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 Orgnization
    Geneva, 1985

         The International Programme on Chemical Safety (IPCS) is a
    joint venture of the United Nations Environment Programme, the
    International Labour Organisation, and the World Health
    Organization. The main objective of the IPCS is to carry out and
    disseminate evaluations of the effects of chemicals on human health
    and the quality of the environment. Supporting activities include
    the development of epidemiological, experimental laboratory, and
    risk-assessment methods that could produce internationally
    comparable results, and the development of manpower in the field of
    toxicology. Other activities carried out by the IPCS include the
    development of know-how for coping with chemical accidents,
    coordination of laboratory testing and epidemiological studies, and
    promotion of research on the mechanisms of the biological action of

        ISBN 92 4 154187 3  

         The World Health Organization welcomes requests for permission
    to reproduce or translate its publications, in part or in full.
    Applications and enquiries should be addressed to the Office of
    Publications, World Health Organization, Geneva, Switzerland, which
    will be glad to provide the latest information on any changes made
    to the text, plans for new editions, and reprints and translations
    already available.

    (c) World Health Organization 1985

         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




    1.1. Introduction






    6.1. Bacteria
    6.2. Fungi
    6.3.  Drosophila
    6.4. Cultured mammalian cells




    9.1. Gene mutation in yeast
    9.2.  Drosophila somatic cell mutation assays
    9.3. Assays for DNA damage SSB (single-strand breaks) and UDS
         (detected via autoradiography or scintillation counting)
    9.4. Assays for the induction of aneuploidy
    9.5. Mammalian cell gene-mutation assays
    9.6. Chromosomal-aberration assays
    9.7. Assays for polyploidy induction
    9.8. Sister chromatid exchange (SCE) assays
    9.9. Transformation assays





Dr D. Amacher, Pfizer Central Research, Groton, Connecticut

Dr P. Arni, Ciba-Geigy, Basle, Switzerland

Dr J. Ashby, Central Toxicology Laboratory, Imperial Chemical
   Industries, Ltd, Macclesfield, Cheshire, United Kingdom

Dr R. Baker, School of Public Health and Tropical Medicine,
   University of Sydney, Sydney NSW, Australia

Dr J.C. Barrett, Laboratory of Pulmonary Function and Toxicology, 
   National Institute of Environmental Health Sciences, Research 
   Triangle Park, North Carolina, USA 

Dr R.H. Barrett, The Boots Company Industrial Division,
   Nottingham, United Kingdom

Dr M.O. Bradley, Merck, Sharp & Dohme, West Point, Pennsylvania, 

Dr T. Brooks, Shell Research, Ltd, Tunstall Laboratory, Kent, 
   United Kingdom 

Dr A. Carere, Higher Institute of Health, Rome, Italy

Dr W. Caspary, National Toxicology Program, National Institute of 
   Environmental Health Sciences, Research Triangle Park, North 
   Carolina, USA 

Dr D.V. Chitavichus, Institute of Medical Genetics, Moscow, USSR

Dr C.L. Crespi, Gentest Corporation, Woburn, Massachusetts

Dr N. Danford, Department of Genetics, University College of
   Swansea, Swansea, United Kingdom

Dr B.J. Dean, Shell Research, Ltd, Tunstall Laboratory, Kent, 
   United Kingdom 

Dr G. Delow, Paterson Laboratories, Christie Hospital & Holt Radium 
   Institute, Manchester, United Kingdom 

Dr F.J. de Serres, National Institute of Environmental Health
   Sciences, Research Triangle Park, North Carolina, USA

Dr G.R. Douglas, Environmental Health Centre, Department of 
   National Health and Welfare, Tunney's Pasture, Ottawa, Ontario, 

Dr M. Draper, International Programme on Chemical Safety, World 
   Health Organization, Geneva, Switzerland 


Dr E. Elmore, Northrop Services, Inc., Research Triangle Park,
   North Carolina, USA

Dr L.R. Ferguson, Cancer Research Laboratory, Pathology
   Department, The Medical School, Auckland, New Zealand

Dr K. Fujikawa, Drug Safety Evaluation Laboratories, Central
   Research Division, Takeda Chemical Industries, Ltd., Osaka, 

Dr R.C. Garner, Cancer Research Unit, University of York, York, 
   United Kingdom 

Dr H. Glauert, Laboratory for Cancer Research, University of
   Wisconsin Medical School, Madison, Wisconsin, USA

Dr D.K. Gulati, EHRT, Inc., Lexington, Kentucky, USA

Dr G. Hatch, Northrop Services, Inc., Research Triangle Park,
   North Carolina, USA

Dr J. Heinisch, Institute for Microbiology, Darmstadt, Federal
   Republic of Germany

Dr C. Howard, Central Toxicology Laboratory, Imperial Chemical
   Industries, Ltd, Macclesfield, Cheshire, United Kingdom

Dr S. Inge-Vechtemov, Department of Genetics and Breeding,
   Leningrad State University, Leningrad, USSR

Dr M. Ishidate, Division of Mutagenesis, National Institute of
   Hygienic Sciences, Tokyo, Japan

Dr A. Knaap, Laboratory of Carcinogenesis and Mutagenesis,
   National Institute of Public Health, Bilthoven, The

Dr T. Lakhanisky, Institut d'Hygiene et d'Epidemiologie,
   Brussels, Belgium

Mr C.G. Lee, Chemical Defence Establishment, Porton Down,
   Wiltshire, United Kingdom

Prof N. Loprieno, Institute of Biochemistry, Biophysics, and
   Genetics, University of Pisa, Pisa, Italy

Dr D. McGregor, Development Toxicology, Inveresk Research Int., 
   Ltd, Musselborough, Scotland 

Dr B. Margolin, Biometry and Risk Assessment Program, National
   Institute of Environmental Health Sciences, Research Triangle 
   Park, North Carolina 


Dr C. Martin, Cancer Research Unit, University of York, York, 
   United Kingdom 

Dr M. Mercier, International Programme on Chemical Safety, World 
   Health Organization, Geneva, Switzerland 

Dr B.C. Myhr, Department of Genetics and Cell Biology, Litton
   Bionetics, Inc., Kensington, Maryland

Dr A.J. Nelmes, Gallaher Limited, London, United Kingdom

Dr S. Nesnow, Carcinogenesis and Metabolism Branch, Health Effects 
   Research Laboratory, US Environmental Protection Agency, 
   Research Triangle Park, North Carolina 

Dr E.R. Nestmann, Environmental Health Centre, Department of
   National Health and Welfare, Tunney's Pasture, Ottawa, Ontario, 

Dr G. Obe, Institute of General Genetics of the Free University of 
   Berlin, Berlin (West) 

Dr T.J. Oberly, Lilly Research Laboratories, Greenfield Laboratory, 
   Greenfield, Indiana 

Dr F. Palitti, Evolutionary Genetics Centre, Institute of Genetics, 
   Citta Universitaria, Rome, Italy 

Dr S. Parodi, Scientific Institute for Tumours, University of
   Genoa, Genoa, Italy

Dr J. Parry, Department of Genetics, University College of Swansea, 
   Singleton Park, Swansea, Wales, United Kingdom 

Dr B.J. Phillips, British Industrial Biological Research 
   Association, Carshalton, Surrey, United Kingdom 

Dr G. Probst, Lilly Research Laboratories, Greenfield Laboratory, 
   Greenfield, Indiana, USA 

Mr C.R. Richardson, Central Toxicology Laboratory, Imperial 
   Chemical Industries, Ltd, Macclesfield, Cheshire, United Kingdom

Dr E. Matthews, Department of Molecular Biology, Litton Bionetics, 
   Inc., Kensington, Maryland 

Dr T. Sanner, Laboratory for Environmental and Occupational Cancer, 
   Norsk Hydro's Institute for Cancer Research, Oslo, Norway

Dr M.D. Shelby, National Toxicology Program, National Institute of 
   Environmental Health Sciences, Research Triangle Park, North 


Dr J.W.I.M. Simons, State University of Leiden, Leiden, The

Dr J. Styles, Central Toxicology Laboratory, Imperial Chemical
   Industries, Ltd, Macclesfield, Cheshire, United Kingdom

Dr W.A. Suk, Northrop Services, Inc., Research Triangle Park,
   North Carolina

Dr G.F. van Went, Division of Toxicology and Chemical Analysis
   of Foodstuffs, National Institute of Public Health, Bilthoven, 
   The Netherlands 

Dr S. Venitt, Chemical Carcinogenesis Division, Pollards Wood
   Research Station, Bucks, United Kingdom

Dr E. Vogel, Department of Radiation Genetics and Chemical
   Mutagenesis, State University of Leiden, Leiden, The Netherlands

Dr R.C. von Borstel, Department of Genetics, The University of
   Alberta, Alberta, Canada

Dr M.D. Waters, Genetic Toxicology Division, Health Effects 
   Research Laboratory, US Environmental Protection Agency,
   Research Triangle Park, North Carolina, USA

Dr G. Williams, Naylor Dana Institute for Disease Prevention,
   American Health Foundation, Valhalla, New York, USA

Dr F. Würgler, Institute of Toxicology, University of Zurich,
   Schwerzenbach, Switzerland

Dr M.Z. Zdzienicka, Department of Radiation Genetics and Chemical 
   Mutagenesis, State University of Leiden, Leiden, The Netherlands

Dr E. Zeiger, Toxicology Research and Testing Program, National 
   Toxicology Program, National Institute of Environmental Health 
   Sciences, Research Triangle Park, North Carolina, USA


    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 Manager of the 
International Programme on Chemical Safety, 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 so that it may be considered in the 
event of updating of the criteria document. 

                        *    *    *

    Partial financial support for the publication of this criteria 
document was kindly provided by the United States Department of 
Health and Human Services, through a contract from the National 
Institute of Environmental Health Sciences, Research Triangle Park, 
North Carolina, USA - a WHO Collaborating Centre for Environmental 
Health Effects. 


    The first part of this project, dealing with  in vitro studies, 
has already been published by Elsevier, Amsterdam.  The second 
part, concerning  in vivo studies, is expected to be completed and 
evaluated by early 1985, with publication about one year later. 

    The rationale for the collaborative study was derived, to a 
large extent, from the major findings of the "International Program 
for the Evaluation of Short-Term Tests for Carcinogens" (IPESTTC) 
(de Serres & Ashby, 1981).  This study, in turn, arose from the 
necessity to evaluate the efficacy of different short-term assays 
proposed for supplementing the traditional long-term assay in the 
rodent.  The results of the IPESTTC clearly confirmed the value of 
 Salmonella reversion assays as suitable primary tests for potential 
carcinogens and mutagens.  However, it was also confirmed that some 
known rodent carcinogens were either not detected, or only detected 
with considerable difficulty, by such assays.  The IPESTTC study 
did not succeed in defining any complementary eukaryotic assay that 
could be used to detect carcinogens, found to be negative in the 
standard  Salmonella reversion assay.  Several assays showed 
promise, but none could be recommended, because it was considered 
that the supporting data base was too small. 

    It was against this background that the Collaborative Study on 
the Assessment and Validation of Short-Term Tests for Genotoxicity 
and Carcinogenicity (CSSTT) was proposed by the International 
Programme on Chemical Safety (IPCS) and the National Institute of 
Environmental Health Sciences (NIEHS) of the USA, as a 
Participating Institution in the IPCS.  The objective of the study 
was to generate a wide range of test results, using a small group 
of carefully-selected chemicals, which would contribute to an 
empirical basis for the selection of one or more  in vitro short-
term tests to complement the widely-used  Salmonella test, developed 
by Professor Bruce Ames. 

    Some 60 investigators presented nearly 90 individual sets of 
assay results to the collaboration study, generating in all some 
2500 dose-response relationships.  Most of the currently available 
 in vitro eukaryotic assay systems were represented.  The following 
8 organic carcinogens known to be either inactive or difficult to 
detect in the  Salmonella assay were chosen:   o-toluidine, 
hexamethylphosphoramide (HMPA), safrole, acrylonitrile, benzene, 
diethylhexylphthalate, phenobarbital, and diethylstilboestrol, 
together with 2 chemicals, caprolactam and benzoin, for which there 
was no evidence of carcinogenicity in 2-year, 2-species rodent 

    The results of the study were evaluated at a meeting of the 
investigators, held at St. Simon's Island, Georgia, USA, on 
22 - 28 October, 1983.  Each group of assays was chaired by a 

coordinator, and all the original data were discussed, evaluated, 
and agreed.  At these assay group meetings, important protocol 
deficiencies were identified and these findings constitute one of 
the most significant developments arising from the collaborative 

    The findings indicate that carcinogens that are inactive or 
difficult to detect in the  Salmonella assay fall into 2 distinct 
groups.  The first group includes genotoxins that are probably non-
mutagenic to  Salmonella because of deficiencies in the available 
metabolic capacity of the assay system.  Thus, HMPA,  o-toluidine, 
safrole, and acrylonitrile were detected by most of the eukaryotic 
assays studied, indicating that there is a range of assays that can 
complement the  Salmonella mutation assay to a limited extent. 

    The other group of carcinogens, benzene, DEHP, DES, and 
phenobarbital, displayed a more selective range of genotoxic 
activities, and none of the assays was selectively sensitive to 

    Of the assays studied, only the induction of chromosomal 
aberrations, cell transformation, and gene mutation in mammalian 
cells, and aneuploidy in yeast gave encouraging overall 
performances for the 8 carcinogens, and, with the exception of the 
first, formidable protocol deficiencies have to be remedied. 

    The 3 carcinogens, DES, phenobarbital, and DEHP, chosen to 
represent the class of chemicals believed to induce tumours in 
rodents without first modifying the integrity of the nuclear DNA, 
each displayed a range of genotoxic activity.  Thus, the term "non-
genotoxic" should be used only when a sufficiently large 
genotoxicity data base has been established. 

    The collaborative study has provided considerable evidence to 
support the view that  in vitro assays should be classified as 
confirmatory, complementary, and supplementary.  This understanding 
of the potential of an assay should provide a basis for the 
elimination of redundancies in proposed combinations of test 

    The major conclusion of the study was that the use of 
chromosomal aberration assays, preferably in an agreed cell type, 
in conjunction with an adequate assessment of the mutagenicity of a 
chemical for  Salmonella, might provide an efficient primary screen 
for possible new carcinogens.  A first priority should be the 
application of resources to establish a generally acceptable and 
applicable protocol for the conduct of this type of assay.  The 
adoption of a chromosomal-aberration assay as a common 
complementary test has additional advantages in that it would allow 
easy comparison of data, ready extension to supplementary 
cytogenetic assays, and the provision of data derived from an 
independent endpoint from the gene mutations of the  Salmonella 
assay.  Finally, looking back at the developments in this field of 
toxicology, it seems clear that to establish an assay at the level 
of international acceptance requires about a decade of meticulous 
scientific endeavour and international collaboration. 


1.1.  Introduction

    It was discovered in the late 1960s and early 1970s that many 
chemicals underwent metabolic changes before they were capable of 
inducing the processes leading to cancer.  This led to the 
development of  in vitro techniques for bringing about such 
metabolic transformations and, consequently, enormous progress was 
made in the fields of mutagenesis and carcinogenesis.  It soon 
emerged that there was a strong correlation between the 
carcinogenic activity of a chemical, particularly in rodents, and 
its mutagenic properties, as demonstrated in a wide variety of  in 
 vitro and  in vivo experimental systems involving bacteria, yeasts, 
insects, rodents, and mammalian cells in tissue culture.  At the 
same time, there was a growing awareness that some chemicals, for 
example, vinyl chloride monomer, posed hitherto unsuspected health 
dangers.  The fact that increasing numbers of chemicals were being 
shown to have toxic properties in the evolving  in vitro 
genotoxicity test systems, together with a series of disasters 
associated with chemicals that resulted in considerable mortality 
and morbidity, brought about a realization that appropriate 
legislative control of chemicals was needed to ensure adequate 
protection of human health.  Thus, in the latter part of the 1970s, 
there was an unprecedented amount of activity, both nationally and 
internationally, in the field of chemical safety. 

    Public perception about the inadvertent exposure of human 
beings to chemical carcinogens was greatly heightened by the 
development of one particular assay system for the detection of 
mutagens.  This was the  Salmonella typhimurium reversion test 
incorporating a metabolic activation system, pioneered by Professor 
Bruce Ames and his colleagues, and now known universally as the 
 Salmonella assay, or the Ames test.  This assay system, which is 
based on fundamental genetic and molecular biological principles, 
produced test results within a week, and it was soon adopted by 
scientists throughout the world, but more particularly in North 
America and Europe.  As a result, many hundreds of chemicals were 
tested and pronounced on, as is now known, without full 
appreciation of the technical difficulties of the test and the 
biological significance of the results.  The fact that many 
chemicals in common use, ranging from food additives and cosmetics 
to household products, were claimed to have mutagenic properties 
and, hence, by implication could be carcinogens, received a 
considerable amount of uncritical attention in the general 
scientific and lay press. 

    It is generally accepted, and was so for the purposes of this 
study, that the strongest evidence that a chemical is a carcinogen 
is derived from either chemical or epidemiological findings in 
human beings showing an unequivocal relationship between exposure 
and the induction of malignant disease, supported by appropriate 
animal studies; or, in the absence of adequate human exposure data, 
the experimental induction in several rodent species of malignant 
tumours following carefully-controlled systematic exposures to the 

chemical for most of their life span.  These rodent bioassays 
require about 3 years of experimental effort to produce a 
conclusion and are extremely costly.  Thus, the prospect of 
obtaining apparently equivalent information in a far shorter time 
and at a fraction of the cost was immensely appealing.  These hopes 
were reinforced by the claims of confirmatory evidence provided by 
an increasing number of different assay systems.  It was these 
developments that gave rise to the term "short-term tests" to refer 
to assay systems that were believed to indicate carcinogenic 
properties.  It was even hoped that, in time, these assays would 
replace, at least in part, the rodent bioassay.  Although, 
understandably, this idea received much uncritical support, many 
scientists were sceptical about some of the claims made for these 
assay systems, pointing out that there were many discrepancies in 
the findings, so-called false positives and false negatives, and 
there was much debate concerning the true predictive nature of 
assay systems, singly and in combination, for rodent carcinogens 
and the relevance of the findings for human disease. 

    By the early 1980s, many nations had adopted legislation to 
control toxic chemicals, incorporating various test requirements 
for acute and chronic effects and other preventive measures, such 
as adequate labelling.  However, the issue of a legal requirement 
for specific short-term tests for mutagenic and carcinogenic 
properties was usually avoided.  In some cases, the problem was 
recognized by the formulation of a discretionary set of 
recommendations involving data from a battery or tier of assays in 
which the  Salmonella assay was a basic requirement.  The enormous 
international trade in chemicals and the realization that, for most 
of the 50 000 or so chemicals in common use, little systematic 
toxicological data existed, brought about an early appreciation of 
the need to use scarce toxicological resources with maximum 
efficiency and under international agreement.  Because of the 
profound implications for international trade in chemicals, the 
Organization of Economic Cooperation and Development (OECD) 
addressed the problem of ensuring that toxicological data developed 
in one member state could be accepted in all member states.  To 
this end, an extensive series of guidelines for toxicity and other 
testing was developed by international experts.  Further, and of 
great importance, a set of procedures was formulated to ensure good 
laboratory practice and quality assurance.  This major 
international collaboration identified and codified the tests that 
were believed to be necessary to provide sufficient data to ensure 
safety in the use of a chemical.  Obligatory measures and 
procedures were set out to ensure that the tests were carried out 
according to the highest standards, as laid down by international 
agreement.  The implementation of these guidelines and procedures 
has proved to be a difficult and onerous task, even for well-
established toxicological laboratories.  The attempt to set some of 
the newly-developed short-term tests into a similar legislative 
framework revealed many uncertainties, which it now appears can 
only be resolved by international collaborative efforts on a scale 
that has few precedents. 

    In the fields of genetics and molecular biology, from which the 
science of mutagenesis has evolved, and which now have assumed 
great importance for the understanding of carcinogenesis, 
scientists have shown great interest and ingenuity in adapting the 
particular biological systems they use for their research studies 
to assay systems of possible general use.  Unfortunately, what can 
be a powerful and flexible tool in the hands of an experienced 
research worker cannot easily be transformed into the somewhat 
inflexible procedure that is required for a test system for routine 
use throughout the world.  Studies to evaluate the efficiency of 
different short-term tests, which were started in a number of 
countries more or less in parallel with the development of the 
systems, were mostly concerned with the validity of the test 
procedure.  By the time legislative measures were being 
consolidated in many countries, notably the Toxic Substances 
Control Act, 1976, in the USA, legislators were faced with a 
plethora of some 40 short-term tests, all claiming some promise for 
revealing mutagenic or carcinogenic potential.  Of considerable 
concern to industry, at this time, was the desire of legislators to 
have clear-cut criteria for yes/no decisions, even in a rapidly-
evolving subject such as carcinogenesis.  If the Delaney clause 
(United States Public Law, 1958), which states, in part, "that no 
additive shall be deemed to be safe if it is found to induce cancer 
when ingested by man or animal, or if it is found after tests which 
are appropriate for evaluation of safety of food additives to 
induce cancer in man or animals", were strictly applied to the 
wider field of chemicals, the consequences for a society so 
dependent on chemicals could be most serious. 

    It is understandable that, against this background, there was 
considerable incentive and support for cooperative efforts to 
resolve these important issues.  The project, which is the subject 
of this report, follows on from an earlier international study that 
was based on the realization that, as a first step, the 
effectiveness of any short-term test in discriminating between 
carcinogens and noncarcinogens had to be established using, as 
reference, chemicals for which extensive rodent bioassay results 
were available. 

    The original project, called the International Programme for 
the  Evaluation of Short-Term Tests for Carcinogenicity (IPESTTC) 
was carried out between 1977 and 1979 and 42 coded chemicals were 
tested by over 60 scientists using some 30 assay systems.  The 
project arose as a result of initiatives by the Health and Safety 
Executive and the Medical Research Council of the United Kingdom 
and the National Institute of Environmental Health Sciences of the 
USA.  At the time of the planning of the IPESTTC, there was general 
agreement concerning the value of the Ames test, which by this time 
had a generally-accepted protocol.  However, it had also become 
apparent that certain carcinogens or classes of carcinogens failed 
to mutate  Salmonella; hence, there was a need to identify other 
test systems that could complement the Ames assay.  Thus, there 
were 3 basic objectives in the programme.  The first was to obtain 
more systematic knowledge concerning the carcinogens that the Ames 
test failed to detect.  The second was to examine the ability of 

selected test systems to discriminate between carcinogens and 
noncarcinogens, and the third was to define assays that 
complemented the bacterial mutation assays.  To this end, 14 
carcinogen/noncarcinogen pairs, together with 11 other carcinogens 
and 3 other chemically-unrelated noncarcinogens, were specially 
prepared with defined purity and distributed "blind" to the 

    The principal results of the IPSSTTC, which have been published 
in full (de Serres & Ashby, 1981), clearly confirmed the value of 
the  Salmonella assay as a suitable primary test for the detection 
of potential mutagens and carcinogens.  However, it was also 
confirmed that some known rodent carcinogens were either not 
detected, or only detected with difficulty, by this assay.  The 
study did not succeed in arriving at clear-cut conclusions 
concerning a single complementary eukaryotic assay that was capable 
of giving a positive response for the carcinogens found negative in 
the standard  Salmonella assay.  Several assays that might serve in 
this capacity were identified, but none was recommended for general 
adoption, because it was considered that the supporting data base 
was too small.  An important practical aspect of the IPESTTC came 
about through the meetings of investigators, where each assay group 
discussed their results with immediate access to the raw data.  
These discussions resolved discrepancies in the findings and 
produced not only consensus views on the findings, but also 
extremely valuable indications of protocol deficiencies, even for 
the well-established  Salmonella assay. 


    The IPESTTC was remarkably successful in attracting the 
voluntary participation of a large number of scientists, together 
with additional support from scientific institutions.  The 
conclusions from this study indicated clearly that priority should 
be given to the identification of assay systems to complement the 
Ames test.  By 1981, the validity and usefulness of the Ames test 
had been well substantiated, but it had also been established that 
a number of important carcinogens were not detected, or detected 
only with difficulty using this assay.  It was, thus, generally 
accepted that no single assay system could be relied on to detect 
all carcinogens.  This led to the proposal of the adoption of 
testing schemes involving multiple  in vitro assays in various 
configurations such as batteries, tiers, or combinations of the 
two.  The basis for the selection and deployment of these multiple 
tests was, and remains, theoretical prudence rather than empirical 
evidence.  That is, a variety of genetic end-points and organisms 
representing different phylogenetic levels were selected with the 
intent of not missing end-points or phylum-specific chemical 
activity.  Furthermore, in general, reliance was placed on the 
deployment of genotoxicity assays, even though, by this time, other 
factors were assuming importance in the biological etiology of 
natural and chemically-induced cancer.  Thus, the molecular targets 
for investigation were no longer dominated by observations of 
readily discernible changes in the sequence or integrity of nuclear 
DNA, but involved consideration of subtle changes in chromosome, 
gene, or oncogene function or expression (Klein, 1981; Reddy et 
al., 1982; Tabin et al., 1982; Weiss, 1982).  This implies that 
some of the genetic end-points monitored in assay systems may 
ultimately be shown not to be directly related to the critical 
events in the etiology of some chemically-induced cancers (Cairns, 
1981).  Amidst these scientific controversies about the reliability 
and biological significance of many short-term tests, registration 
and health authorities were endeavouring to assess the genotoxic 
data from the same short-term tests without adequate scientific 
guidance for the interpretation of the all too frequently 
discordant data. 

    It was against this background that the Collaborative Study on 
Short-Term Tests for Genotoxicity and Carcinogenicity (CSSTT) was 
proposed by the International Programme on Chemical Safety (IPCS) 
and the National Institute of Environmental Health Sciences of the 
USA, as a Participating Institution in the programme.  The general 
goals and designs of the study were outlined by an  ad hoc Working 
Groupa, which met at the invitation of the IPCS in Geneva, on 
30 April - 1 May 1981.  The plans were consolidated by an IPCS 
Working Groupb, which met in Geneva, on 13 - 14 November 1981.  The 

a Participants:  Dr J. Ashby, Professor N.P. Bochkov, Dr B.E. Matter, Professor T. 
  Matsushima, Dr F.J. de Serres, Dr M. Shelby, and Professor F.H. Sobels.
b Participants:  Dr J. Ashby, Dr G.R. Douglas, Dr M. Ishidate, Dr A. Leonard, Dr N. 
  Loprieno, Dr B.E. Matter, Professor T. Matsushima, Dr R. Montesano, Dr F.J. de Serres, 
  Dr M. Shelby, Professor F.H. Sobels, Dr M. Stoltz and Dr M. Waters.
subsequent coordination of the collaborative study was the 
responsibility of a Steering Committee derived primarily from the 
Working Group (J. Ashby; F. de Serres, Chairman; M. Ishidate Jr; 
B. Margolin; B. Matter; M. Shelby; and M.H. Draper, IPCS). 

    The financial burden of the organization of this study was met 
largely by the IPCS, together with some of its Participating 
Institutions, particularly the National Institute of Environmental 
Health Sciences in the USA.  As in the IPESTTC project (de Serres & 
Ashby, 1981), the funding of the assay work was provided, in the 
majority of cases, by the individual investigators managing to 
incorporate the work into their research programmes.  This could 
only occur with the goodwill and belief in the project of the 
senior managements of the approximately 50 involved laboratories 
from universities, research institutes, and industrial research 
facilities, throughout the world.  In addition, a number of 
governments that support the IPCS provided financial assistance for 
this study.  These include the governments of Belgium, Italy, the 
Netherlands, and, in particular, the United Kingdom. 

    The experience gained from the conduct of the IPESTTC and the 
goodwill of the participants in that study were extensively drawn 
on in the planning and organization of the CSSTT.  The major 
objective was defined as the generation of a wide range of test 
results for a small group of carefully-selected rodent carcinogens 
that would contribute to an empirical basis for selecting one or 
more  in vitro short-term tests as complementary to the Ames test.  
The number of chemicals was kept to a minimum, because of the 
experience of handling the 42 chemicals in the IPESSTC study.  It 
was argued that it was better to aim for an extensive data base for 
a few chemicals than a reduced and patchy data base for a large 
number.  The chemicals were selected with much care, and those 
chosen were all known to be particularly difficult to detect in 
assay systems and, thus, would be expected to expose weaknesses and 
inconsistencies in both the assay system and the protocols.  This 
indeed proved to be the case. 

    Some 60 investigators participating in the project carried out 
nearly 90 individual sets of assays, generating, in all, some 2500 
dose-response relationships.  Most of the  in vitro eukaryotic 
tests, currently available, were represented.  The 8 organic 
carcinogens chosen as either inactive, or difficult to detect as 
positive in the  Salmonella assay, were:   o-toluidine, 
hexamethylphosphoramide, safrole, acrylonitrile, benzene, 
diethylhexylphthalate, phenobarbital, and diethylstilboestrol, 
together with 2 chemicals, caprolactam (Huff, 1982) and benzoin 
(NTP, 1980), which had not shown any evidence of carcinogenicity in 
2-year rodent bioassays.  The criteria for the selection of these 
chemicals is an important matter and the reasons for each inclusion 
are given in the following section. 


    Eleven carcinogens were defined as either difficult or 
impossible to detect as bacterial mutagens in the IPESTTC study, 
and 4 of these were selected for the CSSTT.  These were 
hexamethylphosphoramide, safrole, diethylstilboestrol, and 
 o-toluidine.  The carcinogenicity of these agents for rodents is 
well-established but only a small proportion of the  Salmonella  
assays conducted in the IPESTTC study detected them as mutagenic 
(1/15, 4/17, 1/17, and 3/16, respectively).  However, several of 
these responses were not reproducible, and each was weak.  This led 
to the compounds being regarded as essentially non-mutagenic for 
 Salmonella in this study. 

     Hexamethylphosphoramide (HMPA)

    Hexamethylphosphoramide (HMPA) is among the most potent of 
animal carcinogens producing metastasizing nasal tumours in rats 
exposed by inhalation. 

    There is a wealth of data indicating it to be non-mutagenic for 
 Salmonella, yet the results of the IPESTTC study suggested that it 
was a general genotoxin in eukaryotic assays.  One possible mode of 
action for this agent is via the enzyme-mediated formation of 
formaldehyde (Ashby & Lefevre, 1983).  This is also a rat nasal 
carcinogen that is difficult to detect as mutagenic in the 
 Salmonella assay but is a gene mutagen in human cells (Ashby & 
Lefevre, 1983; Goldmacher & Thilly, 1983). 


     o-Toluidine is a relatively weak rodent hepatocarcinogen.  Its 
activity in this respect is interesting because it weakens the 
earlier assumption that single ring aromatic amines, as opposed to 
multiple ring arylamines such as 2-naphthylamine and 4-
aminobiphenyl, are non-carcinogenic.   o-Toluidine was established 
as difficult or impossible to detect in the  Salmonella assay in 
the IPESTTC study, which also showed that it was mutagenic to these 
bacteria, if evaluated in the presence of norharman.  These 
collected findings suggested  o-toluidine to be a general 
genotoxin, which requires specific metabolic activation, rather 
than an agent showing specificity of genetic action. 


    Safrole is a weak rodent liver carcinogen and has been studied 
extensively in the  Salmonella assay.  Although certain 
investigators have reported it to be mutagenic, it is generally 
found inactive in this assay.  Both the alpha-acetoxy and the 
sidechain epoxide derivatives are mutagenic, and these have been 
suggested as the metabolites responsible for the carcinogenic 
action observed.  Safrole may, therefore, be a further example of a 
general genotoxin that requires specific metabolic activation.  Set 
against this is the fact that it appears devoid of genetic activity 
 in vivo; thus, it gave a negative response in both the mouse bone-

marrow micronucleus assay and the  in vivo rat liver unscheduled DNA 
synthesis (UDS) assay.  Consequently, the possibility cannot be 
excluded that the tumours produced by this agent may be mediated 
via some disturbance of normal homeostasis in the test animals 
(i.e., by a non-genotoxic mechanism), despite its ability to induce 
genetic changes in some  in vitro test systems. 

     Diethylstilboestrol (DES)

    Diethylstilboestrol (DES) is carcinogenic in both human beings 
and experimental animals.  It could have been selected for this 
study simply on the basis of a recent paper that showed it to be 
capable of transforming cells and inducing chromosomal damage in 
the apparent absence of gene mutations (Barrett et al., 1981).  
This finding was supported by the fact that, in the IPESTTC study, 
DES was regarded by the investigators as a clastogen that was non-
mutagenic for  Salmonella.  Therefore, DES, together with benzene 
(see below), were included in the study as agents that could 
possibly demonstrate the reality of the genetic specificity of 
action of some chemical carcinogens. 


    Benzene is a unique carcinogen.  Its possible leukaemogenic 
activity in man has been discussed for many years, yet this effect 
has been difficult to reproduce in animals.  The compound is 
nonetheless generally regarded as carcinogenic and extensive data 
exist on its clastogenicity, particularly when evaluated  in vivo.  
Dean (1978) reviewed the literature on the genotoxicity of this 
agent in short-term tests, and this, together with subsequent 
studies, clearly defined it as non-mutagenic for bacteria.  The 
possibility of its complete inability to induce gene mutations  in 
 vitro is implied in some papers, but its gene mutagenicity  in vivo  
has not yet been assessed. 


    Similarities in structure between acrylonitrile and the 
carcinogen vinyl chloride led Venitt to evaluate it for bacterial 
mutagenicity.  The debate that ensued in  Mutation Research (Milvy & 
Wolff, 1977; Venitt et al., 1977) regarding the mutagenic activity 
of this agent in  Salmonella and  Escherichia coli can be summarized 
by describing acrylonitrile as a chemical that could easily be 
found non-mutagenic in a routine screening programme that employed 
only bacteria as marker cells.  The carcinogenicity of this agent 
has been subsequently defined and reviewed.  The question of 
whether acrylonitrile interacts directly with DNA via a Michael 
reaction, or via the intermediate metabolic formation of an epoxide 
derivative, heightens interest in this agent. 

     Diethylhexylphthalate (DEHP)

    Diethylhexylphthalate (DEHP) has been shown to produce 
hepatomas in the rodent liver, yet the majority of experimental 
data indicate it to be non-mutagenic for bacteria.  It has been 

proposed that the carcinogenicity of this agent is associated with 
its ability to proliferate peroxisome microbodies in the rodent 
liver (Moody & Reddy, 1978).  This explanation would not require 
DEHP itself to interact with nuclear DNA.  The carcinogenicity of 
DEHP has, therefore, been considered as possibly "epigenetic" in 
origin, which increases the need to determine accurately its 
genotoxic status  in vitro.  The extent to which DEHP is hydrolysed 
to the corresponding mono-acid derivative (MEHP) could influence 
the outcome of certain assays as the latter chemical, unlike the 
former, is reported to be a clastogen and SCE-inducing agent  in 
 vitro (Phillips et al., 1982; Tomita et al., 1982; Ashby, 1983). 


    Phenobarbital, although active as a rodent liver carcinogen, 
also has significant tumour-promoting properties in the rodent 
liver.  In fact, the issue of whether phenobarbital is a pure 
promoting agent devoid of cancer-initiating activity is of great 
current interest.  In contrast to DES, the rodent carcinogenicity 
of phenobarbital appears not to be reflected in man, despite the 
extensive and controlled exposure of epileptic patients (Clemmesen 
& Hjalgrim-Jensen, 1977).  Although this chemical is generally 
regarded as non-genotoxic, limited evidence exists for its ability 
to induce SCEs  in vitro (Athanasiou & Kryrtopoulos, 1982; Ashby, 
1983).  This property may be related to its ionic composition (cf. 
sodium saccharin, MEHP above, lacchaic acid, sodium benzoate, etc., 
for similar activity profiles) (Ashby, 1983).  An additional point 
of interest in this chemical is that Williams has presented data to 
support the claim that phenobarbital is an example of an epigenetic 
carcinogen (Williams, 1981). 

     The non-carcinogens caprolactam and benzoin

    The selection of non-carcinogens suitable for use in the 
evaluation of short-term tests has presented a stumbling block in 
all validation exercises.  In the early validation studies, non-
carcinogens were simply selected from compounds commonly regarded 
as being non-carcinogenic.  In some cases, no data existed 
regarding their carcinogenicity, and this was taken as indicative 
of inactivity.  In the IPESTTC study, the non-carcinogens selected 
were graded according to the extent and quality of the negative 
data, and although this was an advance, the interpretation of 
unexpected positive assay responses was difficult.  This issue is 
particularly important in relation to the widespread reference to 
false-positive responses occurring in short-term tests; the 
credibility that can be accorded to a false-positive response is 
directly proportional to the certainty associated with the 
compound's classification as a non-carcinogen.  The fact that some 
assumed non-carcinogens may eventually be classified as either weak 
or organ-, strain-, sex-, or species-specific carcinogens might 
lead to the re-evaluation of many previous examples of false-
positive assay responses. 

    In order to circumvent this problem in the current 
collaborative study, particular attention was paid to the selection 
of the 2 chemicals required to act as negative controls.  The 
agents selected were benzoin and caprolactam.  The major criterion 
for their selection was inactivity in recent cancer bioassays 
conducted as part of the US National Toxicology Program.  In the 
reports of these studies (US NTP, 1980, 1982), it was concluded 
that neither compound was carcinogenic in male or female Fischer 
344 rats or B6C3Fl mice dosed at levels up to the maximum tolerated 
dose over their lifetimes.  These 2 studies were taken as 
definitive as they represented the most detailed cancer bioassay 
protocols currently in use.  In addition, these agents were devoid 
of overtly DNA-reactive substituents and were known to be non-
mutagenic for bacteria. 

    The 10 chemicals selected covered a wide range of structural 
types and could, therefore, be considered representative of agents 
encountered in the environment and chemical industries.  In 
addition, several of the carcinogens selected had been associated, 
by other investigators, with possible mechanisms of cancer 
induction other than the DNA-reaction/somatic mutation theory.  
Finally, the 2 non-carcinogens were sufficiently well supported by 
negative carcinogenicity data to ensure that clear decisions could 
be made, regarding the significance of their genotoxic activity,  in 


    As analytical techniques improve, it is possible to find trace 
impurities in materials, formerly considered pure.  Set against 
this is the practical need to obtain large supplies of pure 
chemicals for a study, such as the present collaborative study, 
without inordinate costs and delays.  This dilemma is heightened by 
the history of the conduct of cancer bioassays where the test 
chemical was often, if not usually, assumed to be pure in the 
absence of appropriate analytical data.  Many chemicals bioassayed 
for carcinogenicity have been of technical quality and, therefore, 
probably not more than 95% pure.  Normally, this would not matter, 
but when the cancer bioassay data are to be the ultimate reference 
point, as in the present study, then the relative purity of the  in 
 vitro test chemical becomes of importance.  A 5% impurity, at high 
doses, could be of biological significance.  At one extreme, it can 
be argued that material of similar purity (or impurity) to that 
employed in the cancer bioassay should be assayed, but this may 
lead to a further confounding of the total data base.  At the other 
extreme, it can be suggested that only absolutely pure materials 
should be employed  in vitro, whatever the cost and inconvenience 
incurred in their preparation.  This approach carries the penalty 
that the carcinogenic response observed in mammals may have been 
produced by impurities, in which case, activities observed  in vitro  
may not be correlated with carcinogenic activity (or inactivity). 

    This consideration is particularly relevant for benzene.  The 
most convincing carcinogenicity data for benzene were derived from 
human beings exposed to it together with other chemicals, the 
number and type of which varied from situation to situation.  The 
fact that the carcinogenicity of this chemical is difficult to 
define in rodents has led to the suspicion that it may not be 
benzene, but the chemicals used in association with it, that are 
carcinogenic.  Pure benzene was used in this study; a risk was 
taken by doing so. 

    The purity criteria adopted for the present study entailed the 
following assays of chemical purity: 

    a)  One batch of each chemical of the highest grade commercial 
        samples available, usually 99% or more pure, was obtained.

    b)  The proton nuclear magnetic resonance spectrum, mass
        spectrum, and infrared spectrum were determined and checked 
        for consistency with the proposed structure and for the 
        possible presence of impurities. 

    c)  The elemental analysis (C, H, and N) was determined for 
        both liquids and solids; each was within 0.4% of the 
        theoretical value. 

    d)  The melting point was determined and compared with
        previously reported values for all solids.  Because of 
        differences in thermometer calibrations, variations of less 
        than 4 °C were hard to interpret. 

    e)  In 2 cases (safrole and  o-toluidine), high-pressure liquid 
        chromatography (HPLC) was employed to evaluate trace 
        impurities seen by earlier assay methods. 

    f)  Thin-layer chromatography (TLC) was undertaken on each 
        material, as appropriate.  A variety of eluants and 
        detection systems was employed. 

    On the basis of the above determinations, the present chemicals 
were deemed to be pure to a level of 99%.  These techniques cannot 
eliminate the chance that some activities observed for some of the 
agents (both carcinogens and non-carcinogens) were due to 
impurities.  This admission is necessary, but is not exceptional, 
given the paucity of analytical data usual in such studies, 
including the reference cancer bioassays.  Nonetheless, trace 
impurities may have contributed to some activities, the weak gene 
mutagenicity of phenobarbital in  Salmonella being an example of 
where further purification and reassaying  in vitro might yield 
useful additional data.  Genotoxic impurities should not, however, 
be too easily invoked to explain unexpected genotoxic responses. 
First, similar concerns should apply to positive responses observed 
 in vitro for mammalian carcinogens, and second, such uncertainties 
reflect equally on previous studies, the findings of which 
constitute most of the established data base of this science.  The 
chemicals, made up in 5-g lots, were labelled and distributed to 
the investigators in specially-sealed double containers.  As these 
chemicals were carcinogens and for various other reasons, they were 
not distributed "blind".


    In order to qualify as a complementary assay for routine use in 
conjunction with the  Salmonella plate-incorporation assay, a test 
must have fulfilled the following requirements (Ashby et al., 

    (a)  It should have been successfully employed as a short-term 
         test in a number of laboratories, and should be 
         substantially represented in the literature. 

    (b)  It should have performed well in the detection of the 
         present 8 carcinogens, while concomitantly finding both of 
         the non-carcinogens negative. 

    (c)  Positive responses obtained with the 8 carcinogens tested 
         should have been unambiguous, dose-related, and 

    (d)  Similar qualitative responses should have been observed by 
         the majority of the laboratories using the same assay.

    (e)  It should be appropriate for routine screening purposes, 
         i.e., not unduly demanding as far as resources and 
         technical facilities are concerned. 

    Four categories of assay may thus be defined:

    (1)  Assays suitable for general use in conjunction with the 
          Salmonella assay; 

    (2)  Promising assays, i.e., assays that may be capable of
         fulfilling criteria (a)-(e), but for which data are not 
         available for all of the test chemicals, or where repeat 
         studies are not available; 

    (3)  Relatively new assays that, while not meeting criterion 
         (a), have performed well in the collaborative study, and 
         for which the present 10 chemicals form the greater part 
         of the available data base; these cases would best be 
         handled by the rapid and coordinated acquisition of 
         further information. 

    (4)  Assays that are clearly inappropriate for routine use in 
         testing for potential carcinogens, i.e., that do not meet 
         criteria (b)-(e). 


    As discussed above, the design of the present collaborative 
study reflected the primary purpose of attempting to identify  in 
 vitro eukaryotic assays, which are capable of detecting chemical 
carcinogens, not readily detectable using bacterial assays.  At the 
organism level, 4 categories of assays were employed:  bacteria, 
yeast, fruit flies, and cultured mammalian cells.  Within each of 
these groups of organisms, a variety of test end-points were used. 
Organisms and end-points will be described briefly and are 
presented in Table 1.  Full details are available in the published 
assay working group reports and the reports of individual 
investigators (Ashby et al., 1985). 

Table 1.  IPCS CSSTT test systems
I.   Bacteria

      Salmonella typhimurium

     TA97, TA98, TA100, TA102,  HIS-   HIS+
     TA1535, TA1537, TA1538

     TM677                      AZAS   AZAR

II.   Fungi


           Saccharomyces cerevisiae

          XV185-14C             ARG-   ARG+;  TRP-  TRP+
                                HIS-   HIS+;  HOM-  HOM+

          RM52                  HIS-   HIS+
          D7                    ILV-   ILV+

          D6 and D61-M          ADE-   ADE+   ILV-  ILV+

          D5                    small colonies due to mitochondrial 

           Schizosaccharomyces pombe

          P1                    red    white colonies (ADE)

           Aspergillus nidulans

          35                    methionine metabolism mutants

Table 1.  (contd.)
II.   Fungi (contd.)


           Saccaromyces cerevisiae

          JD1                   gene conversion, tryptophan or 
                                histidine prototrophy

          D7 and D7-144         crossing-over, red and pink 
                                colonies (ADE) gene conversion, 
                                tryptophan prototrophy

          PV-2 and PV-3         crossing-over, canavanine 
                                resistance gene conversion, lysine 

          D6 and D61-M          crossing-over, cycloheximide 
           Aspergillus nidulans

          P1                    crossing over, green yellow 


           Saccharomyces cerevisiae

          D6 and D61-M          red, cycloheximide sensitive white 
                                cycloheximide resist

           Aspergillus nidulans

          P1                    yellow sectors in green colonies

          Illegitimate mating:

           Saccharomyces cerevisiae

          PV-4a and PV-4b       mating type a

III.   Drosophila

           Somatic cell mutations

             wing-mosaicism     wing spots from mutations, 
                                deletions, chromosome breakage, 
                                mitotic recombination or aneuploidy 

             white-zeste eye    eye spots from mutations or 
             mosaicism          deletions

Table 1.  (contd.)
III.   Drosophila (contd.)

             white/white coral  eye spots (same events as wing 
             eye mosaicism      spots above)

IV.   Cultured mammalian cells

     Metabolic cooperation

          V79                   survival of HGPRT- cells


          SHE                   colony assay
          C3H10T1/2             focus assay
          BALB/c 3T3            focus assay
          SHE/SA7               viral enhancement of chemical 
                                transformation-focus assay
          RLV/FRE               enhanced survival of Rauscher 
                                virus-infected rat embryo cells
          CHO                   invasive growth in agar 

   DNA damage

     single-strand breaks

         CHO                    alkaline sucrose sedimentation
         rat hepatocytes        alkaline elution

   Unscheduled DNA synthesis

         HeLa S3                scintillation counting-extraced 
         rat hepatocytes        scintillation counting -DNA
                                extracted from isolated nuclei

         rat hepatocytes        autoradiography

   Cytogenetic damage

     chromosomal aberrations

         CHO                    structural aberrations; micronuclei
         Chinese hamster lung,  structural aberrations; polyploidy
         Chinese hamster,       structural aberrations; polyploidy
         liver,CHl-L            aneuploidy
         rat liver, RL4         structural aberrations; polyploidy
         human lymphocytes      structural aberrations

Table 1.  (contd.)
   Sister chromatid exchange

         rat liver, RL4

   Gene mutations

         L5178Y                 TK+/-  TK-/-
                                OUAS   OUAR

         V79          HGPRT+    HGPRT-
         V79          OUAS      OUAR

         CHO          HGPRT+    HGPRT-    

   Human lymphoblasts                

         TK6          TK+/-     TK-/-

         AHH          HGPRT+    HGPRT-

6.1.  Bacteria

    The carcinogens included in the collaborative study were 
selected on the basis of previously-published results indicating 
their lack of activity in routinely-conducted  Salmonella  
mutagenicity tests.  Five sets of  Salmonella data were obtained in 
this study to confirm the previous results and to provide bacterial 
mutagenicity data on the batches of chemicals used in the current 
study.  Test data are reported for  Salmonella typhimurium strains 
TA97, TA98, TA100, TA102, and TA1535 in both pre-incubation and 
plate incorporation protocols and TA1537 and TA1538 in the plate 
assay only.   S. typhimurium strain TM677 was used to detect 
azaguanine resistant forward mutants, employing a treat and plate 

6.2.  Fungi

    Fungal systems, which offer the advantages of being both 
microbial and eukaryotic, were used to evaluate a wide range of 
genetic end-points.  Test results from  Saccharomyces cerevisiae, 
 Schizosaccharomyces pombe, and  Aspergillus nidulans are reported in 
relation to the following genetic end-points:  nuclear gene 
mutation (both forward and reverse), mitochondrial mutation, gene 
conversion, mitotic crossing over, and aneuploidy. 

6.3.  Drosophila

    Three separate laboratories reported test results from 3 newly-
developed assays for detecting genetic damage induced in somatic 
cells of  Drosophila.  The white-zeste eye mosaicism test detects 
eye spots resulting from mutations or deletions, while the wing 
mosaicism and white/white coral eye mosaicism tests detect wing or 
eye spots resulting from mutations, deletions, chromosome breakage, 
mitotic recombination, or aneuploidy. 

6.4.  Cultured Mammalian Cells

    Five major categories of chemically-induced effects were 
reported for cultured mammalian cells:  inhibition of metabolic 
cooperation, transformation, DNA damage, cytogenetic effects, and 
gene mutations. 

    Test data on the inhibition of metabolic cooperation, an assay 
intended to detect promoting agents, as evidenced by increased 
survival of HGPRT- V79 cells in the presence of an excess of HGPRT+ 
cells and 8-azaguanine or 6-thioguanine, were reported by 3 

    Six distinct transformation assays were reported including 
those from 2 laboratories using Syrian hamster embryo (SHE) cells, 
2 using C3H10T1/2 mouse cells, and single laboratories using the 
Syrian hamster embryo/Simian adenovirus-7 and Rauscher leukaemia 
virus-infected rat embryo cell assays.  In addition, data for 5 
compounds derived from a new assay in which the end-point was 
invasive growth of CHO cells in soft agar were considered.  Two 
investigators, who had offered to generate data using the BHK21 
transformation assay, withdrew from the study because of lack of 
adequate time.  This was disappointing as they had presented the 
prospect of a link with the IPESTTC study in which the BHK21 assay 
represented the sole transformation end-point. 

    The chemical induction of DNA single-strand breaks was 
determined by assessing single-strand breaks using alkaline elution 
or alkaline sucrose sedimentation.  Tests for unscheduled DNA 
synthesis were reported using protocols involving both 
scintillation counting and autoradiography. 

    A large body of test data was reported for the 2 most commonly 
used cytogenetic end-points, structural aberrations, and sister 
chromatid exchanges.  In addition, limited results were reported 
for the induction of micronuclei, aneuploidy, and polyploidy. 

    Gene-mutation induction data were reported for 3 loci: 
thymidine kinase (TK), hypoxanthine guanine phosphoribosyl 
transferase (HGPRT), and NA+, K+ ATPase (Ouabain resistance) in 
mouse, Chinese hamster, or human cells.  These studies included 7 
sets of test results from the L5178Y TK+/- system. 


    The investigators met at St. Simon's Island, Georgia, USA from 
23 - 28 October 1983.  During this meeting, each group of assay 
participants, with the assay coordinator as chairman, discussed the 
results with the raw data in front of them and individual results 
were agreed.  The group then formulated a consensus report on the 
response of each chemical in the assay.  These decisions were 
incorporated into the coordinators' report, prepared during group 
discussions on the overall performance of the assay and any defects 
discovered.  The coordinator's reports were presented and discussed 
at plenary sessions, during which the conclusions and 
recommendations of the study were developed.  The coordinators 
subsequently finalized their reports after further consultations 
with the members of the group.  The reports have been incorporated 
into the text of the publication, which includes all the individual 
reports of results as well as an editorial overview of the study 
and a number of technical appendices (Ashby et al., 1985). 

    Table 2 includes, in summary form, all the agreed results for 
each chemical in each test system in the study, as established in 
the assay group discussions at St. Simon's Island.  In order to 
make some attempt at an overall assessment of assay performances, 
the editorial group proceeded along the following lines.  First, 
the qualitative responses displayed in Table 2 were assumed to be 
correct. Some of these results were unconfirmed and may, therefore, 
represent false-positive or false-negative observations.  Second, 
it was decided that an assay should be capable of detecting at 
least 2 of the selected carcinogenic test agents as positive, 
before it could be assessed for possible use as a complementary 
test.  The extent to which inadequacies of individual test 
protocols, as opposed to the insensitivity of the particular assay 
or its genetic end-point, were responsible for negative responses 
could only be discussed in cases where the same assay had been 
conducted in 2 or more laboratories.  Third, statistical 
comparisons of the overall performance of assays were not 
undertaken because, with the present rather unusual set of test 
chemicals, this could yield meaningless if not misleading 
conclusions, unless undertaken in depth.  The entire data base was 
entered into a computer file at NIEHS and detailed statistical 
analyses may be undertaken, as appropriate. 

    In developing the discussion, it was further accepted that  in 
 vitro assays are, by their constitution, only appropriate for the 
identification of potential carcinogens; that is,  in vitro tests 
can be used to predict possible carcinogens, but not to define 
them; at present, this can only be attempted by  in vivo techniques.  
It is, therefore, to be expected that certain agents will show 
activity  in vitro, but will be unable to express this potential  in 
 vivo, because of their non-absorption, rapid excretion, 
preferential detoxification, inappropriate partitioning, etc., in 
mammals.   In vitro assays cannot, and should not, be expected to 
reveal these possibilities, which are, by definition, unique to 
living animals.  Activity seen  in vitro for the present 2 non-
carcinogens was not used when assessing the overall performance of 

the assays in question (Table 3), but rather, was used to emphasize 
the true role and generic predictive weaknesses of  in vitro assays.  
On occasions, the assimilation of the large data base was made 
easier by considering 4 carcinogens selected for the study as a 
group (HMPA, o-toluidine, safrole, and acrylonitrile).  This was 
because it was known, at the outset, that they were more likely to 
be detected by most assays, as each had already been established as 
being genotoxic, though they were usually inactive in the 
 Salmonella mutation assay.  The remaining carcinogens, with the 
possible exception of benzene, were loosely regarded as non-
genotoxic, prior to this study.  The collaborative study data base 
generally supported the segregation of these 2 groups of 
carcinogens and, thus, enabled a selective assessment of each assay 
to be made.  Some assays performed well with the first 4 
carcinogens but poorly with the others, and some were insensitive 
to this division and performed either generally well or poorly.  A 
possible further subdivision of the second 4 carcinogens became 
evident as the review progressed and this will be discussed in the 
section dealing in more detail with the assays. 
Table 2.  IPCS CSSTT  in vitro study: summary of qualitative resultsa,b
ASSAY                        ACN  TOL  HMPA  SAF  DES  BEN  PB  DEHP  ZOIN  CAP
1.1.1  Salmonella            ?    N    N     N    N    N    P   N     N     N
1.1.2  Salmonella            P    N    N     N    N    N    N   N     N     N
1.1.3  Salmonella            N    N    N     N    N    N    N   N     N     N
1.1.4  Salmonella            P    P    N     N    N    N    P   N     P     N
1.1.5  Salmonella            N    N    N     N    N    N    P   N     N     N

2      FUNGI
2.1     Mutation
2.1.1  D7                    N    N    P     N    N    N    N   N     N     N
2.1.2  Asper 35                   N    N     N         N
2.1.3  D7                    N    N    N     N    N    P    N   N     N     N
2.1.4  XV185                 P    P    P     P         P    P   P     P     P
2.1.5  XV185                      N    P     P                        P     P
2.1.6  P1                    N    N    N     P    N    N    N   N     N     N
2.1.7  D6                    P    N    P     N    N    N    N   N     N     N
2.1.8  D61-M                 P    P    P     P    N    N    N   N     N     N
2.1.9  Mito. D5              P    P    N     P    N    P    N   N     N     N
2.2     Gene conversion
2.2.1  D7                    P    N    P     N    N    N    N   P     N     N
2.2.2  D7                    P    N    N     P    N    N    N   N     N     N
2.2.3  D7-144                P    P    P     P         P    P   P     P     P
2.2.4  PV-3                  N    N    N     N    N    N    N   N     N     N
2.2.5  PV-2                  N    N    N     N    N    N    N   N     N     N
2.2.6  JD-1                  P    N    N     N    N    N    N   N     N     N
2.2.7  D7                    P    N    P     N    N    N    N   P     N     N
2.3     Crossing-over
2.3.1  D7                    N    N    N     N    N    N    N   N     N     N
2.3.2  Asper. 35             N    N    N     N    N    N    N   N     N     N
2.3.3  D6                    P    N    N     N    N    P    N   N           N

Table 2 (contd.)
ASSAY                        ACN  TOL  HMPA  SAF  DES  BEN  PB  DEHP  ZOIN  CAP
2.3.4  D61-M                 P    N    N     N    N    N    N   N     N     N
2.3.5  D61-M                 P    N    N     N    N    N    N   N     N     N
2.3.6  D7                    N    N    N     P    N    N    N   P     P     N
2.4     Aneuploidy
2.4.1  D6                    P    P    P     P    P    P    P   P     N     N
2.4.2  D61-M                 N    N    N     N    P    P    N   N     N     ?
2.4.3  D61-M                 P    P    P     P    P    P    P   P     N     N
2.4.4  Asper. 35             P    N    N     P    N    N    N   N     N     N

3.1.1  Wing spots            P    P    P     P    N    P    N   N     N     P
3.1.2  Eye spots             P    N    P     N    N    N    N   ?     N     P
3.1.3  Eye spots             P    P    P     P    ?    ?    N   ?     N     P

4.     CULTURED MAMMALIAN CELLS (endpoints other than gene mutation)
4.1     Metabolic cooperation
4.1.1  V79                   P    P    N     ?    N    N    ?   P     ?     N
4.1.2  V79                        P    N     N         N                    N
4.1.3  V79                   P    N    N     ?    N    N    N   N     N     N
4.2     Transformation
4.2.1  BALB/C                N    N    N     N    N    N    N   N     N     N
4.2.2  C3H                   P    P    P     P    P    ?    ?   P     P     P
4.2.3  C3H                        P    ?     p         ?                    N
4.2.4  SHE                   P    P    P     ?    N    P    P   P     N     ?
4.2.5  SHE                   P    P    P     P    P    P    N   P     N     P
4.2.6  SHE/SA7                    P    ?     N    ?    N        ?           N
4.2.7  Rl-FRE                     ?    P     N    P             P     N     N
4.2.8  CHO                        N    ?     N         N                    N
4.3     Single-strand breaks
4.3.1  Rat Hepat.            P    P    N     P    P    N    N   N     P     N
4.3.2  CHO                   N    P    N     P    P    P        N     N     N
4.3.3  CHO                   P    P    N     N    P    N    N   N     N     N
4.4     Unscheduled DNA synthesis (UDS)
4.4.1  Rat Hepat. (autorad)  N    N    P     N    N    N    N   N     N     N
4.4.2  Rat Hepat. (autorad)  N    N    N     N    N    N    N   N     N     N
4.4.3  Rat Hepat. (scint.)   P    P    P     P    N    P    N   P     P     N
4.4.4  HeLa (scint.)         N    P    P     P    N    N    N   N     N     N
4.4.5  HeLa (scint.)              P    ?     P         N                    N
4.5     Chromosomal aberrations
4.5.1  CHO                   P    P    N     N    P    N    P   N     N     N
4.5.2  CHO                        N    N     P         P                    N
4.5.3  CHO                   P    N                    N    P               N
4.5.4  LYM                             P          P    P              N     P
4.5.5  CH1-L                 P    P    P     N    P    N    P   N     N     N
4.5.6  CHL                   P    P    P     P    N    P    ?   N     P     P
4.5.7  RL4                   N    P    N     N    N    N    N   N     N     N
4.6     Sister chromatid exchange
4.6.1  CHO                   P    P    P     P    N    N    N   N     N     N
4.6.2  CHO                        P    N     N         N                    N
4.6.3  CHO                   P    N                    N    P               N

Table 2 (contd.)
ASSAY                        ACN  TOL  HMPA  SAF  DES  BEN  PB  DEHP  ZOIN  CAP
4.6.4  CHO                        N    N     N         N        N           N
4.6.5  V79                        P    P     P         N                    N
4.6.6  RL4                   N    P    P     N    N    N    N   N     N     N
4.7     Micronucleus
4.7.1  CHO-MN                P    N    N     N    N    N    N   N     N     N
4.8     Polyploidy
4.8.1  CHL                   N    P    N     N    P    N    N   N     N     P
4.8.2  CH1-L                 N    N    N     N    P    N    N   N     N     N
4.8.3  RL4                   N    N    N     N    P    N    N   N     N     N
4.9     Aneuploidy
4.9.1  CH1-L aneupl.         N    P    P     N    P    P    N   P     N     N
4.9.2  CH1-L spindle         N    N    N     N    P    P    P   P     N     N

5.1     L5178Y
5.1.1  L51-TK                P    P    N     P    P    P    N   N     N     N
5.1.2  L51-TK                P    N    P     N    N    P    P   P     P     N
5.1.3  L51-TK                P    N    N     N    N    N    N   N     N     N
5.1.4  L51-TK                P    N    P     P    N    N    N   N     N     N
5.1.5  L51-TK                P    P    P     ?    P    N    P   N     P     N
5.1.6  L51-TK                     ?    P     P    P    P    ?   N     N
5.1.7  L51-TK                     N    P     ?         N                    N
5.1.8  L51-OUA               N    N    P     P    P    P    ?   N     ?
5.2     V79
5.2.1  V79-OUA                    N    N     N                        N     N
5.2.2  L79-TG                     P    P     ?         P    P         P     ?
5.2.3  V79-TG                N    N    N     N    N    N    N   N     P     N
5.2.4  V79-TG                     N    N     N         N                    N
5.3     CHO
5.3.1  CHO-TG                     N    N     N         N                    N
5.3.2  CHO-OUA                    ?    N     N         N                    N
5.4     Human lymphoblasts
5.4.1  Human lym. TK         P    P    N     P    N    N    N   N     N     N
5.4.2  Human lym. TG         P    P    N     N    P    P    P   N     N     N
P = Positive
N = Negative

a   Results of the study expressed by assay group and test agent.  The
    order of assays is as described in the Introduction and as displayed
    in the associated key.  The order of the chemicals is according to
    decreasing genotoxicity, as shown in Fig. 1.  The results shown are
    those of individual investigators and may, therefore, differ from
    those shown in the assay working group reports.  Reproduced by
    permission, from: Ashby et al. (1985).
b   ACN - acrylonitrile.    PB - phenobarbital.
    TOL -  o -toluidine.     ZOIN - benzoin.
    SAF - safrole.          CAP - caprolactam.
    BEN - benzene. 
    From the summarized total data base, as set out in Table 2, it 
appears that the overall detection rates of the various assays were 
poor, as evidenced by the predominance of negative results.  
However, as one of the objectives was to challenge each assay 
system with particularly difficult chemicals, in order to assess 
its real status as a complementary test, this scatter of results 
was to be expected.  With only 8 "probes", it was not anticipated 
that definitive answers would be obtained, but rather that systems 
offering the best potential for development would be identified. 

    In the detailed examinations of the raw data carried out by the 
investigators at St. Simon's Island, many of the discrepancies in 
the findings were associated with differences in protocol.  Matters 
such as dose levels, sampling times, statistical methods, cell 
lines, and metabolic competence of the systems all proved to be 
critical aspects of protocol.  Had there been agreement on all 
these aspects, it is highly probable that the results would have 
been far more consistent.  Thus, as with the IPESTTC, the CSSTT had 
again demonstrated the importance of protocol detail in assay 
systems as the key to interlaboratory consistency.  Such protocol 
development is a definite and obviously difficult process in the 
evolution of a test system, and the CSSTT has clearly demonstrated 
that, with the exception of the Ames test, this has not been 
achieved satisfactorily for any of the possible complementary assay 
systems.  However, at least a possible mechanism for bringing about 
significant improvements is emerging. 

Table 3.  Overall performance of assays for 8 carcinogensa,b
Class of assay/            Summary of qualitative results expressed as positive tests/total tests         Overall
chemical            ACN   TOL   HMPA  SAF   Overall  DES   BEN   PB    DEHP  Overall  ZOIN  CAP  Overall  performance
                                                                                                          for the 8
2.  Fungi

   2.1 Mutation     4/4   3/5   4/5   4/5   15/19    0/3   2/4   1/4   1/4   4/15     2/5   2/5  4/10     50%
                                            (79%)                            (27%)               (40%)
   2.4/4.9.1        3/5   3/5   3/5   3/5   12/20    4/5   4/5   2/5   3/5   13/20    0/5   0/5  0/10     62%
   Aneuploidy                               (60%)                            (65%)               (0%)

3. Drosophila       3/3   2/3   3/3   2/3   10/12    0/3   1/3   0/3   0/3   1/12     0/3   3/3  3/6      46%
    somatic cells                            (83%)                            (8%)               (50%)

4.  Cultured mammalian cells

   4.3 Single-      2/3   3/3   0/3   2/3   7/12     3/3   1/3   0/2   0/3   4/11     1/3   0/3  1/6      48%
   strand breaks                            (58%)                            (36%)               (17%)
   4.4 UDS          1/2   3/3   2/3   3/3   9/11     0/2   1/3   0/2   1/2   2/9      1/2   0/3  1/5      55%
   (scintillation)                          (82%)                            (22%)               (20%)
   4.5 Chromosomal  4/4   3/5   3/5   2/4   12/18    3/4   3/6   3/4   0/3   9/17     1/4   2/6  3/10     60%
   aberrations                              (67%)                            (53%)               (30%)
   4.6 SCE          2/3   4/5   3/4   2/4   11/16    0/2   0/5   1/3   0/2   1/12     0/2   0/5  0/7      43%
                                            (69%)                            (8%)                (0%)

5.  Mammalian cell   6/7   5/10  7/10  5/10  23/37    5/8   6/10  4/9   1/8   16/35    3/9   0/8  3/17     55%
    mutations (L51,                          (62%)                            (46%)               (18%)
   V79 human cells)                                                                        

Table 3.  (contd.)
Class of assay/            Summary of qualitative results expressed as positive tests/total tests         Overall
chemical            ACN   TOL   HMPA  SAF   Overall  DES   BEN   PB    DEHP  Overall  ZOIN  CAP  Overall  performance
                                                                                                          for the 8
Overall             25/31 26/39 25/38 23/37          15/30 18/39 11/32 6/30           8/33  7/38
activity            (81%) (67%) (66%) (62%)          (50%) (46%) (34%) (20%)          (24%) (18%)
                            99/145                   33/69       17/62                15/72
                            (68%)                    (48%)       (27%)                (21%)
                    -----------------------          -----------------------          -----------
                                    132/214                                  32/133
                                    (62%)                                    (24%)
1.  Salmonella       2/5   1/5   0/5   0/5   3/20     0/5   0/5   3/5   0/5   3/20     1/5   0/5  1/10     15%
                                            (15%)                            (15%)               (10%)

4.2 Transformation  3/3   5/6   4/6   3/6   15/21    3/5   2/5   1/3   4/5   10/18    1/4   276  3/10     64%
                                            (71%)                            (55%)               (30%)
a The responses shown in the reduced data base (Table 4) are represented as number of positive responses/number of 
  observations made.  Questionable responses have been eliminated from the numerator but included in the denominator. 
  The carcinogen sensitivity of each class of assay has been calculated, but not their sensitivity (as only 2 
  carcinogens were employed), nor their accuracy (because of the unique handling of non-carcinogens in this study). 
  The responses of the  Salmonella assay and the transformation assays are shown at the foot of the Table.
b ACN - acrylonitrile   TOL -  o-toluidine    SAF - safrole
  BEN - benzene         PB - phenobarbital   ZOIN - benzoin
  CAP - caprolatam


    The most fundamental assumption made at the outset of the 
collaborative study was that the 8 carcinogens selected were either 
difficult or impossible to detect as positive using the standard 
 Salmonella mutation assay, and that the 2 non-carcinogens would be 
equally inactive.  This assumption was based, in part, on the 
results of the IPESTTC and partly on a general perception of the 
published literature, available earlier in 1981, on chemicals 
picked in November, 1981. 

    As the collaborative study progressed, it was decided to 
re-evaluate the test chemicals in the  Salmonella assay as an 
integral part of the study.  This was triggered by a variety of 
factors.  First, the detailed literature review undertaken by The 
Environmental Mutagen Information Centre (EMIC) revealed reports on 
the mutagenicity of some of the chemicals for  Salmonella.  Second, 
the use of the pre-incubation test was becoming increasingly 
common, and not all of the agents had been tested using this 
protocol.  Third, 2 new strains of  Salmonella were announced by 
Professor Bruce Ames at that time (TA97 and TA102) (Ames et al., 
1975), and the possible activity of these chemicals became of 
interest.  Finally, these 10 agents had not been tested in parallel 
before, nor had any common criteria been applied for the assessment 
of their relative mutagenicity or chemical purity. 

    The bacterial study included both the plate-incorporation and 
pre-incubation assay protocols, a range of S9 mixes and the 7 major 
strains of  Salmonella, including TA97 and TA102.  The  Salmonella  
forward mutation system of Skopek et al. (de Serres & Ashby, 1981) 
was included for purposes of comparison (strain TM 677; assay 
1.1.2), and Zeiger (assay 1.1.4) employed uninduced hamster as well 
as induced rat S9 in his experiments because this was his standard 
practice (Table 2). 

    Negative conclusions were recorded in 347 of 360 tests, the 
exceptions being listed below: 

 o-toluidine     weak activity in TA1535 and TA100, in 1 out of
                5 laboratories, and only when using hamster S9.

acrylonitrile   weak activity in TA1535 and TA100 (+S9), in one
                laboratory, weak activity in TM677, in another
                laboratory (-S9), and questionable activity in
                TA102, in a third laboratory (-S9).

benzoin         weak activity in TA1535 (-S9) and questionable
                activity in TA100 (-S9), in 1 laboratory out of

phenobarbital   S9 independent weak activity in TA1535, in 2
                laboratories, and in TA100 in 1 laboratory.
                Questionable activity was also seen in TA100
                (S9) in one of the laboratories recording
                activity in TA1535.

    HMPA, benzene, safrole, caprolactam, DEHP, and DES showed no 
evidence of mutagenic activity. 

    The assay working group concluded that these data confirm that 
the present 10 test chemicals are either difficult or impossible to 
detect as bacterial mutagens using the routinely-employed test 
protocols of the  Salmonella assay; thus, their selection for the 
present study was endorsed. 


    Application of the criteria set out in section 7 for the 
overall assessment of an assay performance as complementary to the 
Ames test eliminates about half the data in Table 2.  The remaining 
data sets are presented in Table 4.  As the major conclusions of 
the CSSTT are based on this "reduced" data base, some justification 
for the elimination principles employed is necessary.  The 
 Salmonella data, by definition of complementary assays, have not 
been included.  The decision to remove certain classes of assay 
from consideration, because they were not yet suitably developed 
for routine use, or had proved difficult to establish in 
independent laboratories, was a decision of the editors; however, 
their views were usually supported by the conclusions of the 
appropriate assay group reports.  Thus, the transformation assays 
were eliminated, largely on the basis of the conclusions of the 
working group.  However, it is relevant that, in laboratories where 
certain of these assays were performing reliably, they appeared to 
provide an efficient complementary assay.  The metabolic 
cooperation assays were similarly eliminated, as these did not 
appear to be optimal for use as a complementary test.  The removal 
of certain classes of assay, which were generally insensitive to 
the present carcinogens, was automatically justified by the aims of 
this study.  The rat hepatocyte autoradiographic UDS assays and the 
CHO micronucleus test were eliminated on the basis of this 
Table 4.  IPCS CSSTT  in vitro study: summary of reduced data basea,b
ASSAY                        ACN  TOL  HMPA  SAF  DES  BEN  PB  DEHP  ZOIN  CAP
2      FUNGI
2.1     Mutation
2.1.4  XV185                 P    P    P     P         P    P   P     P     P
2.1.5  XV185                      N    P     P                        P     P
2.1.7  D6                    P    N    P     N    N    N    N   N     N     N
2.1.8  D61-M                 P    P    P     P    N    N    N   N     N     N
2.1.9  Mito. D5              P    P    N     P    N    P    N   N     N     N
2.4     Aneuploidy
2.4.1  D6                    P    P    P     P    P    P    P   P     N     N
2.4.2  D61-M                 N    N    N     N    P    P    N   N     N     ?
2.4.3  D61-M                 P    P    P     P    P    P    P   P     N     N
2.4.4  Asper. 35             P    N    N     P    N    N    N   N     N     N

3.1.1  Wing spots            P    P    P     P    N    P    N   N     N     P
3.1.2  Eye spots             P    N    P     N    N    N    N   ?     N     P
3.1.3  Eye spots             P    P    P     P    ?    ?    N   ?     N     P

4.     CULTURED MAMMALIAN CELLS (endpoints other than gene mutation)
4.3     Single-strand breaks
4.3.1  Rat Hepat.            P    P    N     P    P    N    N   N     P     N
4.3.2  CHO                   N    P    N     P    P    P        N     N     N
4.3.3  CHO                   P    P    N     N    P    N    N   N     N     N

Table 4 (contd.)
ASSAY                        ACN  TOL  HMPA  SAF  DES  BEN  PB  DEHP  ZOIN  CAP
4.4     Unscheduled DNA synthesis (UDS)
4.4.3  Rat Hepat. (scint.)   P    P    P     P    N    P    N   P     P     N
4.4.4  HeLa (scint.)         N    P    P     P    N    N    N   N     N     N
4.4.5  HeLa (scint.)              P    ?     P         N                    N
4.5     Chromosomal aberrations
4.5.1  CHO                   P    P    N     N    P    N    P   N     N     N
4.5.2  CHO                        N    N     P         P                    N
4.5.3  CHO                   P    N                    N    P               N
4.5.4  LYM                             P          P    P              N     P
4.5.5  CH1-L                 P    P    P     N    P    N    P   N     N     N
4.5.6  CHL                   P    P    P     P    N    P    ?   N     P     P
4.5.7  RL4                   N    P    N     N    N    N    N   N     N     N
4.6     Sister chromatid exchange
4.6.1  CHO                   P    P    P     P    N    N    N   N     N     N
4.6.2  CHO                        P    N     N         N                    N
4.6.3  CHO                   P    N                    N    P               N
4.6.4  CHO                        N    N     N         N        N           N
4.6.5  V79                        P    P     P         N                    N
4.6.6  RL4                   N    P    P     N    N    N    N   N     N     N
4.9     Aneuploidy
4.9.1  CH1-L aneupt.         N    P    P     N    P    P    N   P     N     N

5.1     L5178Y
5.1.1  L51-TK                P    P    N     P    P    P    N   N     N     N
5.1.2  L51-TK                P    N    P     N    N    P    P   P     P     N
5.1.4  L51-TK                P    N    P     P    N    N    N   N     N     N
5.1.5  L51-TK                P    P    P     ?    P    N    P   N     P     N
5.1.6  L51-TK                     ?    P     P    P    P    ?   N     N
5.1.7  L51-TK                     N    P     ?         N                    N
5.1.8  L51-OUA               N    N    P     P    P    P    ?   N     ?
5.2     V79
5.2.2  L79-TG                     P    P     ?         P    P         P     ?
5.4     Human lymphoblasts
5.4.1  Human lym. TK         P    P    N     P    N    N    N   N     N     N
5.4.2  Human lym. TG         P    P    N     N    P    P    P   N     N     N
P = Positive
N = Negative

a   Results from selected assays shown in Table 2.  This reduction in the
    data base is justified in the text and forms the basis for the
    selection of a generally-applicable complementary in vitro  assay.
    Reproduced by permission, from: Ashby et al. (1985).
b   ACN - acrylonitrile.    PB - phenobarbital.
    TOL -  o-toluidine.      ZOIN - benzoin.
    SAF - safrole.          CAP - caprolactam.
    BEN - benzene.
    Elimination of individual assays that failed to detect any or 
only 1 of the test chemicals, in cases where all 10 had been 
evaluated, requires separate justification.  In most of these 

cases, other investigators using nominally the same assay detected 
several of the carcinogens; the inference is, therefore, that the 
particular protocol used was more at fault than the class of assay.  
The individual assays eliminated by this criterion are listed 
below, and it is clear that different underlying reasons may have 
led to their poor performance. 

    Six assays only detected acrylonitrile as positive.  The fact 
that this carcinogen was the most generally genotoxic of the 8 
carcinogens tested suggests that the sensitivity of the assay 
protocols in question was too low, rather than that the assay class 
as a whole was of no potential value as a complementary assay. 
However, 4 of these assays formed part of 2 classes eliminated for 
reasons of general insensitivity (yeast gene conversion and yeast 
crossing over).  These 6 data sets were: 

          yeast conversion

          yeast crossing over

    4.7.1 CHO micronucleus

    5.1.3 L5178Y TK gene mutation

    The ability of 2 of the polyploidy assays (4.8.2 and 4.8.3) to 
detect only DES may reflect the exceptional potency of this agent 
as a spindle poison.  DES may, therefore, be too potent an agent to 
act solely as the monitor for the sensitivity of aneuploidy assays. 

    Five assays gave isolated positive responses that appeared to 
have no explanation.  These effects may, therefore, reflect the 
technical false-positive responses of these tests.  Whatever the 
reason, the weakness of these responses, together with the 
generally negative context in which they occurred, suggests that 
these activities should only be related with caution to the 
carcinogenicity of the test agents.  These activities involved the 
following chemicals and assays: 

benzoin         assay 5.2.3 (V79 TG gene mutation)

 o-toluidine     assay 4.5.7 (RL4 chromosomal aberrations)

HMPA            assay 4.4.1 (hepatocyte autorad. UDS)
                assay 2.2.1 (yeast D7 gene mutation)

benzene         assay 2.1.3 (yeast D7 gene mutation)

    The above decisions to eliminate certain assays and data were 
designed to optimize the relevance of the conclusions from the 
present collaborative study.  The steps outlined removed from 
consideration assays that were not optimally performed or that were 
representative of a class of assay generally deemed unsuitable for 

routine adoption, as a complementary assay at this stage of their 
development.  Similar exclusion principles have been adopted by the 
several GENETOX review groups, and the resultant data base provides 
a more reliable reflection of the current stature of individual 
assays.  The positive effect that the above decisions had on the 
CSSTT data base is demonstrated in Fig. 1.  From this, it is 
evident that the composite sensitivity of the assays to the 8 
carcinogens has been increased without any corresponding loss of 
specificity (see boxed area of Fig. 1).  This figure also 
demonstrates that the degree of genotoxicity of the present 10 
chemicals, relative to each other, was not affected by the 
elimination procedures employed, and thus provides a compelling 
justification for proceeding with a detailed assessment of the data 
shown in Table 4. 


    The performance of assays on the reduced list was carefully 
assessed by the Editorial Committee, as it was felt that, for the 
guidance of the reader, some overall evaluation of this complex 
project was necessary, even though at this stage the opinions were 
those of scientists, closely associated with the project, and, 
thus, might be subjective.  Close attention was paid to the 
conclusions drawn by the assay groups regarding the usefulness of 
individual assays.  However, not all of these working groups 
approached an assessment of their assays within the context 
outlined originally.  On occasion, the recommendations and 
conclusions derived below are at variance with conclusions of the 
group but, generally, they are consistent with them or represent 
the only ones available. 

9.1.  Gene Mutation in Yeast

    These assays are, phylogenically, the nearest to the  Salmonella  
assay and share a similar genetic end-point.  Previous studies have 
also established that genotoxic agents, found active in  Salmonella,  
are generally also active in the corresponding yeast gene-mutation 
assays.  It was, therefore, expected that these 2 classes of assay 
would share similar sensitivities.  Fourteen positive responses 
were observed within the group comprising HMPA,  o-toluidine, 
safrole, and acrylonitrile, while only 3 positive responses were 
observed for benzene, DEHP, DES, and phenobarbital, and there was 
general agreement on the gene mutagenicity of HMPA (4/5 positive) 
(Table 3, Fig. 2).  The enhanced sensitivity of the yeast gene-
mutation assays, as opposed to the  Salmonella mutation assay, to 
the 4 assumed genotoxins may be due to endogenous metabolism, 
especially evident when the cells are in active growth.  It was, 
therefore, concluded that this class of assay might represent a 
possible alternative to the  Salmonella  assay, and might present 
advantages, on occasion, because of its enhanced sensitivity to 
certain genotoxins.  The mitochondrial mutation assay (2.1.9) has 
been included in this discussion, but should correctly be regarded 
as a distinct assay.  Two potential problems are presented by these 
assays, and by yeast assays in general.  First, they are not yet 
conducted according to an agreed protocol.  For example, the data 
shown in Table 2 represent the consensus gained after conducting 
assays within certain laboratories under a range of test 
conditions:  low or high glucose, stationary, semi-stationary, and 
log-phase cells, etc.  In addition, the largely positive data base 
of assay 2.1.4 includes 2 classes of positive response (see assay 
group report and investigator report).  Second, the  Salmonella  
assay must remain the preferred primary assay for detecting gene 
mutagens  in vitro, owing to the extensive data base available and 
the large number of laboratories involved (ca 2000). 

    It was concluded that appropriate yeast assays present a useful 
method of detecting gene mutagens, and, on occasion, can be more 
sensitive than the  Salmonella test.  However, given the current 
state of the art, they would not be recommended as primary gene-
mutation assays and their usefulness as a single complement to the 
 Salmonella test is reduced by the repetition of the genetic end-
point involved.  The preferential detection by these assays of 
HMPA,  o-toluidine, safrole, and acrylonitrile is referred to later. 

9.2.   Drosophila Somatic Cell Mutation Assays

    New somatic cell mutation assays have recently been developed 
in  Drosophila.  Their novelty is evidenced by the fact that data 
supporting their general sensitivity to genotoxic agents (i.e., 
agents readily detected by the  Salmonella assay) are only now 
being collected for publication.  General sensitivity is, however, 
claimed (see working group report).  Two different types of assay 
were represented.  Assays 3.1.1 and 3.1.3, although based in 
different tissues, were capable of detecting a range of induced 
genetic events, i.e., chromosomal breakage, deletions and 
aneuploidy, gene mutations, and mitotic recombinations.  These 2 
assays were, in fact, more sensitive to the present test chemicals 

than assay 3.1.2, where an inappropriate route of administration 
was employed and the range of detected mutagenic events was limited 
to gene mutations and deletions.  Nevertheless, the general 
concordance between the responses recorded in each of these 3 
assays suggests that they are likely to yield reproducible data 
between laboratories. 


    Assays using  Drosophila require minimal technical facilities 
but may prove relatively time-consuming.  They present the appeal 
of conducting an assay on a multicellular organism with its 
inherent metabolic capability.  It cannot be automatically assumed 
that this capability is significantly related to that of a mammal, 
but the sensitivity of these assays to a range of genotoxic agents 
is consistent with their possession of a broad range of metabolic 

    The 2 most sensitive assays (3.1.1 and 3.1.3) detected 5 and 
perhaps 6 of the test agents, one of which was the non-carcinogen 
caprolactam (see subsequent discussion of CAP and ZOIN).  Activity 
was mainly centred in the first group of carcinogens (Table 4, Fig. 
3).  This performance suggests that somatic mutation in  Drosophila,  
especially the assays capable of detecting a wide range of genetic 
events, presents a promising new assay.  However, their 
preferential detection of HMPA,  o-toluidine, safrole, and 
acrylonitrile reduces their value as a general complementary assay 
(cf. similar discussion of yeast gene-mutation assays, SCE assays, 
and final conclusions).  The present discussion is only related to 
the use of these assays for predicting the possible mammalian 
carcinogenicity of an agent; their use for predicting  in vivo 

mammalian mutagenic events should be considered separately (see 
also discussions of mammalian cell gene-mutation assays). 


9.3.  Assays for DNA Damage SSB (Single-Strand Breaks) and UDS 
(Detected via Autoradiography or Scintillation Counting)

    Although both of these classes of assay are designed to be 
sensitive to the consequences of the primary interaction of the 
test chemical (or a metabolite) with DNA, there are good reasons 
for not considering them as equivalent.  These reasons are 
discussed in the assay working group report and are supported by 
aspects of the present data base (Tables 2 and 3).  For example, 
HMPA was inactive in the 3 SSB assays yet was active in the assays 
for UDS (2+ and 1?).  In contrast, DES was uniformly active in the 
3 SSB assays and inactive in the UDS tests.  The fact that 
unanimity between these 2 assay groups can be achieved is evidenced 
by the responses seen for TOL (6/6+), SAF (5/6+), and PB (4/4-).  
The relative sensitivity of these 3 classes of assay are shown in 
Fig. 3.  It should be noted that some legislative authorities have 
discussed the need for assay data indicating primary interaction of 
a test chemical with DNA.  Their requirement could be met with data 
generated from either of these groups of assay, i.e., the 2 are 
generally regarded as being different methods of assaying a similar 
phenomenon.  As can be seen, this may not be so. 

    Perhaps one of the most surprising findings of the CSSTT is the 
insensitivity demonstrated by the 2 autoradiographic assays (4.4.1 
and 4.4.2, eliminated from Table 4).  These were conducted 
according to similar protocols and are well represented in earlier 
literature.  Their poor performance in the present study suggests 

that they are unsuitable for use as a complementary assay, though 
their use for confirming the activity in mammalian cells of 
previously defined bacterial mutagens remains.  Another intriguing 
aspect of their poor performance is that it provides evidence to 
suggest that the autoradiographic end-point is less sensitive than 
the other end-points under review (with the exception of the 
activity seen for HMPA in one assay, 4.4.1).  This insensitivity is 
unlikely to be due to metabolic incompetence as the hepatocytes 
employed were primary isolates, the metabolic competence of which 
had been previously established.  Furthermore, the possibility that 
this divergence of data was due to these agents producing only 
short-patch repair is reduced, given the range of chemically 
different carcinogens employed.  Whether some of the positive 
responses observed in the SSB assays were a direct result of cell 
necrosis is discussed below. 

    Assays for SSB and UDS would be expected to be primarily 
sensitive to the 4 confirmed genotoxins represented here, HMPA, 
 o-toluidine, safrole, and acrylonitrile, and this was observed (cf. 
Table 3 and the discussion of the yeast and  Drosophila gene-
mutation assays in this section).  The only confirmed success with 
the remaining 4 carcinogens was the ability of all of the SSB 
assays to detect DES.  Non-reproducible activity was seen in both 
classes of assay for benzene (2/6), benzoin (2/5), and DEHP (1/5).  
The latter isolated positive response was strongly endorsed by the 
investigator (4.4.3) and requires further investigation. 

    A topic for future research is the extent to which cell 
necrosis, with the consequent formation of broken DNA, contributes 
to the outcome of SSB assays.  For example, the isolated positive 
response observed for benzene (assay 4.3.2) was only observed at 
higher dose levels than those employed by the other 2 investigators 
in this group.  This response may, therefore, represent a technical 
false-positive response induced solely by DNA strand breakage as 
the result of toxicity.  Equally, the extent to which 
"mitotic" DNA synthesis has been successfully blocked during the 
conduct of a scintillation UDS assay relates directly to the 
credibility accorded to weak induced effects.  The above 2 concerns 
may explain some of the divergencies of data evident within these 3 
classes of assay for the present chemicals. 

    Given the simple requirement to provide evidence of the ability 
of a chemical to induce primary DNA damage in mammalian cells, the 
investigator is faced with the following decisions: 

    (a)  If the agent is already defined as a bacterial mutagen, 
         any of the 3 classes of assay (SSB or UDS via 
         autoradiography or scintillation counting) discussed 
         herein could be used. 

    (b)  If the agent is inactive as a bacterial mutagen, use of 
         either the SSB or scintillation counted UDS assays could 
         present advantages, depending on the chemicals involved.  
         The results obtained for HMPA and DES suggest that the use 
         of both assays should be considered, if an assay for "DNA 

         damage" is required as a complement to bacterial 

    It was, therefore, concluded that the simple legislative 
requirement for "evidence of DNA damage" was too vague to be 
meaningful.  If complementary data are being sought, the 
autoradiographic end-point should be avoided and data should be 
derived from an SSB assay and/or scintillation counted UDS assay.  
The failure of the CSSTT to indicate clearly a single class of 
assay, within this group, for routine adoption suggests that 
attention should be given to developing the sensitivity of DNA 
damage assays to the point that only a single test is required.  
The relative performance of these 3 classes of DNA repair assays is 
shown in Fig. 3. 

9.4.  Assays for the Induction of Aneuploidy

    In the past, assays for the induction of aneuploidy have not 
generally been considered as predictive tests for carcinogenicity.  
In the first place, few appropriate  in vitro assays existed, and 
secondly, the induction of aneuploidy was assumed to have greater 
relevance for the promotional, as opposed to the initiatory stages 
of carcinogenesis.  Thus, while the previous data base has included 
the evaluation of a few reference genotoxins such as MNU, it has 
mainly focused on cancer-promoting agents such as the phorbol 
esters and saccharin, etc.  Two investigators have provided the 
greater part of the present data base for this class of assay by 
presenting data derived from yeast assays (  In addition, 
data derived using mammalian cells were presented (4.9.1) together 
with a set of data observed in  Aspergillus (2.4.4). 

    The performance of the aneuploidy assays in the CSSTT was 
impressive (Table 2 and 3).  In the case of the 2 sets of yeast 
data provided from one laboratory (2.4.1 and 2.4.3), a 100% 
correlation between the reported carcinogenicity of the test agents 
and their activity  in vitro was observed.  In particular, the 
unanimous detection of both DEHP and phenobarbital by these 
aneuploidy assays is unique in this study. 

    The primary conclusion regarding this group of assays is, 
therefore, that they present high promise for development as a 
class of complementary assays.  The development phase should be 
influenced by the following considerations, some of which are of 
sufficient importance to ensure that they are resolved before any 
particular aneuploidy assay is recommended for general adoption. 

    (a)  The findings of the CSSTT suggest that the induction of 
         aneuploidy may be a critical step in the mechanism of 
         action of complete carcinogens.  These observations 
         require independent confirmation in other laboratories, 
         for both the yeast and the mammalian cell systems. 

    (b)  In the case of the D61-M yeast data from the 2 
         laboratories considered, major discrepancies were evident.  
         These were all of the same nature and represented the 
         failure of one laboratory to detect anything beyond 

         benzene and DES.  It is possible that this divergence was 
         due to technical factors affecting the mitotic or 
         metabolic activity of the yeast.  For example, evidence 
         from the fungal studies supported an earlier observation 
         that growing cells are more sensitive to genotoxins than 
         stationary cells.  This could be because of the reduced 
         metabolic activity of stationary cells.  Thus, this 
         divergence might be due to a relatively minor technical 
         consideration, but it must be defined, understood, 
         corrected, and defended in the literature before any 
         general adoption of yeast aneuploidy assays can be 
         recommended.  The question of whether the D6 strain can be 
         regarded as repetitive of the D61-M strain also requires 
         resolution; if it is to be retained, its efficacy must be 
         confirmed in other laboratories. 

    (c)  The isolated set of  Aspergillus data suggests that this 
         assay is relatively insensitive.  Clearly, if the 
         induction of aneuploidy is to be pursued as a useful 
         complementary end-point, and if  Aspergillus nidulans is to 
         be selected as the marker organism, a separate data base 
         will have to be generated to justify this choice. 

    (d)  Requirements similar to those in (c) must apply to the 
         isolated set of aneuploidy data generated in the Chinese 
         hamster liver fibroblast assay (4.9.1).  This assay is 
         relatively novel; thus, apart from an expansion of the 
         data base and the commissioning of independent studies, 
         the following point requires attention.  The cells used in 
         this study were of between the 9 - 15th passage.  The 
         corollary for using such a spread of cell passages is that 
         sufficient evidence must be presented on the chromosomal 
         number and distribution in each cell line before test data 
         can be interpreted with confidence.  Also, the multiple 
         scoring of hyper, hypo, and polyploid cells presents a 
         complex data base, the statistical interpretation of which 
         might not yet have been optimized.  Furthermore, the 
         possible karyotypic instability of higher passage cells 
         suggests that the current practice of scoring only 200 
         cells each in test and control cultures might not be 

    The aneuploidy findings of the CSSTT, therefore, provide an 
exciting area for future research.  However, the role of such 
assays as routinely employed complementary tests must await the 
acquisition of further data and the outcome of developmental 
studies.  The related polyploidy assays are considered in section 

9.5.  Mammalian Cell Gene-Mutation Assays

    Mammalian cell gene-mutation assays, together with chromosomal-
aberration assays, are high on the list of optional secondary 
assays referred to by legislative authorities.  The performance of 
these assays was, thus, of particular interest.  The mammalian 

gene-mutation assays constituted the largest group of assays in the 

    Sixteen sets of data were considered, 6 of which were based on 
nominally the same assay (L5178Y), employing the same selective 
agent (trifluorothymidine). 

    Qualitative inspection of the total data base for the mammalian 
gene-mutation assays (Table 2), and of that produced following the 
primary reduction step (Table 4), was not, as pointed out by the 
working group, initially encouraging.  The overall sensitivity of 
the collected gene-mutation and chromosomal-aberration assays were 
similar (Table 3; 62 and 67% and 46 and 53% overall sensitivity, 
respectively, for the 2 groups of 4 carcinogens).  The value of 
such numerical indices can be questioned, but they are supported by 
the individual comparisons drawn in Table 3 for the activity of 
each chemical in these 2 classes of assay.  Thus, similar overall 
sensitivity was observed in both classes of assay for most of the 
chemicals.  Fig. 4 and 5 show the comparative sensitivity 
histograms with the  Salmonella assays. 



    The working-group discussions on the gene-mutation assays 
identified a number of specific factors that could have led to the 
many disagreements in assay responses and which, if followed up, 
should lead to the development of a greatly improved assay system.  
Each of these points was concerned with protocol details, all of 
which were of relevance to all of these assays, irrespective of the 
cell line, selective agent, etc., employed.  Some of these 
recommendations are of immediate relevance for the interpretation 
of weak assay responses (an important matter that is dealt with  in 
 extenso in the full account of the project).  The common feature of 
the present test chemicals is that they tend to yield weak 
responses  in vitro.   Thus, it is impossible to discern at this 
stage which of the many weak positive responses recorded in this 
assay group were real and which were illusions created by 
inadequate experimentation.  Equally, some of the negative 
responses observed may have been attributable to experimental 
deficiencies rather than to an absolute inadequacy of the assay to 
detect the carcinogen.  Suggested protocol modifications, 
especially those associated with the repetition of experiments and 
the use of appropriate cell numbers, should reduce the incidence of 
"technical" false-positive and false-negative responses.  The data 
base that would have been produced in this study, had such 
technical modifications already been incorporated, would doubtless 
have been much more consistent.  However, in addition to possible 
protocol deficiencies, which it is anticipated can be remedied, 
there remain 2 other important questions. 

    The first is the extent to which this class of assay can 
realistically be reduced to a common cell line using one or more 
common selective agents.  This would obviously represent a major 
advance, as it would enable meaningful interlaboratory comparisons 

of data to be made and engender a feeling of confidence that once a 
test chemical had been evaluated for mammalian cell gene 
mutagenicity, and a conclusion drawn, this would not be reversed on 
retesting in another, but similar assay.  The need to try to devise 
such a system, suitable for general screening was, in fact, the 
final recommendation of the assay working group.  Some of the 
divergencies in test results, evident in this study, were probably 
due to the differential sensitivity and metabolic competence of the 
several cell lines employed, and this problem cannot be resolved by 
adoption of a single common cell line.  However, the range of 
genetic loci observed and selective agents used probably led to 
many instances of true differential sensitivity.  The need to use 
at least 2 different genetic markers, when a wide range of 
genotoxic agents are to be detected has been discussed elsewhere, 
and instances of 2 loci being monitored concomitantly have been 
reported (Barrett et al., 1981).  A considerable amount of 
developmental work will be required to discover the degree to which 
this class of assay can be standardized with a retention of overall 

    The second concern is related to whether these assays are 
suitable, by definition, to act as complements to the  Salmonella 
assays.  The results from the yeast and  Drosophila assays were not 
encouraging and indicated that these particular assays were not 
optimal for this purpose, because of the similar nature of the 
genetic event being monitored.  Similar concerns apply to mammalian 
cell gene-mutation assays.  Thus, the concept that some agents may 
be specifically clastogenic is an argument against the 
incorporation of only gene-mutation assays in a battery.  Given 
that there is usually a requirement in legislative guidelines to 
assess a compound for clastogenic activity, the precise role of 
mammalian cell gene-mutation assays needs to be reviewed.  These 
considerations do not necessarily detract from their use as 
confirmatory assays.  Thus, if it is considered necessary to 
evaluate whether a bacterial gene mutagen is similarly active in 
mammalian cells, their use is obvious. 

    The future use of mammalian gene-mutation assays is, therefore, 
likely to be influenced by a variety of considerations.  The first 
and most fundamental is the need for a precise exposition of their 
proposed role in the detection of potential carcinogens  in vitro.  
If they are to be employed only to confirm activity observed in the 
 Salmonella assay, then some justification should be provided for 
the implied existence of a group of established non-carcinogens 
that are mutagenic to  Salmonella, yet inactive in a well-conducted 
mammalian cell gene-mutation assay.  If they are to be used as a 
complementary assay to the  Salmonella assay, urgent attention 
should be given to protocol development along the lines outlined in 
the assay working group report.  The selection of a single assay, 
or the development of a single multi-locus assay, may form part of 
this development process.  However, the wisdom of employing 2 gene-
mutation assays for screening purposes now needs careful 

    Set against these rather demanding requirements are the 
following 2 more positive observations.  First, it is significant 
that mammalian cell gene-mutation assays formed the largest group 
of closely-related assays considered in this study.  This is 
indicative of the current widespread use of these tests and implies 
that, in many laboratories, they are performing to an acceptable 
standard for reference mutagenic carcinogens such as 2AAF, MNU, 
etc.  The large number of laboratories carrying out these tests is, 
of course, mainly the result of legislative stimulus, but had the 
tests proved generally insensitive or unreliable, it would not have 
been possible to collect so many together for this study.  This is 
in contrast to most of the other assays in the study and provides a 
strong stimulus for further work to refine them to the point that 
they can be used with confidence for the detection of weak 
responses.  Second, although this study was concerned primarily 
with the detection  in vitro of potential mammalian carcinogens, the 
possible mutagenic properties of chemicals for  in vivo somatic and 
germ cells are becoming of increasing interest.  If the specific 
requirement to assay for possible mammalian  in vivo mutagens is a 
consideration, the role of  in vitro gene-mutation assays assumes a 
different complexion. 

9.6.  Chromosomal-Aberration Assays

    As indicated above, the performance of this class of assay was 
similar to, but slightly better than, that of the mammalian cell 
gene-mutation assays.  As with the gene-mutation assays, 
chromosomal-aberration tests are generally recommended in almost 
all legislative guidelines as suitable for use in conjunction with 
a bacterial-mutation test, when screening chemicals for new 
potential carcinogens.  However, while it is currently normal 
practice to select the L5178Y cell line when conducting a gene-
mutation experiment, a range of possible cell lines could be chosen 
for cytogenetic studies.  Thus, in the CSSTT, 4 distinct cell lines 
(CHO cells, human lymphocytes, Chinese hamster liver fibroblasts 
(CHI-L), and Chinese hamster lung cells (CHL)) were selected, each 
of which had been used previously for such studies, and each of 
which could be considered suitable for use, when submitting 
genotoxicity test data on a new chemical. 

    A distinct attribute of the cytogenetics assay data was that, 
although a range of cytogenetic disturbances can be classified as 
chemically-induced damage, the data base was more ordered than 
might have been anticipated, because of the common reporting format 
employed by most investigators.  This was achieved, in advance, by 
the assay group Chairman and was based substantially on the 
guidelines for assay data reporting suggested in the UKEMS criteria 
document (UKEMS, 1983).  Further sources of protocol variation can, 
therefore, be associated directly with technical decisions made by 
individual investigators, including such issues as doselevels, 
sampling times, methods of statistical analysis, cell line adopted, 

    As would be expected from the similar overall detection rates 
of the gene-mutation and chromosomal-aberration assays (Table 3, 
and Fig. 4 and 5), simple inspection of the chromosomal-aberration 

data base (Table 2 and 4) indicates a disappointing scatter of 
positive and negative observations.  Certain consistencies were, 
however, evident.  For example, general agreement was observed in 
the 3 cell lines studied for the clastogenicity of both 
acrylonitrile (4/4+) and phenobarbital (3+ and 1? of 4).  With the 
exception of a questionable response in one of the 3 CHO assays 
(4.5.3), DEHP gave negative results.  Benzoin gave 3 negative 
responses and a single, but confirmed, positive response in the CHL 
assay.  A partially consistent data set was observed for HMPA, 
which gave a negative response in each of the 2 CHO assays but was 
active in the 3 other cytogenetic assays.  This may reflect the 
general technical inadequacy of these assay protocols or represent 
a general failure of this cell line to metabolize HMPA, a genotoxic 
species.  One of the 2 CHO SCE assays gave a positive response for 
this agent, and this weakens, but does not destroy, the metabolic 
deficiency rationale. 

    Marked divergencies in assay responses were evident for the 
remaining 5 chemicals ( o-toluidine, safrole, benzene, caprolactam, 
and DES).  An approximately equal incidence of positive and 
negative responses was observed for these agents and each was 
regarded as clastogenic in the assay working-group "consensus" 
results ( o-toluidine 3/5 + safrole 2/4 +, benzene 3/6 + caprolactam 
2/6 +, and DES 3/4 +).  Important protocol deficiencies were 
identified in the assay working-group discussions.  These 
deficiencies were generally associated with the selection of 
inappropriate sampling times or dose levels, and if further work 
could prove them to be the critical contributors to false-negative 
assay responses, then the performance of these assays would be 
dramatically improved. 

    For the purpose of defining a complementary assay within the 
present context, the 4 classes of chromosomal-aberration assays 
cannot be considered as equivalent.  For example, the CHL assay has 
been extensively studied and reported on in the literature, but its 
use is centred mainly in a single laboratory.  Assays involving 
human lymphocytes, especially if a rat liver S9 mix is incorporated 
into the protocol, are few (Buckton & Evans, 1982).  The study data 
using Chinese hamster liver fibroblasts formed the greater part of 
the accepted data base for this assay.  In contrast, the CHO assay 
probably represented the one most widely used in industrial and 
academic laboratories.  This assay can, therefore, be regarded as 
particularly relevant for the objectives of the project, as 
probably more data are generated using it than any other, when 
compounds are assessed prior to legal registration.  Given that 
some of the negative responses observed in CHO cells might be 
corrected by the suggested modifications to the assay protocol, 
there is still the problem that, at present, different responses 
may be observed in different laboratories using nominally the same 
assay.  This is similar to the situation currently evident with the 
L5178Y gene-mutation assays and, as in that case, urgent attention 
to further protocol design and data interpretation is indicated. 

    The CHL assay (4.5.6), or perhaps the protocol by which it was 
conducted, was more sensitive than the others in this group.  By 
the same criteria, the partial data base evident for the human 

lymphocyte assay was also encouraging and worthy of elaboration.  
As in other assay systems, it was noted how "assays" were judged, 
or misjudged, by the test protocols adopted.  This important aspect 
of test systems must be resolved, because of the real possibility 
that, if more detailed or refined protocols had been used, many 
assays would have appeared in a much more favourable light. 

    An advantage of the chromosomal assays is the extent of the 
existing data base and the technical skill that is available for 
the conduct of these assays.  Thus, it can be expected that 
acceptable protocol modifications could be rapidly developed and 
disseminated for general adoption.  A second advantage is that a 
range of supplementary and probably "complementary" end-points can 
be easily assessed, once the decision has been taken to conduct 
cytogenetic studies within a laboratory.  These include the 
recording of polyploid cells, an assessment of aneuploidy, and the 
studying of sister chromatid exchanges (SCE).  Aneuploidy induction 
in mammalian cells is considered as a separate assay, and the 
induction of SCEs is discussed below.  However, the induction of 
polyploidy is pertinent to the present discussion as it can be 
assessed either concomitantly with, or in parallel to, chromosomal-
aberration assays, depending on the sampling time of the assay. 

9.7.  Assays for Polyploidy Induction

    The chemical induction of polyploidy was discounted as a 
separate assay by the working group, but the data generated were 
considered because of the interest of the results that were 
available.  Three complete sets of polyploidy data were presented 
(Table 2; assays 4.8.1 (CHL), 4.8.2 (CH liver), and 4.8.3 (RL4, rat 
liver)), and each detected DES as positive.  This activity is 
probably associated with the ability of DES to induce aneuploidy, 
spindle effects, and chromosomal aberrations.  In the last 2 assays 
(4.8.2 and 4.8.3), the remaining 9 compounds were found to be 
negative while, in the CHL assay (4.8.1),  o-toluidine and 
caprolactam were detected as active.  These responses indicate that 
this genetic end-point is unlikely to produce false-positive 
responses; in fact, activity seems to be rare. 

    The mode of action of DES, benzene, and DEHP as carcinogens is 
less clearly defined than that of most other agents.  Not only are 
they inactive in the  Salmonella assay, but they were difficult to 
detect as positive in many of the eukaryotic assays used in the 
CSSTT.  Furthermore, on the basis of their chemical structure and 
known or anticipated metabolism, these agents do not appear to be 
likely to interact covalently with DNA.  It was, therefore, 
perceived by many investigators to be of particular interest that 
each of these agents was capable of inducing both aneuploidy and 
cell transformation  in vitro.  Furthermore, if "associated" 
chromosomal-assay data are considered in conjunction with the 
aneuploidy findings, a potentially useful pattern emerges, as shown 
in Table 5. 

Table 5.  Comparative performance of DES, benzene, and DEHP in the classes of 
assay showna
         Cell            Aneuploidy  Clastogenicity  Polyploidy  Spindle studies
         transformation                                          (assay 4.9.2)
DES      +               +           +               +           spindle damage

benzene  +               +           +               -           modified spindle

DEHP     +               +           -               -           modified spindle
a Results are based on the consensus conclusions given by the working groups.
    From Table 5, it can be seen that DES was clearly active in 
each of the 3 categories of assay, and that its mode of action in 
each might be closely related to its ability to damage the 
microtubules of the metaphase spindle.  In contrast, DEHP was 
inactive in the 3 polyploidy assays, as well as in the chromosomal 
assays.  The question therefore arises of whether the cell-
transforming and aneuploidy-inducing properties of DEHP are 
mechanistically unrelated to those of DES.  If totally different 
mechanisms of action are involved, then the relationship of these 
findings to the carcinogenicity of these agents may equally be 
different.  In this connection, the difference between modification 
of spindle integrity and spindle function may be important.  Thus, 
it is concluded that if an assay for chromosomal aberrations is to 
be used as a complementary test, the determination of changes in 
ploidy may be a useful additional parameter to measure.  In some 
assays, it may not be possible to score both chromosomal 
aberrations and ploidy, at the same time, mainly because of the 
need to progress to the second mitosis after treatment, for 
polyploidy to become evident.  However, the maintenance of a 
parallel culture after treatment for separate assessment of ploidy 
should present few difficulties. 

9.8.  Sister Chromatid Exchange (SCE) Assays

    The relationship between the chemical induction of SCEs and the 
production of chromosomal aberrations is not clear, and the 
mechanism of this phenomenon is obscure.  The role of SCE assays in 
the detection of carcinogens has yet to be defined.  For these 
reasons, and because their respective test protocols are quite 
distinct, SCE assays have been considered separately from the 
aberration assays in the present discussion. 

    One set of data was omitted from Table 4 owing to the recording 
of 6 out of 6 negative responses (4.6.4) and, in the remaining 5 
SCE assays, only 12 positive responses were observed out of 41 
determinations.  These positive responses were not randomly 
distributed among the compounds as can be seen in Table 3.  Two-
thirds of the SCE assays made on HMPA,  o-toluidine, safrole, and 
acrylonitrile gave positive responses, while only a single positive 

response was observed among the 14 determinations made on benzene, 
DEHP, DES, or phenobarbital.  This grouping of positive responses 
is very similar to that seen with the  Drosophila mutation assays 
and to a lesser extent by the yeast gene-mutation and UDS 
(scintillation) assays.  This supports earlier suggestions that the 
induction of SCEs is more closely related to gene mutagenicity than 
to clastogenicity.  No activity was observed for the 2 non-

    Given that the greatest number of positive responses was 
observed in this study for the first 4 "genotoxic" carcinogens 
shown in Table 4, the SCE findings are unexceptional.  The 
inability to detect the remaining 4 carcinogens by SCE assays may 
provide clues (along with the  Drosophila assays) regarding the 
mechanism of carcinogenic action of these agents.  An observation 
of particular interest is that 3 carcinogens, benzene, DES, and 
phenobarbital (but not DEHP) were each active in one or more of the 
other chromosome-based assays; their inactivity in the SCE assays, 
therefore, becomes pointed. 

    The SCE assays, therefore, appear to be of limited advantage as 
complementary assays.  Together with several other assays, they 
successfully detected the "agreed" genotoxins in the CSSTT and, as 
such, the assay seems to be more closely related to the  Drosophila  
and yeast gene-mutation assays.  The low detection rates evident 
for the 4 remaining carcinogens suggest that SCE assays are less 
generally useful as a complementary test than the chromosomal-
aberration and mammalian cell gene-mutation assays. 

9.9.  Transformation Assays

    The working group, for reasons briefly referred to earlier, did 
not consider this class of assay to be suitable for a generally-
applicable complementary assay.  However, the good overall 
performance of the assay in the CSSTT (Table 3) warrants further 
consideration.  The chemical induction,  in vitro, of the 
transformed cell phenotype is appealing, because the derived cell 
morphology and behaviour  in vivo and  in vitro bear a striking 
resemblance to the malignant phenotype.  The many different 
definitions of the transformed cell phenotype (Weinstein et al., 
1976), together with the technical difficulties that accompany the 
conduct of such assays have proved to be major factors contributing 
to the slow progress that has been made in the general use of these 
tests for the detection of possible new carcinogens.  The initial 
promise shown by the BHK-21 cell-transformation assay was due to 
the compatibility of these cells with S9 mix (Styles, 1977), but 
progress was impeded by the karyotypic instability of this cell 
line and inter-laboratory reproducibility problems encountered at 
an early stage (de Serres & Ashby, 1981).  The present study 
represents the first occasion on which the mouse C3H 10T 1/2 assay 
and the Syrian hamster embryo (SHE) assay have been directly 
compared in a general study, yet each has an impressive history in 
a few laboratories.  Although both of these classes of assay 
performed well in the CSSTT, there was a marked reluctance among 
the assay investigators when it came to suggesting any of these 

tests as suitable for routine adoption as a complementary assay.  
The insensitivity to the present carcinogens of the Balb C assay 
(4.2.1) and the CHO assay (4.2.8) confirmed previous observations 
that established cell lines tend to become metabolically 
incompetent, yet the metabolically competent primary cell systems 
suffer from other problems associated with the criteria used to 
select clones of cells for study, and the selection of appropriate 

    Recent studies associated with the probable role of activation 
of the oncogene in the etiology of cancer have invoked the 
transformed cell phenotype as a critical marker; thus, the future 
of this class of assay in cancer research seems assured.  
Nonetheless, the practical problems that would probably be 
encountered by a new laboratory endeavouring to establish such an 
assay are probably still sufficient to warn against its use as a 
routinely-employed complementary test. 


    No single assay was defined by the CSSTT as giving high 
carcinogen sensitivity and specificity together with an established 
ability to perform optimally in several laboratories.  It was 
concluded, however, that there was compelling evidence both from 
within and beyond the collaborative study, for the adoption of a 
chromosomal-aberration assay as the reference complementary test.  
This conclusion was drawn on the basis of the following data and 

    (a) Eight classes of tests, as listed in Table 3, were assessed 
        for performance as complementary assays, in the foregoing 
        discussion.  Of these, the gene-mutation assays in yeast 
        and  Drosophila were not favoured because of their 
        repetition of genetic end-point and their poor performance 
        with the last 4 carcinogens (benzene, DES, phenobarbital, 
        and DEHP).  Assays for aneuploidy were not recommended 
        because of their novelty, which necessarily prevented 
        independent studies from being commissioned.  The DNA-
        repair assays showed poor sensitivity for the last 4 
        carcinogens, and, occasionally, dramatic differences in 
        sensitivity between the single-strand breakage and 
        scintillation UDS tests were evident (Fig. 3).  The data 
        base for the SCE assays suggested marked insensitivity to 
        the last 4 carcinogens (Fig. 6).  Several of the above 
        assays were, therefore, defined as capable of acting as a 
        partial complement to the  Salmonella assay, but none could 
        be recommended for routine use as a generally sensitive 
        complementary test. 


    (b) The transformation working group recommended that assays 
        for cell transformation were not suitable for routine 
        adoption because of the technical difficulties of carrying 
        them out and the remaining uncertainty regarding the 
        definition of the transformed phenotype.  Nonetheless, 6 of 
        these assays showed high sensitivity to the present 8 
        carcinogens; in fact, as a class, they performed better 
        than any other category of test (Table 3).  It is clear 
        that both the C3H 10T 1/2 and the SHE assay are sensitive 
        assays, and each performed similarly, in the CSSTT, in 2 
        different laboratories.  Nonetheless, the advice of the 
        assay group is endorsed here; namely, that transformation 
        assays are not to be recommended for routine adoption, at 
        present, especially by laboratories new to such techniques. 

    (c) Two classes of assay remain for consideration, each of 
        which was well represented in the CSSTT, i.e., mammalian 
        cell gene-mutation and chromosomal-aberration assays.  Both 
        of these classes of assay showed a similar level of 
        carcinogen sensitivity, when considering the reduced data 
        base (Table 3 and 4). 
        The apparently similar performance of these 2 classes of 
        assay as complementary tests is shown in Fig. 4 and 5.  The 
        performance of the  Drosophila and SCE assays is shown in 
        Fig. 6 for purposes of comparison and to illustrate that 
        some assays only partially complement the  Salmonella assay.  
        When considering the total data base (Table 2) for the 
        gene-mutation and chromosomal-aberration assays, the latter 
        appear to be more sensitive, yielding 22 positive responses 
        out of 43 determinations made on the 8 carcinogens (51%), 
        as opposed to only 40 out of 103 for the mutation assays 
        (39%).  The low sensitivity of some of the mutation assays 
        led to 6 of the 16 being eliminated in Table 3 as opposed 
        to only 1 of the 7 cytogenetic assays.  The normalized (%) 
        performances of these 2 classes of assays as complementary 
        tests are compared in Fig. 7.  Similar efficiencies are 
        evident with the chromosomal assays proving marginally more 
        sensitive to the carcinogens. 

    Figs. 4, 5, and 7 enable the problem of reproducibility of 
response to be visualized and it is clear that harmonization and/or 
extension of assay protocols is indicated for both classes of test.  
The effort required to produce a common and sensitive assay 
protocol, combined with a realistic assessment of whether such 
efforts are likely to be rewarded, constitutes the final factor in 
the recommendation of the chromosomal-aberration assays.  Four 
points appear to be particularly relevant: 

    (a)  chromosomal-aberration assays offer an independent genetic 
         end-point from the  Salmonella assay; 

    (b)  chromosomal-aberration assays have a more extensive 
         history than the mammalian cell gene-mutation assays; this 
         means that more laboratories have used them for a longer 
         period; thus, any protocol corrections required should be 
         more clearly perceived and easier to institute; 

    (c)  the recommendations for assay protocol modification made 
         by the chromosomal-aberration working group were more 
         clearly defined, less fundamental, and seemingly easier to 
         institute than those made by the gene-mutation working 
         group; and 

    (d)  once the decision is made to conduct a chromosomal-
         aberration assay in a laboratory, the expertise and 
         facilities required will serve equally for the assessment 
         of aneuploidy, polyploidy, and SCEs. 


    It is therefore recommended that major efforts be focused on 
designing a minimum effective protocol according to which 
chromosomal-aberration assays should be conducted, and that this 
could constitute a complementary assay for the  Salmonella test.  
Recourse to other assays may be indicated, in order to assess fully 
the  in vitro activity of certain test chemicals.  Acquisition of 
test data from a minimum of these 2 standard assays, if optimally 
performed, will enable a greater degree of comparability of data 
between both independent laboratories and different countries, and 
this can only help the efficient detection of possible new human 

    In many testing schemes, resources are spread across a large 
number of  in vitro assays, and this may lead to assays on a new 
chemical being conducted suboptimally.  The alternative suggested 

here is that with 2 well-conducted and common assays, there could 
be enhanced comparability and reliability of data, together with an 
increase in carcinogen sensitivity.  In addition, the  Salmonella  
assay should be conducted according to what is now regarded as the 
best current practice, i.e., employing strains TA1535, 1537, 1538, 
98, and 100 in the presence and absence of an induced rat liver S9 
mix (10%) (Ames et al., 1975; UKEMS 1983) and also using a 
preincubation protocol with strains TA98 and 100 in the case of 
initial negative responses.  Use of a different source of S9 mix 
(e.g., hamster) and additional levels of rat liver S9 mix (e.g., 4% 
and 30%) should also be considered (Venitt et al., 1984).  In the 
case of volatile chemicals, an enclosed test system should be 

    The particular cell line employed for cytogenetic assays may be 
critical but should not necessarily be restricted to the widely-
used Chinese hamster ovary (CHO) line.  Good results were obtained 
with the Chinese hamster lung (CHL) line in the CSSTT, but it is 
not commercially available at present, and while human lymphocytes 
present a sensitive and readily-available source of primary euploid 
cells, further studies will be required to demonstrate the general 
usefulness of this system (Buckton & Evans, 1982).  The Chinese 
hamster liver system protocol was designed to measure both 
aberrations and aneuploidy and, likewise, will require further 
independent studies to confirm the promise shown in the CSSTT. 


    The dilemma that preceded the formalization of final 
conclusions was difficult to resolve.  Several carcinogens of 
uncertain genotoxicity were selected for the CSSTT, in order to 
ascertain which among the many available eukaryotic  in vitro  
assays for carcinogens would be most efficient and best suited for 
adoption as a complement to the generally available  Salmonella  
mutation assay, by laboratories engaged in the routine screening of 
chemicals for potential carcinogens.  The organizing committee was 
aware of the impact that any conclusions derived from this study 
might have on legislative guidelines, but was primarily interested 
in the optimal use of the resources of laboratories that may be 
attempting to identify possible new human carcinogens.  In order to 
provide a comprehensive study, the number of types of assays 
included was not limited.  This non-restrictive method of 
collecting assays presented a unique problem; by implication, some 
investigators had been invited to submit their assay for scrutiny 
for a purpose other than they had originally intended.  For 
example, many oncologists conduct cell-transformation, aneuploidy, 
UDS, etc., assays for research purposes, in order to study the 
relationship between genetic end-points and the subtle nuances of 
the cancer process.  Such research studies may have supported the 
selective use of a particular assay that was not claimed to be of 
general usefulness for the present purposes, and which was not 
being recommended as such by the investigators.  It is, therefore, 
important to emphasize that the failure of an assay to qualify as a 
complementary assay does not preclude its use for research into the 
intricate processes of cancer initiation/promotion.  This caution 
is most important to remember when whole classes of assays such as 
"cell transformation", "metabolic cooperation", etc., have been 
dismissed from consideration in this collaborative study.  On the 
one hand, is the immediate and worldwide need for efficient and 
reliable means for screening new chemicals, leading to their 
registration as safe new products (Ashby et al., 1983); on the 
other, the need to use selected  in vitro assays to probe the 
etiology of chemically-induced cancer.  The present study is aimed 
solely at the former endeavour, but its findings might expedite the 
achievement of the latter.  Ancillary findings of the CSSTT are 
first, that certain classes of  in vitro assays appear to be 
insensitive to the cancer-related activities of certain chemical 
carcinogens.  Second, it is suggested that a clear distinction 
should be made between the use of eukaryotic  in vitro assays to 
confirm genotoxic activity, seen previously for a chemical in 
bacteria, and their use to complement bacterial-mutation assays, 
i.e., to detect genotoxins found inactive in bacteria; the choice 
of assay and the interpretation of data are influenced by the need 
originally perceived. 

1.  Significant differences exist among individual investigators 
    conducting nominally identical assays.  A direct result of the 
    CSSTT is that many recommended protocol changes were identified 
    for all assays and some have already been instituted by a 
    number of investigators.  Inadequate test protocols clearly 
    contributed to the fact that some assays appeared to be 

    insensitive.  It is likely that the need, currently perceived, 
    to conduct a range of genotoxicity assays on a new chemical has 
    been stimulated, in part, by the inadequacy of some test 

2.  Urgent attention should be given to the definition of criteria 
    for judging positive and negative test responses in all assays.  
    Methods by which weak responses can be interpreted with 
    confidence were identified, including the adoption of 
    appropriately sensitive test protocols, the acquisition of a 
    sufficiently large control data base, and the use of 
    appropriate statistical methods for data assessment. 

3.  Carcinogens that are inactive or difficult to detect in the 
     Salmonella assay fall into 2 distinct groups in this study.  
    The first include genotoxins that are probably only non-
    mutagenic to  Salmonella because of protocol deficiencies 
    associated with the overall metabolic capacity of the assay 
    system.  These agents, HMPA,  o-toluidine, safrole, and 
    acrylonitrile, were detected in most of the eukaryotic assays 
    studied.  Thus, a range of assays can act as limited 
    complements to the  Salmonella mutation assay.  The other group 
    of carcinogens, benzene, DEHP, DES, and phenobarbital, 
    possessed a more selective range of genotoxic activities, and 
    none of the assays was selectively sensitive to them.  Of the 
    assays studied, only the induction of chromosomal aberrations, 
    transformation, gene mutations in mammalian cells, and 
    aneuploidy in yeast gave encouraging overall performances for 
    the 8 carcinogens (Conclusion 9). 

4.  A surprising finding of the CSSTT was the apparent mutagenicity 
    observed for the 3 carcinogens, benzene, DES, and phenobarbital 
    in some of the mammalian cell gene-mutation assays.  These 
    activities were not seen by the majority of investigators in 
    this assay group and were not generally confirmed in either the 
     Drosophila or fungal gene-mutation assays.  If they prove 
    repeatable, 3 cell-specific gene mutagens will have been 
    defined.  If they are not repeatable, this will emphasize the 
    need, already perceived by the assay working group, to improve 
    the protocol and data assessment processes for this class of 
    assay.  The concomitant monitoring of 2 gene loci in a single-
    cell line, and the development of appropriate and common data 
    assessment procedures represent 2 areas for research with 
    mammalian cell gene-mutation assays.  The precise role of these 
    assays in routine screening programmes requires clarification; 
    in particular, whether they are to be used to anticipate 
    possible mammalian carcinogens or possible mammalian mutagens, 
    or both.  The use of negative results in a mammalian cell gene-
    mutation assay to negate positive observations made in the 
     Salmonella assay may be premature. 

5.   A range of cell-transformation assays was studied, and the 
    overall performance of several of them was impressive.  The 
    duplicated studies with the Syrian hamster embryo (SHE) assay 
    and the mouse C3H10T 1/2 assay gave particularly encouraging 

    results.  Nonetheless, these assays are not recommended for 
    routine screening purposes, at present, because of the 
    expertise required to carry them out and because of the 
    problems that remain with clone selection and judgement of when 
    chemically-induced transformation has occurred. 

6.  Both of the non-carcinogens showed low incidences of activity 
     in vitro, several of which were confirmed and most of which 
    were weak.  The clastogenicity of caprolactam for human 
    lymphocytes and its somatic cell mutagenicity in  Drosophila  
    were particularly clear-cut.  The activities observed for these 
    agents in certain  in vitro assays do not, therefore, correlate 
    with their rodent carcinogenicity, and thus, they represent 
    false-positive predictions of carcinogenicity.  This underlines 
    the fact that  in vitro assays are of use for predicting, rather 
    than defining carcinogenic activity. 

7.  Three of the carcinogens, DES, phenobarbital, and DEHP, were 
    chosen to evaluate the hypothesis that some chemicals can 
    induce tumours in rodents without first modifying the integrity 
    of nuclear DNA.  Although these chemicals were generally less 
    active in the tests than the other carcinogens, they each 
    displayed a range of genotoxic activities  in vitro.  DEHP 
    showed least activity, with levels similar to those observed 
    for the 2 non-carcinogens.  Nevertheless, it induced 
    significant aneuploidy and cell-transformation effects in an 
    almost total absence of evidence of a direct interaction with 
    DNA.  The data generated for DES confirmed earlier reports that 
    it is an aneuploidy-inducing and cell-transforming agent, and 
    the present study also revealed that it was capable of causing 
    primary damage to DNA; it cannot, therefore, be regarded as 
    non-genotoxic.  Phenobarbital was perhaps the previously best 
    documented non-genotoxin of those studied, but it has been 
    defined herein as a gene mutagen, a clastogen, and an 
    aneuploidy-inducing agent.  It is concluded that the phrase 
    "non-genotoxic" should only be used when a sufficiently large 
    genotoxicity data base has been collected. 

8.  Many of the assays evaluated may have an important research 
    role.  However, if their main function is to act as 
    confirmatory or complementary assays in conjunction with the 
     Salmonella test, then substantial redundancy among  in vitro  
    eukaryotic genotoxicity assays is evident. 

9.  In relation to the stated aim of the CSSTT, it is recommended 
    that resources be applied to the definition of a generally 
    acceptable and applicable protocol for the conduct of 
    chromosomal-aberration assays, preferably in an agreed cell 
    type.  Use of this class of assay in conjunction with an 
    adequate assessment of the mutagenicity of a chemical for 
     Salmonella should provide an efficient primary screen for 
    possible new carcinogens.  Certain chemicals that induce 
    tumours in rodents at elevated dose levels will remain 
    undetected by these tests, as exemplified by DEHP in the 
    present study.  The "detection",  in vitro, of such agents may 

    be achieved by the incorporation of additional assays, but the 
    price in terms of a reduction in overall specificity and the 
    increased burden on technical resources may be high. 
    Consideration of whether the joint inactivity of a chemical in 
    an adequate  Salmonella assay and an adequate  in vitro 
    cytogenetic assay provides evidence of its inactivity with DNA 
    and, thus, its probable non-carcinogenicity for mammals, at low 
    dose levels, is indicated.  The adoption of a chromosomal-
    aberration assay as a common complementary test would have the 
    additional advantages of allowing easy comparison of data, 
    ready access to supplementary cytogenetic assays, and the 
    provision of test data derived using a genetic end-point, 
    independent of that of the  Salmonella (gene-mutation) assay. 


AMES, B.N., MCCANN, J., & YAMASAKI, E.  (1975)  Methods for 
detecting carcinogens and mutagens with the salmonella/mammalian-
microsome mutagenicity test.  Mutat. Res., 31: 347-364. 

ASHBY, J.  (1983)  The distinction between genetic and non-genetic 
mechanisms in chemical carcinogenesis. In: Zbinden, G., Cohadon, 
F., Detaille, J.Y., & Mazué, G., ed.  Current problems in drug 
 toxicity, London, Paris, John Libby Eurotext. 

ASHBY, J. & LEFEVRE, P.A.  (1983)  Genetic toxicology with 
formaldehyde and closely related chemicals. In: Gibson, J.E., ed. 
 Formaldehyde toxicity, Hemisphere Pub. 

B.E., & SHELBY, M.  (1983)  The two IPCS collaborative studies on 
short-term tests for genotoxicity and carcinogenicity.  Mutat. Res., 
109: 123-126. 

MARGOLIN, B.H., MATTER, B.E., & SHELBY, M., ed.  (1985)   Evaluation 
 of short-term tests for carcinogens. Report of the International 
 Programme on Chemical Safety Collaborative Study on in vitro 
 assays, Amsterdam, Elsevier Science Publishers (Progress in 
Mutation Research, Vol. 5). 

ATHANASIOU, K. & KRYRTOPOULOS, S.A.  (1981)  Induction of SCE by 
non-mutagenic carcinogens.  NATO Adv. Study Inst. Ser A, 40: 557-562 
(Chem. abst., 97 18839 (1982). 

BARRETT, J.C., WONG, A., & MCLACHLAN, J.A.  (1981)  
Diethylstilboestrol induced neoplastic transformation without 
measurable gene mutation at two loci.  Science, 212: 1402-1404. 

BUCKTON, K.E. & EVANS, H.J.  (1982)  Human peripheral blood 
lymphocyte cultures. An  in vitro assay for the cytogenetic effects 
of environmental mutagens. In: Hsu, T.C., ed.  Cytogenetic assays of 
 environmental mutagens, Osman, New Jersey, Allanheld. 

CAIRNS, J.  (1981)  The origin of human cancers.  Nature (Lond.), 
289: 353-357. 

CLEMMENSEN, J. & HJALGRIM-JENSEN, S.  (1977)  On the absence of 
carcinogenicity to man of phenobarbital. In:  Statistical studies in 
 malignant neoplasms, Vol. 5, pp. 38-50. 

DEAN, B.J.  (1978)  Genetic toxicology of benzene, toluene, 
xylenes, and phenols.  Mutat. Res., 47: 75-97. 

DE SERRES, F.J. & ASHBY, J., ed.  (1981)   Evaluation of short-term 
 tests for carcinogens: report of the international collaborative 
 program, North-Holland, New York, Elsevier. 

GOLDMACHER, V.S. & THILLY, W.G.  (1983)  Formaldehyde is mutagenic 
for cultured human cells.  Mutat. Res., 116: 417-422. 

HUFF, J.E.  (1982)  Caprolactam: condensation of the carcinogenesis 
bioassay technical report.  Environ. Health Perspect., 45: 199-200. 

KLEIN, G.  (1981)  The role of gene dosage and genetic 
transpositions in carcinogenesis.  Nature (Lond.), 294: 313-318. 

MILVY, P. & WOLFF, M.  (1977)  Mutagenic studies with 
acrylonitrile.  Mutat. Res., 48: 217-278. 

MOODY, D.E. & REDDY, J.K.  (1978)  Hepatic peroxisome (microbody) 
proliferation in rats fed plasticizers and related compounds. 
 Toxicol. appl. Pharmacol., 45: 497-504. 

PHILLIPS, B.J., JAMES, T.E.B., & GANGOLLI, S.D.  (1982)  
Genotoxicity studies of di(2-ethylhexyl)phthalate and its 
metabolities in CHO cells.  Mutat. Res., 102: 297-304. 

point of mutation is responsible for the acquisition of 
transforming properties by the T24 human bladder carcinoma 
oncogene.  Nature (Lond.), 300: 149-152. 

STYLES, J.A  (1977)  A method for detecting carcinogenic organic 
chemicals using mammalian cells in culture.  Br. J. Cancer, 36: 558-563. 

E.H.  (1982)  Mechanism of activation of a human oncogene.  Nature 
 (Lond.), 300: 143-149. 

TOMITA, I., NAKAMURA, Y., AOKI, N., & INUI, N.  (1982)  
Mutagenic/carcinogenic potential of DEHP and MEHP.  Environ. Health 
 Perspect., 45: 119-125. 

UKEMS  (1983)   Report of the UKEMS sub-committee on guidelines for 
 mutagenicity testing, Part I, The United Kingdom Environmental 
Mutagen Society. 

UNITED STATES PUBLIC LAW  (1958)   United States Code, title 21, 
 section 348(c)(3)(A). 

US NTP  (1980)   Benzoin - NTP technical report on the toxicology 
 and carcinogencity of benzoin, Research Triangle Park, North 
Carolina, National Toxicology Program (TR No. 204). 

US NTP  (1982)   Caprolactam - NTP technical report on the 
 toxicology and carcinogenicity of benzoin, Research Triangle Park, 
North Carolina, National Toxicology Program (TR No. 214). 

VENITT, S., BUSHELL, C.T., & OSBORNE, M.  (1977)  Mutagenicity of 
acrylonitrile (cyanoethylene) in  Escherichia coli. Mutat. Res., 45: 

VENITT, S., CROFTON-SLEIGH, C., & OSBORNE, M.B.  (1984)  The hair 
dye reagent 2-(2,4-diaminophenoxy) ethanol is mutagenic to 
 Salmonella typhimurium. Mutat. Res., 135: 31-47. 

WEINSTEIN, I.B., WIGLER, M., & STADLER, U.  (1976)   Analysis of the 
 mechanism of chemical carcinogenesis of epithelial cell cultures, 
Lyons, International Agency for Research on Cancer, (IARC 
Scientific Publication No. 12, Screening tests in Chemical 

WEISS, R.  (1982)  The myconcogene in man and birds.  Nature 
 (Lond.), 299: 9-10. 

WILLIAMS, G.M.  (1981)  Liver carcinogenesis: the role for some 
chemicals of an epigenetic mechanism of liver-tumour promotion 
involving modification of the cell membrane.  Food Cosmet. Toxicol., 
19: 577-583. 

    See Also:
       Toxicological Abbreviations