meetings and
helped throughout with the preparation of the manuscript.

The final editing was carried out by a group headed by Professor
N. Nelson who, indeed, presided over the whole project and to whom
special thanks are due for, without his ideas, enthusiasm and, above
all, profound knowledge of the subject, there would have been no
treatise.

^a Participated in the Meeting of Main Authors, Geneva, 28 July to
1 August 1975.

^b Participated in the Scientific Group on Methods of Toxicity
Evaluation of Chemicals, Lyons, 1-5 December 1975.

REFERENCES

IARC (1976) Screening tests in chemical carcinogenesis -- Proceedings
of a Workshop organized by IARC and the CEC, Brussels 1975.
IARC Sci. Publ. No. 12.

WHO (1957) WHO Technical Report Series No. 129 (General principles
governing the use of food additives: First report of the Joint
FAO/WHO Expert Committee on Food Additives.) 22 pp.

WHO (1958) WHO Technical Report Series No. 144 (Procedures for the
testing of intentional food additives to establish their safety
for use: Second report of the Joint FAO/WHO Expert Committee on
Food Additives.) 19 pp.

WHO (1961) WHO Technical Report Series No. 220 (Evaluation of the
carcinogenic hazards of food additives: Fifth report of the Joint
FAO/WHO Expert Committee on Food Additives.) 33 pp.

WHO (1966) WHO Technical Report Series No. 341 (Principles for
pre-clinical testing of drug safety: Report of a WHO Scientific
Group.) 22 pp.

WHO (1967a) WHO Technical Report Series No. 348 (Procedures for
investigating intentional and unintentional food additives:
Report of a WHO Scientific Group.) 25 pp.

WHO (1967b) WHO Technical Report Series No. 364 (Principles for the
testing of drugs for teratogenicity: Report of a WHO Scientific
Group.) 18 pp.

WHO (1968) WHO Technical Report Series No. 403 (Principles for the
clinical evaluation of drugs: Report of a WHO Scientific Group.)
32 pp.

WHO (1969) WHO Technical Report Series No. 426 (Principles for the
testing and evaluation of drugs for carcinogenicity: Report of a
WHO Scientific Group.) 26 pp.

WHO (1971) WHO Technical Report Series No. 482 (Evaluation and testing
of drugs for mutagenicity: principles and problems -- Report of a
WHO Scientific Group.) 18 pp.

WHO (1973) WHO Technical Report Series No. 535 (Environmental and
health monitoring in occupational health: Report of a WHO Expert
Committee.) 48 pp.

WHO (1974a) WHO Technical Report Series No. 539 (Toxicological
evaluation of certain food additives with a review of general
principles and of specifications: Seventeenth report of the Joint
FAO/WHO Expert Committee on Food Additives.) 40 pp.

WHO (1974b) WHO Technical Report Series No. 546 (Assessment of the
carcinogenicity and mutagenicity of chemicals: Report of a WHO
Scientific Group.) 19 pp.

WHO (1975a) WHO Technical Report Series No. 563 (Guidelines for
evaluation of drugs for use in man: Report of a WHO Scientific
Group.) 59 pp.

WHO (1975b) WHO Technical Report Series No. 571 (Early detection of
health impairment in occupational exposure to health hazards:
Report of a WHO Study Group.) 80 pp.

WHO (1975c) WHO Technical Report Series No. 560 (Chemical and
biochemical methodology for the assessment of hazards of
pesticides for man.) 26 pp.

WHO (1975d) Methods used in the USSR for establishing biologically
safe levels of toxic substances. Geneva, WHO, 171 pp.

WHO (1975e) Methods for studying biological effects of pollutants
(A review of methods used in the USSR). Copenhagen, WHO
Regional Office for Europe, 80 pp. (EURO publication 3109(4).)

WHO (1976a) WHO Technical Report Series No. 586 (Health hazards from
new environmental pollutants: Report of a WHO Study Group.)
96 pp.

WHO (1976b) Background and purpose of the WHO Environmental Health
Criteria Programme. (Reprint from Environmental Health
Criteria 1 Mercury.) Geneva, WHO, 9 pp.

WHO (1977) WHO Technical Report Series No. 601 (Methods used in
establishing permissible levels in occupational exposure to
harmful agents: Report of a WHO Expert Committee with the
participation of ILO.) 68 pp.

PRINCIPLES AND METHODS FOR THE TOXICITY EVALUATION OF CHEMICALS

Editorial Group

Dr F. A. Fairweather, Department of Health & Social Security, London,
England

Professor F. Kaloyanova, Institute of Hygiene & Occupational Health,
Sofia, Bulgaria

Dr G. N. Krasovskij, Laboratory of Water Toxicology, A. N. Sysin
Institute of General & Communal Hygiene, Moscow, USSR

Dr R. Kroes, Central Institute for Nutrition & Food Research, Zeist,
Netherlands

Dr R. Montesano, Unit of Chemical Carcinogenesis, International Agency
for Research on Cancer, Lyons, France

Professor S. D. Murphy, Division of Toxicology, Department of
Pharmacology, The University of Texas Health Sciences Center,
Houston, TX, USA

Professor N. Nelson, Institute of Environmental Medicine, New York
University, NY, USA (Chairman)

Professor D. V. Parke, Department of Biochemistry, University of
Surrey, Guildford, England

Professor I. V. Sanockij, Department of Toxicology, Institute of
Industrial Hygiene & Occupational Diseases, Moscow, USSR

Dr I. P. Ulanova, Department of Toxicology, Institute of Industrial
Hygiene & Occupational Diseases, Moscow, USSR

Dr V. B. Vouk, Control of Environmental Pollution and Hazards,
Division of Environmental Health, World Health Organization,
Geneva, Switzerland (Secretary)

Professor J. G. Wilson, Department of Pediatrics, Children's Hospital
Medical Center, University of Cincinnati, Cincinnati, OH, USA

PRINCIPLES AND METHODS FOR THE TOXICITY EVALUATION OF CHEMICALS

Contributors to Part 1

^b Dr H. Bartsch, Unit of Chemical Carcinogenesis, International
Agency for Research on Cancer, Lyons, France (Chapter 7)

Dr S. M. Charbonneau, Toxicology Research Division, Health Protection
Branch, National Department of Health & Welfare, Ottawa, Canada
(Chapter 3)

^a Dr R. T. Drew, Medical Department, Brookhaven National Laboratory,
Upton, NY, USA (Chapter 6)

^a Dr H. L. Falk, National Institute of Environmental Health
Sciences, Research Triangle Park, NC, USA (Chapter 2)

Dr V. J. Feron, Central Institute for Food Research, Zeist,
Netherlands (Chapter 5)

^a Dr P. Gehring, Toxicology Research Laboratory, Dow Chemical USA,
Midland, MI, USA (Chapter 4)

^b Dr H. C. Grice, Toxicology Research Division, Health Protection
Branch, Department of National Health & Welfare, Ottawa, Canada

^a,b Professor F. Kaloyanova, Institute of Hygiene & Occupational
Health, Sofia, Bulgaria

^a Dr G. N. Krasovskij, Laboratory of Water Toxicology, A. N. Sysin
Institute of General & Communal Hygiene, Moscow, USSR (Chapter 1)

^a,b Dr R. Kroes, Central Institute for Nutrition & Food Research,
Zeist, Netherlands (Chapters 1 & 5)

Dr J. E. LeBeau, Toxicology Research Laboratory, Dow Chemical USA,
Midland, MI, USA (Chapter 4)

^a,b Dr S. Manyai, Biochemical Department, Institute of Occupational
Health, Budapest, Hungary (Chapter 4)

^a,b Dr R. Montesano, Unit of Chemical Carcinogenesis, International
Agency for Research on Cancer, Lyons, France (Chapters 1 & 7)

^a Dr I. C. Munro, Toxicology Research Division, Health Protection
Branch, Department of National Health & Welfare, Ottawa, Canada
(Chapters 1, 2 & 3)

^a Professor S. D. Murphy, Division of Toxicology, Department of
Pharmacology, The University of Texas Health Sciences Center,
Houston, TX, USA (Chapters 1 & 2)

^a,b Professor N. Nelson, Institute of Environmental Medicine, New
York University, NY, USA (Chapters 1 & 7)

^b Professor G. Nordberg, Institute of Hygiene & Social Medicine,
Odense University, Odense, Denmark

^a,b Professor D. V. Parke, Department of Biochemistry, University of
Surrey, Guildford, England (Chapters 1 & 2)

^a,b Dr E. A. Pfitzer, Department of Toxicology, Research Division,
Hoffman-La Roche Inc., Nutley, NJ, USA (Chapters 1 & 2)

^a Dr M. A. Pinigin, A. N. Sysin Institute of General & Communal
Hygiene, Moscow, USSR (Chapter 1)

Dr J. C. Ramsey, Toxicology Research Laboratory, Dow Chemical USA,
Midland, MI, USA (Chapter 4)

^b Professor I. V. Sanockij, Department of Toxicology, Institute of
Industrial Hygiene & Occupational Diseases, Moscow, USSR
(Chapters 1 & 2)

Dr K. K. Sidorov, Department of Toxicology, Institute of Industrial
Hygiene & Occupational Diseases, Moscow, USSR (Chapter 6)

^b Dr L. Tomatis, Unit of Chemical Carcinogenesis, International
Agency for Research on Cancer, Lyons, France (Chapter 7)

^b Professeur R. Truhaut, Laboratoire de Toxicologie et d'Hygiène
industrielles, Faculté des Sciences pharmaceutiques et
biologiques. Université René Descartes, Paris, France

^a Dr I. P. Ulanova, Department of Toxicology, Institute of
Industrial Hygiene & Occupational Diseases, Moscow, USSR
(Chapter 6)

^a,b Dr V. B. Vouk, Control of Environmental Pollution and Hazards,
Division of Environmental Health, WHO, Geneva, Switzerland
(Chapter 1)

Dr Z. Zawidski, Toxicology Research Division, Health Protection
Branch, Department of National Health & Welfare, Ottawa, Canada
(Chapter 3)

^a Participated in the Meeting of Main Authors, Geneva, 28 July to
1 August 1975.
^b Participated in the Scientific Group on Methods of Toxicity
Evaluation of Chemicals, Lyons, 1-5 December 1975.

1. SOME GENERAL ASPECTS OF TOXICITY EVALUATION

1.1 Introduction

Toxicology is concerned both with the nature and mechanisms of
toxic lesions and the quantitative evaluation of the spectrum of
biological changes produced by exposure to chemicals. Every chemical
is toxic under certain conditions of exposure. An important corollary
is that for every chemical there should be some exposure condition
that is safe as regards man's health (Lazarev, 1938; Pravdin, 1934;
Smyth, 1963; Weil, 1972a) with the possible exception of chemical
carcinogens and mutagens (WHO, 1974a).

The quantitative evaluation of the biological changes caused by
chemicals aims at the establishment of dose-effect and dose-response
relationships that are of fundamental importance for health risk
evaluation.

1.1.1 Defining toxicity, hazard, risk, and related terms

In a general sense, the toxicity of a substance could be defined
as the capacity to cause injury to a living organism (NAS/NRC, 1970;
Sanockij, 1970). A highly toxic substance will damage an organism if
administered in very small amounts; a substance of low toxicity will
not produce an effect unless the amount is very large. Thus, toxicity
cannot be defined without reference to the quantity of a substance
administered or absorbed (dose), the way in which this quantity is
administered (e.g. inhalation, ingestion, injection) and distributed
in time (e.g. single dose, repeated doses), the type and severity of
injury, and the time needed to produce that injury.

There is no generally agreed definition of "hazard" associated
with a chemical, but the term is used to indicate the likelihood that
a chemical will cause an adverse health effect (injury) under the
conditions in which it is produced or used (Goldwater, 1968; NAS/NRC,
1970, Pravdin, 1934).

Risk is a statistical concept and has been defined by the
Preparatory Committee of the United Nations Conference on the Human
Environment^a, as the expected frequency of undesirable effects
arising from exposure to a pollutant. Estimates of risk may be
expressed in absolute terms or in relative terms. The absolute risk is
the excess risk due to exposure. The relative risk is the ratio
between the risk in the exposed population and the risk in the
unexposed population (BEIR, 1972; ICRP, 1966).

^a Preparatory Committee of the United Nations Conference on the
Human Environment, Third Session, 13-24 September 1971
(A/Conf. 4818, pp. 45 & 46).

Safety is a term that has been used extensively but is difficult
to define. One definition is that "safety" is the practical certainty
that injury will not result from the substance when used in the
quantity and in the manner proposed for its use (NAS/NRC, 1970). This
definition is of little use unless "practical certainty" is defined in
some way, for example, in terms of a numerically specified low risk.
Another view is that "safety" should be judged in terms of socially
"acceptable" risks. Such judgments are largely outside the scope of
scientific evaluation but nevertheless require assessment both of the
probabilities^a of various adverse effects and of their severity in
terms of human health or other concerns (NAS, 1975).

1.1.2 Laboratory testing

Human data on the toxicity of chemicals are obviously more
relevant to safety evaluation than those obtained from the exposure of
experimental animals (see section 1.4). However, controlled exposures
of man to hazardous or potentially hazardous substances are limited by
ethical considerations and information obtained by clinical or
epidemiological methods must be relied on. Where such information is
not available, as in the case of all new synthetic chemicals, data
must be obtained from tests on experimental animals and other
laboratory procedures. The degree of confidence with which human
health risks can be estimated from laboratory data depends on the
quality of the data, and the selection of appropriate laboratory
testing procedures is the main subject of this monograph.

1.1.3 Toxicological field studies

In the laboratory, only a small number of animal species are
available for testing. The testing of wild species, living in cages
under field conditions, may be useful but sometimes presents a variety
of problems. Successful trials require a large enough site (about
8 ha; 20 acres) with adequate and varied populations of birds,
mammals, fish, insects, and other species, and the area studied must
be considerably greater than that treated (Brown & Papworth, 1974).
Data obtained from field trials of chemicals are of considerable value
in supplementing data obtained with laboratory animal species and in
validating the projection of experimental results to the ecosystem,
including man. Studies of random events in natural ecosystems can also
provide useful data.

^a i.e. the expected frequencies.

Sensitive analytical techniques now make it relatively simple to
conduct field studies in man by monitoring levels of a chemical or its
metabolites in blood, urine, hair, or saliva; this biological
monitoring together with environmental monitoring provides important
information on the exposure of man^a. Regular periodic determination
of the profile of certain plasma enzymes and other biochemical
variables in the subject provides another valuable method for
monitoring health effects particularly under occupational exposure
conditions (WHO, 1973, 1975a, 1975b); changes in these profiles may
provide early warning of damage by toxic chemicals (Cuthbert, 1974).

1.1.4 Ecotoxicology

A new subdivision of toxicology, "ecotoxicology", has emerged
following observations that some persistent chemicals can exert toxic
effects at several points in an ecosystem. The appearance of a
chemical or the manifestation of a toxic effect may occur far away
from its initial point of introduction into the environment. Methods
for assessing the extent and significance of the movement of
pollutants and their degradation products through the environment to
target systems are discussed in a recent publication (NAS, 1975).

1.1.5 Priorities in the selection of chemicals for testing

In principle, all new chemicals require safety evaluation before
manufacture and sale, but, because of the large number of chemicals
that represent a possible hazard to human health and limited
resources, it is necessary to give priority to those that are directly
consumed by man, such as drugs and food additives, and those that are
widely used such as pesticides or household consumer products.
Industrial chemicals that can escape into the working or general
environment or can contaminate other products are another category of
concern.

Compounds of suspected high acute, chronic, or delayed toxicity
(such as carcinogenicity) or of high persistence in the environment,
or compounds which contain chemical groups known to be associated with
these properties, deserve the highest priority. This also applies to
compounds known to inhibit metabolic deactivation of chemicals as they
may represent a more insidious form of toxicity.

^a Report of the Meeting of a Government Expert Group on Health
Related Monitoring. Unpublished WHO document CEP/77.6.

Chemicals resistant to metabolism, especially metabolism by
microflora, will have a high environmental persistence. Many
halogenated compounds come into this category, and should, therefore,
have some degree of priority. Compounds that accumulate in food chains
or are stored in the body, e.g. methylmercury and DDT, will be a
matter of concern. Such compounds are often highly lipid-soluble or
strongly bound to tissue proteins, or may undergo enterohepatic
recirculation with consequent slow excretion resulting in accumulation
in the organism.

Physicochemical properties can be an important consideration in
setting priorities for testing potential environmental pollutants. For
example, biomagnification of stable, fat-soluble substances may lead
to contamination of human food supplies as well as to adverse effects
in wildlife at the higher levels of food chains, even though the
intended use and sites of application of the substance would suggest
that primary exposure of these species is unlikely (Edwards, 1970).
Physicochemical properties such as vapour pressure, and particle size
and density are important in predicting the atmospheric transport of
chemicals (Fuchs, 1964; OECD, 1977). Adsorption of a chemical on soil
particles may increase the likelihood that the material will become
airborne or be transported by watercourses and subsequently deposited
in areas remote from its site of application (Cohen & Pinkerton,
1966), or it may retard the movement of a chemical through ground
water and thus reduce the likelihood of contamination of ground water
supplies near the site of application (Edwards, 1970; Hamaker et al.,
1966).

Even though certain predictions and comparisons of environmental
distribution and biomagnification of chemicals in the environment may
be made theoretically on the basis of the physicochemical properties
of the substances in question, more definitive information of this
nature can be obtained experimentally by the use of model ecosystems
such as those described by Metcalf et al. (1971) and Lu & Metcalf
(1975). These model ecosystems may be oversimplified, and they should
not replace experimental field studies or programmes for monitoring
environmental contaminants. However, their use in an early phase of
the overall evaluation of the toxicity of environmental chemicals may:
( a) help to determine order of priority of chemicals for study,
( b) identify the components of the environment (food, water, air)
most likely to be a source of human exposure and ( c) suggest whether
the chemicals are likely to accumulate in human tissues.
Furthermore, the systematic application of such model systems to
structure-distribution studies may help in determining with greater
certainty those physicochemical properties of substances that are most
useful in predicting the distribution and effects of chemicals in
ecosystems (Lu & Metcalf, 1975).

Information on production, use, and disposal are of great
importance in determining the sources and quantities of a chemical
released into the environment, in assessing the possible extent of
human exposure, and in identifying human populations that are likely
to be exposed.

In conclusion, essential criteria for priority in the selection
of chemicals for testing are: (a) indication or suspicion of hazard to
human health and type and severity of potential health effects; (b)
probable extent of production and use; (c) potential for persistence
in the environment; (d) potential for accumulation in biota and in the
environment, and (e) type and size of populations likely to be
exposed. A chemical of first priority for testing would rate highly
with respect to all or most of these criteria.

1.1.6 The extent of toxicity testing required

The extent of the toxicity testing required will depend on a
variety of considerations, and generally valid procedures cannot be
proposed. One scheme, proposed by Sanockij (1975a), for chemicals that
are being developed, is shown in Table 1.1. As a first step, it may be
useful to make an approximate estimation of toxicity based on the
chemical structure and the physical and chemical properties of the
substance, and on known correlations of these variables with
biological activity (Andreyeshcheva, 1976; WHO, 1976a). These
considerations may be of value for decisions on safety measures to be
taken during initial laboratory work. Extrapolation and interpolation
in homologous series may also be of value for decisions on safety
measures to be taken during initial laboratory work (Ljublina &
Miheev, 1974), but for some series of chemicals this is not
applicable.

A preliminary evaluation of toxicity should start when chemicals
are synthesized in the laboratory stage of the development of an
industrial process. The full evaluation of the chemicals involved,
both in respect to occupational and general population exposure, and
assessment of possible air, water, and food contamination, should be
initiated later, when it has been decided to proceed with full-scale
production of the chemical. Toxicity data obtained during the
development stages of a technological process could provide
information concerning the health hazards not only of the raw
materials and products, but also of the various other substances used
or produced as intermediates in the technological process, and of
gaseous and other wastes. Toxicological evaluation may also help in
the selection of an alternative technological process, less hazardous
to health.

Waste disposal by dispersion in air and water, the ease of
environmental degradation of the chemical, and the toxicity of the
degradation products, are other problems that need attention at an
early stage in the toxicological evaluation of new chemicals. For
example, resistance to degradation has to be taken into account when
formulating health criteria regulating the application and disposal of
pesticides (Medved & Spynu, 1970).

This phasing of toxicological studies may be useful in
coordinating testing at national and international levels.

Environmental and health standards will need to be defined
preferentially for those chemicals that show a significant degree of
toxicity and represent a health hazard, and are likely to be used
widely in industry, agriculture, or in consumer products.

Changes and developments in industrial processes, the development
of new chemicals, and changes in the use of existing chemicals, may
lead to new or increased hazards. This calls for a continuous
re-evaluation of priorities.

1.2 Dose-Effect and Dose-Response Relationships

1.2.1 Dose

Most commonly, the term "dose" is used to specify the amount of
chemical administered, usually expressed per unit body weight. If the
dose is administered into the stomach, on the skin, or into the
respiratory tract, transport across the membranes may be incomplete
and the absorbed dose will not be identical with the dose
administered. In environmental exposures, an estimate of the dose can
be made from the measurement of environmental and food concentrations
as a function of time, and involves the assessment of food intake,
inhalation rate, and the appropriate deposition and retention factors.

The doses in the organs and tissues of interest may be estimated
from:

(a) administered dose or intake;

(b) measurement of the concentrations in tissues and organ
samples;

Table 1.1 The extent of toxicological evaluation required in relation to technological process development

Stages of technological Stages of toxicological Toxicological studies
development evaluation

1. Theoretical concept Preliminary toxicological Analysis of literature data on toxicity and
and process flow assessment hazards of raw materials, reagents,
diagram catalysers, semiproducts and additives

Assessment of toxicological parameters on
the basis of metabolic analogies, persistence,
the relationship between chemical
structure, chemical and physical properties.
and biological activity. Interpolation and
extrapolation in homologous series

2. Laboratory development Acute toxicity Acute and subacute experiments on
of the technological animals. Toxicological evaluation of
process technological unit processes

3. Pilot plant stage Subacute toxicity Subacute toxicity experiments on animals.
Studies of delayed effects. Medical
examination of workers.

Detailed toxicological Chronic toxicity studies and, when indicated,
evaluation effects on reproduction, carcinogenicity,
mutagenicity. Formulation of medical and
industrial hygiene requirements for
full-scale production

Table 1.1 (cont'd).

Stages of technological Stages of toxicological Toxicological studies
development evaluation

4. Design of industrial Additional studies Studies of the mechanism of action, early
scale process and differential diagnosis, experimental
therapy

5. Production and use Field studies Assessment of working and environmental
of chemicals conditions and of health status of workers
and general population

Epidemiological studies

Clinical evaluation of experimental
prophylactic, diagnostic and therapeutic
methods

Adjustment and correction of requirements
for health and environmental protection

The use of these three types of information for the purposes of tissue
and organ dose estimation requires the postulation of models to
describe the absorption, distribution, retention, biotransformation,
and excretion of the original chemical or its metabolites, as a
function of time (see Chapter 4).

When the site of toxic action is located at, or very near, the
site of application, for example, the skin, then the tissue dose
estimate may be very reliable. However, when the site of toxic action
is remote, for example, a liver cell, then the estimates of
toxicologically significant doses are much less reliable.

The presence of a chemical in the blood indicates absorption;
however, the blood concentration of a chemical is in a dynamic state,
reaching higher levels with increasing absorption but decreasing as
the distribution, tissue storage, metabolic transformation, and
excretion increase. The blood concentration of a chemical is useful as
an indicator of the dose only when it is related in a defined manner
to the concentration at the site or sites of action (organs and
tissues) (Task Group on Metal Toxicity, 1976).

1.2.2 Effect and response

"Effect" and "response" are often used interchangeably to denote
a biological change, either in an individual or in a population,
associated with an exposure or dose. Some toxicologists have, however,
found it useful to differentiate between an effect and a response by
applying the term "effect" to a biological change, and the term
"response" to the proportion of a population that demonstrates a
defined effect (Pfitzer, 1976; Task Group on Metal Toxicity, 1976).

In this terminology, response means the incidence rate of an
effect. For example, the LD₅₀ value may be described as the dose
expected to cause a 50% response in a population tested for the lethal
effect of a chemical. This distinction will be made in the present
monograph, although it should be recognized that this terminology is
not generally accepted.

An effect can usually be measured on a graded scale of intensity
or severity and its magnitude related directly to the dose. Certain
effects, however, permit no gradation and can be expressed only as
"occurring" or "not occurring". Such effects are usually called
"quantal" (see for example, Finney, 1971). Typical examples of quantal
effects are death or occurrence of a tumour^a.

The toxic action of chemicals usually affects the whole organism
but the primary damage may be localized in a specific target organ or
organs in which the toxic injury may manifest itself in terms of
dysfunction or overt disease (NIEHS, 1977). According to Sanockij
(1975a), the specificity of acute toxic action can be expressed in
terms of a "zone of specific action" (Z_sp) which is the ratio
between the threshold^b dose of an acute effect at the level of the
total organism and the threshold dose for an acute effect at a
specific organ or system. If Z_sp > 1, the toxic action is specific;
if Z_sp < 1, it is non-specific.

Acute effects are those that occur or develop rapidly after a
single administration (Casarett, 1975) but acute effects may appear
after repeated or prolonged exposure as well. Chronic effects may also
result from a single exposure but more often they are a consequence of
repeated or prolonged exposures. Chronic effects are characterized not
only by their duration but also by certain pathological features. They
may arise from the accumulation of a toxic substance or its
metabolites in the body, or from a summation of acute effects. The
latent period (or the "time-to-occurrence" of an observable effect)
may sometimes be very long, particularly if the dose or exposure is
low. Other aspects of the nature of toxic effects are discussed in
section 2.6.

^a A similar classification of effects is used in radiological
protection where a distinction is made between "nonstochastic" and
"stochastic" effects (ICRP, 1977). Nonstochastic effects are those
for which the severity of effect varies with the dose. Stochastic
effects are those for which the probability of occurrence, rather
than their severity, is regarded as a function of dose. Hereditary
effects and carcinogenesis induced by radiation are considered to
be stochastic.

^b The threshold concept is discussed in section 1.3.2.

Not every effect is necessarily adverse or harmful. In some
cases, a graded effect may be either within the so-called "normal"
range of physiological variation, or an "adverse" effect, depending on
its intensity. The distinction between a physiological change and a
pathological effect (adverse effect) is sometimes very difficult to
make and there is much disagreement on this subject which will be
discussed in detail in section 1.3.1. The concept of biochemical
lesion introduced by Peters and his collaborators (Gavrilescu &
Peters, 1931; Peters, 1963, 1967), and based on the ideas of Claude
Bernard (Bernard, 1898), is of fundamental importance in this respect.
A biochemical lesion can be defined as the biochemical change or
defect which directly precedes pathological change or dysfunction
(Peters, 1967).

The Task Group on Metal Accumulation (1973) and the Task Group on
Metal Toxicity (1976) have defined the critical concentration for a
cell as the concentration (of a metal) at which undesirable (adverse)
functional changes, reversible or irreversible, occur in the cell.
Critical organ concentration has been defined as the mean
concentration in the organ at the time any of its cells reaches
critical concentration and critical organ as that particular organ
which first attains the critical concentration of a metal under
specified circumstances of exposure and for a given population. This
definition of "critical organ" differs from the generally accepted use
of the term, i.e. that the critical organ is the organ whose damage
(by radiation) results in the greatest injury to the individual (or
his descendants) (ICRP, 1965). However, some toxicologists question
the usefulness of the concept of a critical organ or tissue because it
diverts attention from the role that the various regulatory systems of
the body may have in relation to a toxic injury.

1.2.3 Dose-effect and dose-response curves

Dose-effect curves demonstrate the relation between dose and the
magnitude of a graded effect, either in an individual or in a
population. Such curves may have a variety of forms. Within a given
dose range they may be linear but more often they are not. Finney
(1952a) has discussed various transformations that can be used to make
dose-effect curves linear.

Dose-response curves demonstrate the relation between dose and
the proportion of individuals responding with a quantal effect. In
general, dose-response curves are S-shaped (increasing), and they have
upper and lower asymptotes, usually but not always 100 and 0% (see for
example Cornfield, 1954). One way of explaining the shape of
dose-response curves is that each individual in a population has a
unique "tolerance" and requires a certain dose before responding with
an effect. There exists, in principle, a low dose to which none will
respond and a high dose to which all will respond.

For each effect there will usually be a different dose-response
curve. Loewe (1959) and Hatch (1968) have discussed the relationship
between dose, effect, and response and its graphical representation in
a three-dimensional model.

If the experiment or observation is well designed (Chapters 2 and
3), the dose-response relationship will be based on data from many
individuals over a range of doses from minimum to maximum response.
Mathematical and statistical procedures are then used to establish the
curvilinear relationship that provides the best fit to all of the
data, expressed as mean values with their standard deviations at
different doses. Mathematical expressions for dose-effect and
dose-response relationships and the merits of applying normal,
log-normal, and other types of distributions are discussed in the
Appendix.

It should be pointed out that the shape of the dose-response
curve for the same substance and the same animal species may vary with
changes in experimental conditions, such as changes in the way in
which the dose is distributed in time (Weil, 1972a).

In evaluating human exposure to environmental chemicals, the dose
will usually be estimated as a function of concentration and time. In
some cases the concentration will be fairly constant and then the
time-effect and time-response relationships will be similar to the
dose-effect and dose-response relationships. However, in many cases
the concentration will vary, as will the time of exposure to specific
concentrations, and integrated relationships of dose-concentration-
time must be considered as well as dose-effect and time-effect
relationships (Druckrey, 1967; Golubev et al., 1973; Lazarev, 1963;
Weil, 1972a).

Haber's rule (ct=k) states that the product of concentration
(c) and time (t) results in a constant intensity of effect (k)
for some gases. This formula was later changed to ct^b = k (where
b is constant) which fitted other biological data better (Lazarev &
Brusilovskaja, 1934), although it also has its limitations. The
extrapolation of concentration-time relationships has been used
successfully to obtain predictions of response following long-term
inhalation exposure to low concentrations (Pinigin, 1974).

Concentration-time relationships, such as the variation of the
fraction of the dose with time as in combinations of short-term peak
concentrations and prolonged low-level concentrations in air
pollution, and variable cycles of exposure, may influence the toxic
effect. Few systematic attempts to evaluate these factors have been
made, although Sidorenko & Pinigin (1975, 1976) have described some
principles for setting air quality standards from this viewpoint, and
Pinigin (1974) has dealt with the problems of intermittent inhalation
exposure. This problem has also been discussed by Ulanova et al.
(1973, 1976).

1.2.4 Toxic effects due to a combination of chemicals

When an organism is exposed to two or more chemicals, their joint
action may be:

(a) independent -- when the chemicals produce different effects
or have different modes of action;

(b) additive -- when the magnitude of an effect or response
produced by two or more chemicals is numerically equal to
the sum of the effects or responses that the chemicals would
produce individually;

(d) less than additive (antagonism, inhibition).

More specific terminology may be used when the mechanisms of
joint action are known or when definite assumptions are made about
them (Finney, 1971; Hewlett & Pluckett, 1961). The time intervals and
sequences between exposures to different chemicals are extremely
important, and the quality as well as the degree of joint action may
depend on these variables (Kagan, 1973; Kustov et al., 1974; Williams,
1969). Furthermore, the joint action at lethal dose levels may be
quite different from that at low dose levels, when the effects or
responses are often only additive or independent (Smyth et al., 1969;
Ulanova, 1969).

Most statistical models for joint action have been developed for
situations in which two or more chemicals are administered
simultaneously or within a short (few minutes) time interval. A model
proposed by Finney (1952b, 1971) is often used for predicting the
acute joint toxicity of chemicals. The model is strictly applicable to
mixtures of chemicals that act at the same site, producing the same
type of acute toxic effect and having parallel regression lines of
probits against log doses (see Appendix). For a mixture of, for
example, three chemicals, the equation for the median effective dose
(ED₅₀) is

1 contour integral_A contour integral_B contour integral_C
= + +
ED₅₀ (A,B,C) ED₅₀ (A) ED₅₀ (B) ED₅₀ (C)

where contour integral_A, contour integral_B and contour integral_C are
the fractions of substances A, B, and C in the mixture. When all
the values on the right hand side of equation (1) are known, a predicted
ED₅₀ (assuming additive joint action) can be calculated and compared
with the actual ED₅₀ of the mixture determined experimentally. A
smaller than predicted ED₅₀ demonstrates a more than additive
response (synergism), a greater than predicted ED₅₀ indicates a less
than additive response (antagonism). Smyth et al. (1969) demonstrated
that this equation can give satisfactory results under conditions that
are less restrictive than stated above, for example in identifying the
type of acute joint action among randomly selected industrial
chemicals. Ball (1959) applied the equation to the estimation of
maximum allowable concentrations for occupational exposure to mixtures
of substances that exercise a "similar joint action", e.g. benzene and
toluene. Another model for estimating the results of joint action has
been developed using the isoeffective concentrations instead of ED₅₀
in equation (1) (Pinigin, 1974).

The possibility of predicting the type of joint action is
enhanced if there is information on the metabolism and disposition of
the chemicals (Murphy, 1969; Williams, 1969). Basic principles
concerning the kinetics of reactions of chemicals with primary sites
(tissue receptor sites) and with secondary sites are important in
considering the joint action of chemicals (Gaddum, 1957; Schild et
al., 1961; Veldstra, 1956; Williams, 1969). The relevant factors seem
to be the relative affinities at the sites of action (e.g. target
enzymes, neuroeffector sites, and other vital target sites), and at
the sites of loss or sinks (e.g. detoxifying enzymes, nonvital tissue
binding sites, pathways of excretion, and storage sites), and the
intrinsic activity at the sites of action^a. Since there is a limited
number of sites of action and sinks within any organism, there will be
a limited dose range within which synergism or antagonism can be
demonstrated. This, of course, is only one area where more information
could help in predicting the effects of the joint action of chemicals.
Other areas where knowledge is insufficient are the possible effects
of low-level, prolonged exposures to mixtures of chemicals and the
effects of multiple stresses including chemicals, physical factors
such as heat and noise, and pre-existing disease (NIEHS, 1970).

^a Relative affinity -- reciprocal of the dissociation constant for
the chemical-receptor complex. Intrinsic activity -- the capacity
of the chemical to produce an effect when it combines with a
reactive tissue site. For precise definitions see for example
Ariëns et al. (1957).

Simultaneous exposures to the same chemical in different media
(e.g. air, water, food) which is called "complex action" by some
toxicologists (Korbakova et al., 1971; Kustov et al., 1974; Pinigin,
1974; Spynu et al., 1972) is another aspect of multiple stresses which
has considerable practical importance.

1.3 Interpretation of Laboratory Data

It is essential that all experiments to evaluate toxicity should
be designed to be scientifically meaningful, and should not be
conducted merely to comply with statutory regulations. Thus, the
evaluation of each new chemical will not be an identical task and
procedures will differ, to some extent, from one compound to another.
The protocol for an experiment will evolve gradually during the
experiment, in accordance with earlier findings. It is useful to have
laboratory data validated by a study of the mechanisms involved in the
development of the toxic lesion. Furthermore, numerous endogenous and
environmental variables can modify the toxicity of chemicals, as
discussed in subsequent chapters. In some instances, the influence of
these variables is known, and can be controlled, but often this is not
the case and this may cause serious difficulties in the interpretation
of laboratory toxicity data.

In the present context, we are mainly interested in the
interpretation of laboratory data with a view to their application in
the evaluation of the health risk to man. The discussion will
therefore be limited to a few topics that are particularly relevant in
this respect.

1.3.1 Distinction between adverse and nonadverse effects

An adverse, or "abnormal" effect has often been defined in terms
of a measurement that is outside the "normal" range. The "normal"
range, in turn, is usually defined on the basis of measured values
observed in a group of presumably healthy individuals, and expressed
in statistical terms of a range representing 95% confidence limits of
the mean or, for individuals, in terms of 95% "tolerance" limits^a
established with a derived degree of confidence (95% or 99%). An
individual with a measured value outside this range may be either

^a Tolerance limits are defined as m ± ks where m is the sample
mean, s is the sample standard deviation and k is a coefficient
that depends both on the size of the sample (N) and the required
degree of confidence. If the "normal" mean has been determined on
the basis of a very large sample, the 95% limits will be equal to
µ ± 1.96sigma where µ and sigma are the "true" or population values
of the mean and the standard deviation, respectively (see for
example Owen, 1955).

"abnormal" in fact, or one of that small group of "normal" individuals
who have extreme values. According to Sanockij (1970), the distinction
between "normal" and "abnormal" values based on statistical
considerations may be used as a criterion for adverse effects, if the
exposed population consists of adult, generally healthy individuals,
subject to periodical medical examination, such as workers. Departures
from "normal" values associated with a given exposure will then be
considered as adverse effects, if the observed changes are:

(a) statistically significant ( P < 0.05) in comparison with a
control group, and outside the limits (m ± 2 s) of
generally accepted "normal" values;

(b) statistically significant ( P < 0.05) in comparison with a
control group, but within the range of generally accepted
normal values, provided such changes persist for a
considerable time after the cessation of exposure; and

(c) statistically significant ( P < 0.05) in comparison with a
control group, but within the "normal" range, provided
statistically significant departures from the generally
accepted "normal" values become manifest under functional or
biochemical stress.

This statistical definition of adverse effects is less suitable
for the general population which includes some groups that may be
specially sensitive to environmental factors, particularly the very
young, the very old, those affected with disease, and those exposed to
other toxic materials or stresses. In this case, it is practically
impossible to define "normal" values, and any observable biological
change may be considered as an adverse effect under some
circumstances. For this reason, attempts have been made to set
criteria for adverse effects based on biological considerations and
not only on statistically significant differences with respect to an
unexposed population (control group). Although there is no general
agreement on such criteria and the ultimate decision on what is an
adverse effect will have to depend, in each case, on experience and
expert judgment, it may nevertheless be useful to give examples of
such criteria, which illustrate at the same time how different such
criteria may be.

A Committee for the Working Conference on Principles of Protocols
for Evaluating Chemicals in the Environment (NAS, 1975) defined
nonadverse effects as the absence of changes in morphology, growth,
development, and life span. Furthermore, nonadverse effects do not
result in impairment of the capacity to compensate for additional
stress. They are reversible following cessation of exposure without
detectable impairment of the ability of the organism to maintain
homeostasis, and do not enhance susceptibility to the deleterious
effects of other environmental influences.

On the other hand, adverse effects may be deduced as changes
that:

"1. occur with intermittent or continued exposure and that result in
impairment of functional capacity (as determined by anatomical,
physiological, and biochemical or behavioural parameters) or in a
decrement of the ability to compensate additional stress;

2. are irreversible during exposure or following cessation of
exposure if such changes cause detectable decrements in the
ability of the organism to maintain homeostasis; and

3. enhance the susceptibility of the organism to the deleterious
effects of other environmental influences."

Soviet toxicologists emphasize that criteria for differentiating
between adverse and nonadverse effects should not be based on overt
pathology (e.g. inflammation, necrosis, hyperplasia), and have
proposed, inter alia, a number of criteria based on metabolic and
biochemical changes. Such changes are considered to be adverse if:

(a) the metabolism of a substance becomes less efficient or the
elimination of a substance (expressed in terms of biological
half-time, T_´) slows down with increasing doses of the
substance (Sanockij, 1956);

(b) enzymes that have a key significance in metabolism are
inhibited (Kustov & Tiunov, 1970);

(c) the inhibition of a certain enzyme results in an increase in
the concentration of the corresponding natural substrate in
the body and/or in a decreased capacity to metabolize the
specific substrates in a loading test (Kustov & Tiunov,
1970);

(d) the relative activities of different enzyme systems are
changed (e.g. the ratio of the activities of asparagine and
alanine transaminases (Kustov & Tiunov, 1970)).

Pokrovskij (1973) also attaches great importance to the changes
in the pattern of isoenzymes in the blood, and to the changes in the
subcellular membranes (e.g. lysosomal membranes) resulting from the
action of toxic substances.

Differentiation between "nonadverse" and "adverse" effects
requires considerable knowledge of the importance of reversible
changes and subtle departures from "normal" physiology and morphology
in terms of the organism's overall economy of life, ability to adapt
to other stresses, and their possible effects on life span. Newer and

improved methods of research have increasingly provided more sensitive
tests for subtle biological deviations such as induction of enzymes of
the smooth endoplasmic reticulum of the liver, or reversible
hypertrophy of the liver. These types of changes are produced by
relatively low doses of many chemicals and they are considered by some
authors to be adaptive and generally useful to health, and by others
to be indicative of injury (Hermann, 1974; Kustov & Tiunov, 1970;
Parke, 1975). One of the most challenging areas for basic research in
toxicology today is the acquisition of data that can be used to
estimate whether, or under what conditions, subtle changes in enzyme
activities, nerve action potentials, altered behavioural reaction etc.
indicate impairment of physiological function or predict impending
development of more serious irreversible injury, should exposure to
the chemical continue.

In addition to all these considerations, the possibility must be
kept in mind that an effect may not be seen because the number of
animals studied was inadequate, the observation time was too short, or
for other reasons.

1.3.2 Threshold: practical and theoretical considerations

The concept of "threshold" is complex and the term has to be
carefully defined, so that statements concerning this concept in
relation to the protection of human health are not confused by
semantic differences. A distinction should be made between the
threshold for individuals and thresholds for limited groups of
individuals or general populations.

The dependence of effect or response on the dose of a chemical
has already been discussed (section 1.2). As a rule, the intensity of
the effect or response decreases with reduction in dose, and a
biological reaction often reaches zero before the dose becomes equal
to zero. Below a certain limiting exposure level, or dose, i.e. below
the threshold, a chemical substance may not elicit a toxic effect. The
threshold for an adverse effect of a chemical is defined by some
toxicologists as the minimum exposure level or dose that gives rise to
biological changes beyond the limits of homeostatic adaptation. True
homeostatic adaptation should be carefully distinguished from
pathological processes (Sanockij, 1975a).

The existence of a threshold for all adverse effects is, however,
still a matter for discussion. Sanockij (1975b) has provided data
which show that small quantities of environmental chemicals may not
reach their receptor because the rate of elimination or metabolic
degradation is relatively more effective with smaller doses. It has
also been suggested that where effective repair processes are present,
even if a substance interacts with the receptor, it need not
necessarily produce an adverse effect.

For some toxic effects, such as neoplastic disease or mutations
of genetic material, it has been assumed by some authors that a single
molecule of a chemical is sufficient to initiate a process that may
progressively lead to an observed, harmful effect. For this reason, it
may not be possible to demonstrate that a threshold dose for a
carcinogen or a mutagen exists (Saffiotti, 1973).

Other scientists view carcinogenic or mutagenic chemicals as
toxic entities that may have special properties with regard to the
nature and characteristics of their adverse effects, but are subject
to the same physicochemical and biological interactions that are
considered to result in a threshold dose for other chemicals (Dinman,
1972; Sanockij, 1970; Stokinger, 1972; Weil, 1972a).

The question of the existence of a threshold for carcinogens and
mutagens was recently discussed by a WHO Scientific Group (WHO,
1974a), which concluded that "the existence of a threshold may be
envisaged. Nevertheless, the difficulties of determining a threshold
for a population are great. Therefore, mathematically derived
conclusions that it is impossible to demonstrate no-effect levels
experimentally cannot be ignored". A "no-effect" level for a group of
animals may occur because the dose is really below the theoretical
no-effect level (i.e. below the threshold) or because the number of
animals is too small. For example, in an experiment with 20 animals,
it is possible that none of the animals will show an effect whereas in
an experiment with 100 animals some response might be seen. However,
an upper limit for the probable response can be estimated
statistically. For instance, if in an experiment with 100 animals, no
response has been observed, it can be shown that there is a 95%
probability that, under the conditions of the experiment, the upper
limit of response is 3%, and that there is a 99% probability that the
response will not exceed 4.5%. Even in an experiment with 1000 animals
showing no response, the upper 95% confidence limit of response is 3
animals showing an effect per 1000 treated animals (Food & Drug
Administration, 1970).

Another reason for not having seen a response in an experiment
may be that the time of observation was too short. This may be the
case, for example, when the quantal effect considered is a cancer,
with a long latent period between exposure and appearance of tumours.

For these reasons the "no-effect level" has no real meaning and a
better term is "no-observed-effect level" (NAS, 1975).

1.3.3 Extrapolation of animal data to man

In many cases, studies with laboratory animals make it possible
to predict the toxic effects of chemicals in man. However, it is
important to realize that experimental animal models have their
limitations, and that the accuracy and reliability of a quantitative
prediction of toxicity in man depend on a number of conditions, such
as choice of animal species, design of the experiments, and methods of
extrapolation of animal data to man.

Hoel et al. (1975) considered the criteria for the adequacy of an
experiment to be used for the extrapolation of animal data to man.
They include: test animal species and strain (the animal should be
susceptible to induction of the effects under consideration); the
number of animals; the route of administration (which should include
the routes of human exposure); and the physical state and chemical
form of the agent. The side effects of the chemical and its organ
specificity should also be taken into account in the design of the
experiment. In interpreting the results, attention should be paid to
adequate survival of the animals, to possible intercurrent disease,
the quality and extent of pathological data, the quality and extent of
relevant data collection during the experiment, and the availability
of data at the time of interpretation.

1.3.3.1 Species differences and related factors

The most difficult problem in the extrapolation of animal data to
man is the conversion from one species to another. For most
substances, the pathogenesis of poisoning is the same in man and other
mammals, and for this reason the signs of intoxication are also
analogous. Thus, quantitative rather than qualitative differences in
toxic response are most common. Man may be more sensitive than certain
laboratory animals but there are also many cases where some animal
species are more sensitive than man. For example, the mouse is most
sensitive to atropine, the cat is less sensitive, while the dog and
the rabbit tolerate atropine in doses 100 times higher than the lethal
dose for man. However, the dog is more sensitive to hydrocyanic acid
than man (Elizarova, 1962).

Species differences in sensitivity can often be explained by
differences in metabolism, in particular by quantitative and
qualitative differences in the ability of an enzyme to detoxify
chemicals, and also by differences in the rates of absorption,
transport, distribution, and elimination of chemicals (Curry, 1970;
Ecobichon & Cormeau, 1973; Flynn et al., 1972; Hucker, 1970; Portman
et al., 1970; Sato & Moroi, 1971). Rall (1970) discussed various
factors to be considered in the selection of animal models for
pharmacotherapeutic studies in relation to the steps that intervene

between administration of the drug (or chemical) and the arrival of
the compound at the ultimate sites of action. After oral
administration, absorption in standard laboratory animals is generally
considered to be very similar to man, although there are quantitative
differences for some compounds. For example, species differences in
the absorption and action of some compounds are related to differences
in the bacterial flora of the gastrointestinal tract (Williams, 1972).
Rall further concluded that the distribution and storage of drugs are
reasonably consistent in mammalian species, including man, although
plasma binding tends to be more extensive in man than in small
mammalian species. Urinary excretion in different animal species
depends to some extent, on their different diets, since diet
influences urinary pH and thus the extent of ionization of compounds.
Biliary excretion is quite variable from species to species and
apparently is more extensive in mice and rabbits than in rats or man.
Species differences in response to chemicals appear to be mainly
related to rates of biotransformation which are generally more rapid
in small laboratory animals than in man.

One of the most potent bladder carcinogens, 2-naphthalenamine
(2-naphthylamine), produces bladder cancer in the dog, hamster, and
man, but not in the rat, rabbit, or guineapig. Species differences in
the carcinogenicity of 2-fluorenylacetamide (2-acetaminofluorene) have
been attributed to the different extents of metabolism to the
proximate carcinogen, the N-hydroxy derivative (Miller et al.,
1964). Similarly, strain differences in metabolism may also affect
toxicity (Mazze et al., 1973).

If metabolic information is available, differences in absorption,
distribution, biotransformation, and elimination of toxic substances
in man and animals should be taken into account when selecting
experimental animals.

Species differences in toxicity may also be due to differences in
cellular transport. Aflatoxin, which is more toxic to rats than to
mice, both as an acute poison and as a carcinogen, is transported more
slowly into the liver cells and is metabolized more rapidly in the
mouse than in the rat (Portman et al., 1970).

In determining the required duration of an animal experiment, it
is often useful to compare the life span of the animal with that of
man. Using the "body weight rule", the average life span for 70
species of mammals showed a linear correlation with body weight, but
the average life span of man was found to be an exception (Krasovskij,
1975). The regression equation obtained from a study of many mammals
showed that the average life span for a mammalian representative,
having the same body weight as man (70 kg) was equal to 15 years.

Thus, if this assumption is accepted the average life span of a rat
(about 2.5 years) corresponds to only 15-17 years of a man's life.
This inconsistency in the life spans of man and experimental animals
should be taken into account in the design and interpretation of
animal experiments for the evaluation of toxicity to man.

There are other problems in the evaluation of toxicity to man
from experiments on animals, such as where an effect is difficult to
measure or where similar conditions are difficult to obtain in animal
models, for example, intelligence and the more esoteric behavioural
changes. Furthermore, in animal experiments, the effects of social
factors, so important to man, cannot be evaluated.

For these reasons, when extrapolating from animals to man it is
prudent to apply a species conversion factor which should be
determined on the basis of biological considerations and the available
information on the test species (Hoel et al., 1975). There is no
definite rule for the species conversion factor. If the extrapolation
of data is based on the most sensitive species tested, some
toxicologists use a factor of 1 (Sabad et al., 1973), but others
recommend a factor as large as 10 (Weil, 1972a).

The unit of dose to be used has also to be considered in the
extrapolation of data to man and it has recently been recommended that
the dose per unit surface area approximately equivalent to the weight
raised to the power 2/3 should be used. If the dose is given in terms
of dietary concentration, there seems to be no need to make the
surface area adjustment (Hoel et al., 1975; Mantel & Schneiderman,
1975).

A separate problem, to which there appears to be no satisfactory
answer at present, is the conversion from an inbred animal strain to a
genetically highly heterogeneous human population (Hoel et al., 1975).

1.3.3.2 Safety factors

In almost all instances, laboratory data on the toxicity of
chemicals are drawn from experiments in which the adverse effect
occurs at a considerably higher incidence rate than would be
acceptable in man. For this reason alone, and apart from the
biological differences between laboratory species and man, an
extrapolation from a known dose-response range to an unknown range is
necessary. Indeed, essentially the same problem arises when a human
accident or epidemiological data are used as the starting point.

Traditionally, a safety factor has been introduced to provide for
uncertainties in extrapolation from animals to man, and from a small
group of individuals to a large population. Such safety factors have
ranged from 1 to as much as 5000. Because of the current uncertainty
regarding the mathematical and biological reliability of methods for
extrapolating from high doses to low doses, primary dependence on
somewhat arbitrary safety factors continues. However, means of
extrapolating from high to low doses are being intensively studied at
the present time, especially with respect to carcinogenicity.

Most regulatory authorities rely on the use of safety factors but
there are no precise guidelines for deciding the appropriate size of
such a factor. Sanockij (1962) and Sanockij & Sidorov (1975) have
discussed the rationale for different safety factors. In general, the
size of the safety factor will depend on (a) the nature of the toxic
effects, (b) the size and type of population to be protected, and (c)
the quality of toxicological information available. A factor of 2 to 5
or less may be considered as sufficient if the effect against which
individuals or a population are to be protected is not regarded as
very severe, if only a small number of workers are likely to be
exposed, and if the toxicological information is derived from human
data. On the other hand, a safety factor as large as 1000 or more may
be required if the possible effect is very serious, if the general
population is to be protected, and if the toxicological data are
derived from limited experiments on laboratory animals. In some cases,
the safety factor may be a value that has been used with reasonable
success and is, therefore, perpetuated.

For most food additives that are not considered to be
carcinogenic, it has been the accepted practice to divide the
no-adverse-effect dose (i.e., the maximum ineffective dose) in animals
by 100, to arrive at an acceptable daily intake (ADI) for man
(Vettorazzi, 1977; WHO, 1958). For pesticides and certain
environmental chemicals, safety factors ranging from less than 100 to
several thousand have been used (Vettorazzi, 1975). For some
occupational exposures, and for certain air pollutants (WHO, 1977)
much smaller safety factor have been proposed in the range of 2-5.
Safety factors have also been proposed for carcinogenic chemicals
ranging from 100 (Druckrey, 1967; Janyseva, 1972) to about 5000 (Weil,
1972a) but they have not been generally accepted.

1.3.3.3 Low-dose extrapolation

Low-dose extrapolation is based on mathematical models that are
used to predict the response at a given low dose or to predict that
dose which gives a predetermined low response. Such models may relate
the incidence of a quantal effect to dose, or they may consider the
distribution of the "time to occurrence" of a condition and its
relation to dose. In both cases, the results of extrapolation are

strongly dependent on the choice of the model. For example, the
Advisory Committee on Food Additives (FDA, 1970) noted that
dose-response data may fit several models equally well in the 2% to 5%
range, but the doses extrapolated to very low responses would differ
very strikingly: the ratio of ED₁ to ED_0.000001 would be either
100, 100 000, or 1 000 000 for the probit, logistic, or one-hit
curves, respectively.

Several extrapolation procedures have been proposed which will
give an upper limit to the dose corresponding to a low response. In
other words, the result of extrapolation will not be the best estimate
of the unknown dose required to give the desired response but a dose
that is most likely to be below the dose required to give this
response. Two procedures based on this approach have received
particular attention: one is based on the one-hit model, the other on
the probit model.

The one-hit model assumes that an effect can be induced after a
single susceptible target has been reached by a single biologically
effective unit of dose (see for example Cornfield, 1954). At low
doses, this model is numerically equivalent to the linear
dose-response model which is compatible with animal data for some
carcinogens (Druckrey, 1967) and with some human data such as the
incidence of lung cancer in relation to the number of cigarettes
smoked per day (Doll, 1967). In their simplest form (i.e. when the
true response at zero dose is assumed to be zero), the currently used
extrapolation procedures based on this model (Gross et al., 1970; Hoel
et al., 1975; Schneiderman, 1971) operate as follows: (1) the upper
99% confidence limit (UCL) is estimated for the observed response at a
dose d; (2) a desired limit is set for a low response ( R) e.g. 1
in 1 000 000; and (3) the dose ( d_e) that would produce a response
which is, with a 99% probability, lower than R is calculated from
the equation d_e = d*R/(UCL). Such procedures are more
conservative than the procedures based on any other currently used
dose-response model (probit, logit, or extreme-value models). In
addition, the one-hit model seems to have a reasonable biological
basis for carcinogenesis at low doses (Hoel et al., 1975).

Mantel & Bryan (1961) proposed the use of probits (see Annex to
this Chapter) and a log-normal distribution to describe the
variability of the sensitivities (tolerances) of individuals in a
population. The probit model gives a dose-response curve that is
concave at low-dose levels, and is less conservative than the linear
model based on the one-hit hypothesis. The Mantel-Bryan procedure (see
for example Schneiderman & Mantel, 1973) involves (1) the choice of a
desired limit of response (R) (e.g. 1 in 1 000 000); (2) the
estimation of the upper 99% confidence limit (UCL) for the observed
response at dose d, and (3) imposing a probit-log dose straight line

through UCL, with a slope (ß) equal to 1 (i.e. one probit per 10-fold
dose-range). The choice of the slope (ß) is critical in this
procedure. ß = 1 has been proposed because a slope greater than 1 is
usually (but not always) observed in carcinogenesis experiments. The
Mantel & Bryan procedure has been modified to take into account
response levels in control groups (Mantel et al., 1975).

A second category of models is based on the observation that the
median "time to occurrence" (latent period) of an effect such as
cancer may increase as the dose decreases but not proportionally. A
thousand-fold change in the dose usually causes an approximately
ten-fold change in the median time to tumour appearance and, with
decreasing dose levels, a dose may be reached which would predict
tumour occurrence beyond the life expectancy of the exposed
individuals. This would still be consistent with the hypothesis that
molecular changes in the cells, occurring in proportion to the
concentration of carcinogens, are the initiating event (for a recent
review see Jones & Grendon, 1975). One model (Altschuler, 1973; Blum,
1959; Druckrey, 1967) considers a log-normal distribution of the
time-to-occurrence with median time depending on dose but with
standard deviation independent of the dose. The dose d is related to
the median time (t) by d= c/tⁿ, where n is assumed to be greater than
one, and c is a constant. Peto et al. (1972) compared this model
with another model in which the time to occurrence is considered to
have a Weibull distribution (Day, 1967; Peto & Lee, 1973). They found
that the Weibull distribution agreed with experimental data better
than the log-normal, but this no doubt depends on the type of cancer
involved. The low-dose extrapolation using these models would also be
strongly dependent on the choice of the model (Chand & Hoel, 1974).
Dose-"time-to-occurrence"-response relationships in cancer risk
assessment have also been considered by Janyseva & Antomonov (1976).

The application of all these procedures presents practical
difficulties (Hoel et al., 1975). Low-dose extrapolation is thus a
very difficult problem that cannot be solved by statistical methods
alone. Great caution should be exercised in using the existing methods
and their inherent limitations should always be kept in mind. Good
experimental data, combined with human data if available, and an
understanding of the mechanisms of toxic action are essential if the
task of low-dose extrapolation is to be accomplished satisfactorily.

1.3.3.4 Other methods of extrapolation

A method for extrapolation from one species to another based on
an established relationship between the indices of toxicity and body
weight for different animal species has also been suggested
(Krasovskij, 1976a). In mammals, the weights of internal organs, and
many physiological variables (pulse and respiration rates, consumption

of oxygen, food, and water, liver microsomal enzyme activity) show a
log-log linear relationship with body size of the animals (allometric
ratios). This regularity appears to be valid for more than 100
different variables including the period of gestation, litter size,
erythrocyte life span, and latent period of tumour development but
there are also other variables to which it does not apply. This "body
weight rule" (Krasovskij, 1975) may be expressed as follows: the
logarithms of biological variables of mammals show a linear regression
to the logarithms of the body weight.

Krasovskij (1976b) showed that values for the lethal dose for
dogs of several chemicals obtained from regression analysis of
toxicity in four other species of small mammals compared well with
predictions made from direct extrapolation from albino rats or from
the most sensitive animal species, or from the relationship of body
surface area, and also with predictions obtained by the method of
Ulanova (1969) and Van Noordwijk (1964). For the calculation of
extrapolation coefficients from regression equations, see Krasovskij
(1976b).

1.4 Human Data

1.4.1 Ethical considerations

In research involving human subjects, a number of elements, such
as the assessment of risk, potential benefit, and quality of consent,
have to be evaluated to ascertain whether ethical considerations are
satisfied. The essential provisions for protecting human subjects in
experimentation and research have been expounded by many international
and national organizations. Key factors include the right to informed
consent and freedom from coercion. The international instruments in
dealing with this matter are the Declaration of Helsinki (as revised
in Tokyo in 1975) and Article 7 of the International Covenant on Civil
and Political Rights, adopted by the United Nations General Assembly,
December 1966. Article 7 provides that "no-one shall be subjected
without his free consent to medical or scientific experimentation"
(Cranston, 1973; WHO, 1976b). Some countries possess specific codes of
ethics relating to human experimentation, and special problems of
experimentation that involve the use of fetuses, children, the
mentally ill, and prisoners require special consideration.

It is essential that human experimentation should only be
undertaken when there is adequate evidence from animal and other
studies that both the chemical and the circumstances of administration
are safe. Every experiment with human volunteers should be subject to
prior review and approval by a local ethical committee in order to
ensure that the intended study complies with the ethical principles
embodied in the Declaration of Helsinki and with other requirements of
national and local bodies.

Ideal conditions of truly informed consent may not always be
achieved in practice, consequently the burden of responsibility rests
mainly with the investigator and, to a lesser extent, with the peer
review body. Because of these difficulties, the guidelines and
procedures for the protection of human subjects should be constantly
reviewed and updated (WHO, 1976b).

In any case, collection of data from human subjects must be
accomplished with due respect for human rights and dignity. The use of
ethics committees with broad representation to review and approve all
such experimentation is recommended to protect the rights of human
subjects and to ensure responsible investigation.

1.4.2 Need for human investigations

Although there is general repugnance at the idea of using human
subjects to assess the safety of environmental chemicals, the question
is not whether or not human subjects should be used in toxicity
experiments but rather whether such chemicals, deemed from animal
toxicity studies to be relatively safe, should be released first to
controlled, carefully monitored groups of human subjects, instead of
being released indiscriminately to large populations with no
monitoring and with little or no opportunity to observe adverse
effects (Paget, 1970).

The prediction and prevention of possible toxic hazards that may
arise from the introduction of chemicals into the environment can be
made more valid if data from studies of the chemical in human subjects
are available. Three particular aspects of human toxicology have need
of such information, namely: (a) the selection, through comparative
consideration of metabolism, of the most appropriate animal species
for studies to predict the human response; (b) investigation of a
specific, reversible effect of the compound in the most sensitive
animal species, to determine whether there is a correlation with a
similar effect in man; and (c) study of effects specific to man.

Certain types of information about the effects of chemicals can
only be obtained by direct observations on man. Often, carefully
controlled experiments can provide significant information at doses
well below those anticipated to be "safe"; measurement of subtle
changes of reaction time, behavioural functions, and sensory responses
may be examples. In other cases, useful information may be obtained by
careful studies on human cells or tissue maintained by culture
techniques.

Human toxicological data include both the data obtained from
epidemiological surveys of populations exposed to a toxic chemical
under normal conditions of use, in cases of acute accidental poisoning
and in occupational exposure, and the data from experiments in
volunteers. Although an experiment is defined as observations under
controlled conditions of exposure, there is, at times, only a grey
area that distinguishes an experiment with human subjects from
observations on human subjects under natural conditions. For example,
some segments of human populations are at higher risk and should be
particularly closely monitored, e.g., those exposed to chemicals at
work or those receiving continuous treatment with medicines. The
periodic clinical evaluation of workers is normally the responsibility
of the employer and careful records of these examinations coupled with
measurement of exposure conditions often exist. If accidental
excessive exposure of an individual or a population should occur, it
is both ethical and pertinent to learn as much as possible,
recognizing always the right of the patient. Because of the wide
individual variation in the toxicity of chemicals to man, the final
evaluation should be based on information obtained from as widely
varied a human population as is compatible with the various ethical
principles involved.

1.5 The Use of Toxicological Data in Establishing Environmental
Health Standards

1.5.1 Environmental health standards

The aim of environmental health standards is to protect
individuals, human populations, and their progeny from the adverse
effects of hazardous environmental factors, including chemicals. A
sound principle of health protection is to keep all exposures as low
as reasonably achievable, subject to the condition that the
appropriate exposure limits, defined by the standard, are not
exceeded.

Environmental health standards for chemicals may be formulated
either in terms of concentrations in environmental components (e.g.,
air, water, food, consumer products) or in terms of amounts of
substances that may be taken into the body. These concentrations and
amounts should be sufficiently low that the threshold dose (if it
exists and can be determined) will not be reached, or that the
population of concern will not be subject to "unacceptable" risk, even
following life-time or working life-time exposure. In some cases, as
for irritant air pollutants, the distribution of exposure
concentrations in time should also be considered. Standards may also
prescribe the quantity of a substance to be used at any time and the
manner of its use.

Social, cultural, and economic considerations should be taken
into account in setting standards, but never to the detriment of
health protection which should be of primary concern.

It is obvious that a standard setting process will necessarily
involve many considerations besides toxicology. This process is often
very different in different countries and different types of society.
In general, however, it involves appraisal of toxicological data,
particularly of dose-response relationships, including the effects on
non-human targets (plants, animals, materials); social and economic
analysis, policy analysis and review of experience elsewhere, leading
eventually to an administrative or policy decision concerning the
standard. Other relevant questions include the technological
feasibility of achieving a standard, the cost and benefit of
implementing it, means of enforcement, other public health priorities
etc. Many of these topics are outside the scope of the present
monograph.

1.5.2 Assessment of health risk and evaluation of benefits

Assessment of health risk from a given exposure to an
environmental factor is an essential step in any procedure for setting
environmental health standards. Assessment of health risk involves
more than routine application of "safety factors", or low dose
extrapolation which provides estimates of response that are, strictly
speaking, applicable only to the conditions of the experiment. The
application of a "species conversion" factor has been discussed and
the difficulties pointed out (section 1.3.3.2). Questions such as the
incidence of effects in various age groups and the degree of life
shortening in affected individuals are all relevant to standard
setting. For this reason the study of "time-to-occurrence" models in
the extrapolation of data should be encouraged (Albert & Altschuler,
1976; Hoel et al., 1975). In addition, the seriousness of adverse
effects will have to be evaluated from the public health and social
viewpoint. Attention should also be paid to the heterogeneity of human
populations, and, at present, it is not clear how the existence of
susceptible groups, and the influence of nutrition and pre-existing
disease in human populations should be taken into account. The
existing methods of extrapolation from animal data to man deal with
exposures to single substances whereas the actual human environment
contains a large number of hazardous chemicals and other factors that
can interact and considerably modify the effects, for example, in
cancer induction (Bingham & Falk, 1959; Montesano et al., 1974). From
this viewpoint, the importance of epidemiology and systematic
surveillance of high risk groups cannot be overemphasized.

The acceptability of a given risk should also be considered in
standard setting. This, as well as the judgment on safety (which
involves decision on the acceptability of risk) exceeds the expertise
of toxicologists. This is a domain where society at large has a role
to play. Political decisions are also required on various social,
economic, and ecological concerns. The same applies to the evaluation
of benefits. As pointed out by a WHO Expert Committee (WHO, 1974a)

"the expertise needed for the evaluation of risk is different from
that needed for benefit evaluation. On the risk side, concern is
focused on adverse health effects on man, damage to the environment,
and misuse of natural resources. On the benefit side, the emphasis is
on value to the consumer and the country". The interaction of all
these factors is often described by the term "risk and benefit
analysis" (see for example Falk, 1975), which is only partly within
the area of toxicological expertise. The final judgment as to whether
the benefit does or does not justify a risk is for society to make.

1.5.3 An example of toxicological information used in standard
setting

Although standard setting procedures will differ from country to
country, and the requirements for toxicological information will vary
to a considerable extent, it may be useful, nevertheless, to describe,
as an example, the procedure used in the USSR for setting s