Genetic effects in human populations, guidelines for the study of (EHC 46, 1985)

INTERNATIONAL PROGRAMME ON CHEMICAL SAFETY

ENVIRONMENTAL HEALTH CRITERIA 46

GUIDELINES FOR THE STUDY OF GENETIC EFFECTS
IN HUMAN POPULATIONS

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
chemicals.

ISBN 92 4 150186 5

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
letters.

CONTENTS

GUIDELINES FOR THE STUDY OF GENETIC EFFECTS IN HUMAN POPULATIONS

CONTRIBUTORS

PREFACE

1. INTRODUCTION

2. METHODOLOGICAL AND EPIDEMIOLOGICAL ISSUES

2.1. General considerations
2.2. Methodological issues
2.3. Epidemiological considerations - common components
2.3.1. Study samples
2.3.2. Comparison samples
2.3.3. Confounding variables
2.3.4. Interaction
2.3.5. Methods for estimating exposure
2.3.6. End-points
2.3.7. Dose-response relationships
2.3.8. Time-response relationships
2.3.9. Problems in sample size and interpretation of
findings
2.4. Long-term medical follow-up
2.4.1. Relevance and limitations
2.4.2. Personal identification
2.4.3. Starting-point records
2.4.4. End-point records
2.4.5. Methods required for integrating source files

3. MUTATIONS IN SOMATIC CELLS

3.1. General considerations
3.1.1. End-points
3.1.2. Study requirements
3.2. Chromosomal end-points
3.2.1. Structural aberrations
3.2.1.1 Methods of culture
3.2.1.2 Fixation and slide preparation
3.2.1.3 Analysis of cells
3.2.1.4 Data analysis
3.2.1.5 Radiation-induced aberrations and
estimates of exposure
3.2.1.6 Chemically-induced aberrations and
estimates of exposure
3.2.1.7 Conclusions
3.2.2. Sister-chromatid exchange (SCE)
3.2.2.1 Formation of sister-chromatid exchange
3.2.2.2 Relevance of sister-chromatid exchange

3.2.2.3 Factors potentially influencing SCE
frequency
3.2.2.4 Methods for sister-chromatid exchange
analysis
3.2.2.5 Data processing and presentation
3.2.2.6 Conclusions
3.3. Gene mutations
3.3.1. Principles and basis for the methods
3.3.1.1 Autoradiographic method
3.3.1.2 Cloning method
3.3.2. Relevance and limitations
3.3.2.1 Relevance
3.3.2.2 Limitations
3.3.3. Procedures for assay of TG^r T-lys arising in vivo
in human beings
3.3.3.1 Autoradiographic method
3.3.3.2 Cloning method
3.3.4. Data presentation and analysis
3.3.4.1 Autoradiographic method
3.3.4.2 Cloning method
3.3.5. Conclusions

4. GERMINAL MUTATIONS

4.1. Introduction
4.1.1. Approaches for detecting germinal mutations
4.1.1.1 Detection of chromosomal mutations
4.1.1.2 The biochemical approach to detecting
point mutations
4.1.1.3 Indicator phenotypes
4.1.2. Methodological considerations and strategies
4.1.2.1 Sample acquisition and storage
4.1.2.2 Timing of studies
4.1.3. Summary
4.2. Germinal chromosomal abnormalities
4.2.1. Principles and basis of the method
4.2.2. Relevance and limitations
4.2.2.1 Studies of induced abortions
4.2.2.2 Studies of fetal deaths
4.2.2.3 Studies of prenatal diagnosis specimens
4.2.2.4 Studies of live births
4.2.2.5 Studies of indicator phenotypes of
chromosomal abnormalities
4.2.3. Procedures
4.2.3.1 Fetal specimens
4.2.3.2 Live births and other offspring
4.2.3.3 Detection of "indicator" phenotypes for
germinal chromosomal mutations - trisomies
4.2.3.4 Data presentation
4.3. Biochemical approaches to detecting gene mutations in
human populations

4.3.1. Biochemical methods for monitoring for gene
mutations
4.3.1.1 One-dimensional electrophoresi
4.3.1.2 Two-dimensional electrophoresi
4.3.1.3 Enzyme activity
4.3.1.4 Other biochemical approaches

4.3.2. Analytical strategy and methodological
considerations
4.3.2.1 One-dimensional electrophoresis
4.3.2.2 Two-dimensional electrophoresis
4.3.2.3 Enzyme activity
4.3.2.4 Sample acquisition and storage
4.3.3. Data management
4.3.4. Considerations in screening for germinal mutations
4.3.4.1 Sample size
4.3.4.2 Distinction between "true" and "apparent"
mutations
4.3.4.3 Implementation of gene mutation screening
programmes
4.3.5. Summary
4.4. Sentinel phenotypes
4.4.1. Introduction
4.4.2. Basis of method
4.4.3. Relevance and limitations
4.4.4. Procedures
4.4.4.1 Surveillance of sentinel anomalies
4.4.5. Data intrepretation
4.4.5.1 Surveillance of sentinel anomalies
4.4.6. Conclusions
4.5. Fetal death
4.5.1. Introduction
4.5.2. Procedures
4.5.3. Data processing and presentation
4.5.4. Conclusions

REFERENCES

CONTRIBUTORS

The following experts participated in the preparation of this
monograph:

Dr R.J. Albertini^b, University of Vermont, College of Medicine,
Burlington, Vermont, USA

Dr K. Altland^b, Institute for Human Genetics, University of
Giessen, Giessen, Federal Republic of Germany

Dr A.V. Carrano^a,b, Biomedical Sciences Division, L-452, Lawrence
Livermore National Laboratory, Livermore, California, USA

Dr A. Czeizel^b, Department of Human Genetics, National Institute of
Hygiene, Budapest, Hungary

Dr G.R. Douglas^a,b, WHO Collaborating Centre, Mutagenesis Section,
Department of National Health and Welfare, Ottawa, Ontario,
Canada

Dr H.J. Evans^b, Medical Research Council, Clinical and Population
Cytogenetics Unit, Western General Hospital, Edinburgh,
Scotland, United Kingdom

Dr E.B. Hook^a,b, Bureau of Maternal and Child Health, New York
State Department of Health, Albany, New York, and Department of
Pediatrics, Albany Medical College, Albany, New York, USA

Dr J.R. Miller^a,b,c, Central Research Division, Takeda Chemical
Industries, Ltd., Osaka, Japan

Dr H. Mohrenweiser^a,b, Department of Human Genetics, University of
Michigan, Ann Arbor, Michigan, USA

Dr H.B. Newcombe^b, Deep River, Ontario, Canada

Dr R.J. Preston^a,b, Biology Division, Oak Ridge National
Laboratory, Oak Ridge, Tennessee, USA

Dr. E.M.B. Smith^b, International Programme on Chemical Safety,
World Health Organization, Geneva, Switzerland

Ms M. Smith^a,b, Vital Statistics and Disease Registries Section,
Statistics Canada, Ottawa, Ontario, Canada

Dr D. Warburton^b, Department of Human Genetics and Development,
Columbia University, New York, New York

-------------------------------------------------------------------
^a WHO/DNHW Consultation, July 27-28, 1981, Ann Arbor, Michigan,
USA
^b WHO/DNHW Task Group Meeting, September 18-21, 1984, Ottawa,
Ontario, Canada
^c Chairman, Ann Arbor and Ottawa meetings

NOTE TO READERS OF THE CRITERIA DOCUMENTS

Every effort has been made to present information in the
criteria documents as fully and as accurately as possible. 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 and re-evaluation of the conclusions contained in
the criteria documents.

PREFACE

Monitoring and assessment of effects on human health from
exposure to environmental agents of all types, with monitoring to
confirm that control measures are working effectively, are key
aspects of the World Health Organization's Environmental Health
Programme. Without objective data on effects, attempts to assign
causes can only be speculative. Speculation does not provide a
reliable basis for remedial action and the control of health
hazards. Knowledge of the presence and identity of mutagenic
agents in the environment and the extent of exposure to them is
therefore extremely important in interpreting and using the results
obtained from monitoring populations for genetic effects.

Within the World Health Organization, there are a number of
programmes that are concerned with the monitoring and assessment of
human exposures and effects on health.

Control of environmental health hazards is one such programme.
This has the objective that, by 1989, the countries actively
involved will have formulated and developed national policies and
programmes for the protection of the health of their populations
against environmental hazards. A number of projects on the
monitoring of air and water quality, food contamination, and of
selected pollutants in human tissues are carried out by the World
Health Organization. These activities are complemented by a more
recent component, the Human Exposure Assessment Location (HEAL)
project, which is devoted directly to the monitoring of human
exposure to certain pollutants in all environmental media. All of
these monitoring projects are implemented in conjunction with the
United Nations Environment Programme's Global Environmental
Monitoring System (GEMS).

The International Programme on Chemical Safety is a cooperative
activity of the World Health Organization, the International Labour
Office, and the United Nations Environment Programme. It resulted
from concern expressed on the hazards of exposure to chemicals and
the pressing need to identify and assess them, so that control
measures could be applied and safe use achieved. One of the means
of achieving safe use is the preparation and dissemination of
documents of practical use to those involved in implementing
chemical safety. Publication in the International Programme on
Chemical Safety environmental health criteria series emphasizes the
importance of a systematic approach to monitoring human populations
for genetic effects as an integral part of environmental health
management.

These guidelines are intended as a source of practical
information on the design and conduct of genetic studies on human
populations exposed, or suspected of being exposed, to mutagenic
agents. As they are guidelines, they do no include comprehensive
protocols for studies. However, attention is directed to important
details that must be included, as well as pitfalls to be avoided.
It is envisaged that this document will be of use both to those
already involved in the assessment of human genetic hazards and to

those who wish to become better informed on this subject. Although
future requirements are not ignored, these guidelines principally
encompass methods that are considered practicable at present.

The original concept for these guidelines arose from the
recommendations of a WHO Consultation on Genetic Monitoring held in
Ottawa, Canada, on October 17, 1980, organized by the WHO
Collaborating Centre on Environmental Mutagenesis, Department of
National Health and Welfare (DNHW) Canada, and held in conjunction
with an International Conference on Chemical Mutagenesis,
Population Monitoring, and Genetic Risk Assessment. The drafting
of the monograph, implemented initially by the late DR K.C. BORA
(DNHW), and subsequently directed by DR G.R. DOUGLAS (DNHW), who
was also responsible for editing the final draft, began with a
WHO/DNHW Consultation held in Ann Arbor, Michigan, USA, on
July 27 - 28, 1981. This meeting produced a draft document
containing material that formed the basis for this monograph.
Subsequently, a joint WHO/DNHW Task Group Meeting, held in Ottawa,
Canada, on September 18 - 21, 1984, completed the project.

* * *

Financial and other support for the meeting was provided by the
Department of National Health and Welfare, Canada. The United
Kingdom Department of Health and Social Security covered the costs
of printing.

1. INTRODUCTION

The extent to which human somatic and germinal mutation
frequencies may be increased by exposure to ionizing radiation
and to the variety of chemicals that characterize modern societies
has been a matter of concern in recent years. Somatic mutations,
either genic or chromosomal, are not transmitted to the offspring
of an exposed individual. However, increases in the frequency of
these mutations may contribute to an increase in the frequency
of acquired disorders, for example, cancer. Increases in the
frequency of germinal mutations, genic or chromosomal, are likely
to contribute to inherited defects in the offspring of individuals
exposed to mutagenic agents. There is, therefore, a clear need to
develop and apply methods to study exposed populations at risk of
increased levels of somatic or germinal mutations.

Any effort to determine whether an increase in mutation rate
has occurred in a given population must contend with formidable
problems, one of the most significant of which is defining the
genetic end-points suitable for study. Germinal mutations, either
genic or chromosomal, can give rise to a plethora of phenotypes,
only a few of which are useful and meet the criteria required
for studies in mutation epidemiology. Likewise, using current
techniques, only a subset of somatic-cell mutations are amenable
to study. However, since it is much easier to obtain large samples
of somatic rather than germinal events, it would be ideal if some
measure of germinal mutations and the health effects of their
consequent phenotypes could be conveniently extrapolated from
studies of somatic events. Although such an extrapolation has
been attempted for the effects of ionizing radiation (Brewen et
al., 1975), it is not possible, at present, to attempt the same for
chemical effects. Despite these difficulties, it is essential that
efforts be made to initiate studies aimed at measuring somatic and
germinal genetic changes and assessing the relationships between
the two.

Although similar mutational changes will occur in somatic
and germ cells,the methods for detecting them are quite different.
An increase in the frequency of somatic mutations, genic or
chromosomal, can be established from relatively few individuals,
provided that a large number of cells is analysed from each sample
(Bloom, 1981). In contrast, determining an increase in the
frequency of mutation in germ cells by examining affected offspring
involves large study populations; the smaller the increase to be
detected (or excluded), the larger the sample needed (Neel, 1980;
Vogel & Altland, 1982). International cooperation may be necessary
to obtain adequate samples. However, because of the costs involved
in mounting large-scale epidemiological surveys, it is essential
that they should be designed to make the most efficient use of
resources and that the tests used should be designed to give
meaningful results. The general principles are the same for
all countries, and comparability can be achieved through
standardization of many of the procedures. Since the practical
details of the strategy and tactics in any one country will be

influenced by local circumstances, the emphasis in these guidelines
will be placed on general principles and examples of possible
procedures.

2. METHODOLOGICAL AND EPIDEMIOLOGICAL ISSUES

2.1. General Considerations

Certain general principles influence the design of
epidemiological studies for genetic effects and the nature of
the procedures for data gathering and analysis. The information
requirements are basically similar, whether changes in the genetic
make-up of somatic cells or germ cells in the exposed individuals,
or inherited changes affecting the offspring of such individuals
are being investigated. The key elements for any such study should
include:

(a) means of identifying the exposed (or affected) population
to be studied;

(b) availability of a reasonably similar comparison group, to
serve as the unexposed (or unaffected) controls;

(d) prior knowledge of the likely end-points that could serve
as indicators of genetic damage;

(e) where possible, some separation into different levels of
exposure with which to investigate a likely dose-response
relationship; and

(f) a means of observing the time course of the response.

Not all of these will be equally available for any study.
However, none of the elements that are available should be
overlooked in the initial documentation or the subsequent
analysis. In general, the need to guard against oversights in
necessary information will tend to increase with the interval of
time between the causal events and the resulting expressions of
harm.

Wherever long-term follow-up is involved, particular attention
should be paid to the personal identification of the individuals so
that starting-point data and end-point data will be appropriately
and unambiguously matched.

2.2. Methodological Issues

Not all studies will have a broad information base at the
outset. However, useful work can still be done, even where there
is no prior indication of harmful exposures, no suggestion of which
persons might be exposed, and no clear idea of the end-point
effects to expect. In principle, watch can be kept over a wide
range of potential indicators of genetic damage, among large
populations, such as the new-born, and this should be continued
over a long period of time. In this broad kind of continuing
observation, it is the temporal changes in the frequency of

end-points, above familiar background rates, that alert the
investigator to a possible mutagenic effect. The term
"surveillance" is used to describe this type of activity.

Methods for determining the frequency of a number of end-points
are discussed in this monograph. Various techniques are used for
determining germinal mutations. Indicators include potential
germinal effects, adverse reproductive outcomes, or other health
problems that may have a genetic basis.

Because of size considerations, it will often be necessary to
combine data from many studies. This will be greatly facilitated
if all investigators adhere to standardized laboratory procedures
and record-keeping practices including identification of the
individuals, with standard nomenclature for causes of death,
congenital anomalies, and protein and chromosomal variants.

2.3. Epidemiological Considerations - Common Components

The following comments apply to all methods and end-points.

2.3.1. Study samples

The appropriate epidemiological study design for examining the
possible hazards of agents must be determined. A cohort design
will be used in most studies where there is known exposure to a
suspected mutagenic agent. Alternatively, a case-control study
design may be used when exposure is common (e.g., smoking,
caffeine), or rare phenotypes are the end-points of interest
(e.g., new mutant phenotypes). In a case-control study, groups
of individuals are selected on the basis of whether they have the
genetic end-point under study. Also, in a case-control study, a
number of exposures can be evaluated in relation to a selected end-
point, in contrast to a cohort study in which a number of diseases
are evaluated in relation to one or more exposures.

In determining the population to be studied, several different
approaches can be taken. For example, questionnaires can be used
to determine if individuals in a particular setting have been
exposed to a putative hazard, and to gather data on possible
controls, or existing data sources (e.g., vital records, census
data, employment and work history records, special registries,
populations of specific geographical locations) may be used.
Regardless of the approach used, sufficient personal identifying
information must be recorded regarding the individuals to allow for
follow-up and for searching any large files.

2.3.2. Comparison samples

The identification and use of appropriate controls is
essential. In both cohort and case-control studies, a basis for
comparison is required, in order to evaluate whether the outcome
observed differs from that expected, had there not been any
increased risk from the agent under study. In certain cases a
population may serve as its own control by sampling before, during,

and after exposure (internal controls in a longitudinal study).
Since the frequency of a given end-point may vary according to age,
sex, socioeconomic status, and other variables, it is important
that this is taken into consideration when making the comparisons.

2.3.3. Confounding variables

Confounding can be described as the mixing of effects caused by
variables that are associated with both the exposures and end-
points to be studied, or a distortion of the apparent effect of an
exposure on risk brought about by the association with other
factors that can influence the outcome. In this regard, the
following should be considered:

(a) variables known to cause mutagenic effects should be
identified as far as possible (e.g., cigarette smoking,
individuals known to have undergone radiotherapy);

(b) cases and controls should be appropriately matched (e.g.,
for age and sex) in such a way that they have the same
distribution with respect to known confounding variables;
and

2.3.4. Interaction

If an exposure is associated with an increased risk of a
particular end-point, it is important to determine whether the
risk is additive to that of other known causes of disease, or
synergistic in its effect.

2.3.5. Methods for estimating exposure

The following methods can be used to estimate either the levels
of exposure, or values that are proportional to the levels of
exposure:

(a) estimates of length of exposure as well as ambient levels
of the agent may be the best or only estimate of dose in a
working environment; it is recognized that such estimates
are crude, but, in some situations, they may be all that
is available;

(b) direct measurement of an agent in the environment (e.g.,
air or water sampling) may be used;

(c) direct measurement of an agent or its metabolites in body
fluid and tissues to estimate body burden (e.g., blood,
urine, hair, teeth);

(d) observation of other pathological evidence of organ or
tissue damage (e.g., liver damage, chloracne) may be
helpful in estimating doses; and

(e) the use of questionnaires related to work histories and
lifestyles may assist in determining the exposure of
concern or other exposures.

In some cases, there may be ample time to obtain exposure
data, but, with emergency situations, the information may be only
transiently available. Thus, it must be collected early and the
people involved, recruited, or identified before they disperse.

2.3.6. End-points

Effects of somatic or germinal origin are considered in
sections 3 and 4 of this document, respectively.

2.3.7. Dose-response relationships

The existence of a dose-response relationship - that is, an
increase in disease incidence with increase in amount of exposure -
supports the view that an association is a causal one. Thus, it is
desirable to attempt to quantify exposure as far as possible.
However, under certain circumstances, in order to make the best use
of limited resources, initial studies could be restricted to those
with high exposure doses. Individuals so exposed would be compared
to an unexposed population. If no differences were found, the
study could be terminated, but it should be noted that, at high
doses, excess cell lethality might mask genetic effects. If
differences were observed, effects at intermediate dose levels
should be studied, paying special attention to possible confounding
synergistic effects.

Estimates of doses for populations that have been exposed
accidently will most likely be less accurate than for those exposed
occupationally and medically. In such cases, it may be more useful
to estimate the upper boundary of the dose (i.e., maximum possible
dose) rather than the mean, or to stratify the population into 2 or
3 groups whose bounds, though wide, probably do not overlap enough
to wipe out the suspected differences between them.

2.3.8. Time-response relationships

Consideration of the optimal sampling time is critical for
quantifying the somatic end-points in the exposed individual.
Appropriate sampling times should be selected to maximize an
effect, whenever this is possible. This is discussed further in
section 3.

2.3.9. Problems in sample size and interpretation of findings

The ability to detect an effect of a given incidence will
depend on the sample size that can be studied and the base-line
frequency of a given end-point in the unexposed population. The
more frequent the end-point, the smaller the sample size needed to
detect a given effect. The required sample size will also depend
on the chosen values of alpha, the probability of falsely rejecting
the null hypothesis, and beta, the probability of falsely accepting

the null hypothesis. These are sometimes known as Type 1 and Type
2 errors. It is customary, in most experimental situations, to
choose alpha = 0.05 and beta = 0.20 (2-tailed test). However, in
order to be more certain that a real effect is not missed, then an
alpha = 0.10 (or 0.05 for a one-tailed test) and beta = 0.05 could
be chosen. However, this choice will increase the sample size
required to rule out a given effect incidence.

It is possible, for given values of alpha and beta, to
calculate the sample size required to rule out a given effect
incidence, for an end-point of known frequency in the exposed
population. Conversely, the effect size detectable with a given
sample size can be calculated. Table 1 shows the relationship
between sample size and effect incidence for a given degree of
assurance that the effect is real.

In any study, a negative result should be expressed in terms of
the size of the effect that can be ruled out at a given power. For
example, "this study rules out a 2-fold increase in the frequency
of end-point X, with 80% confidence" or "this study has a 50% power
to detect a doubling of the frequency of end-point X".

Further considerations of sample size will be discussed under
the sections devoted to each specific test end-point.

Table 1. Sample sizes needed to detect a given increase in
frequency, with alpha = 0.05^a
-----------------------------------------------------------
Frequency Sample size to detect Sample size to detect
of a doubling a tripling
end-points beta = 80 beta = 95 beta = 80 beta = 95
-----------------------------------------------------------
.000001 23 5551 082 39 001 131 7 850 356 13 000 369
.00001 2 355 076 3 900 058 785 021 1 300 011
.0001 235 475 389 951 78 487 129 976
.001 23 515 38 940 7834 12 972
.01 2319 3839 769 1272
.05 435 719 141 232
.10 200 329 62 102
.15 121 199 36 58
-----------------------------------------------------------
^a Two-tailed probability without correction for continuity.
For discussion of continuity correction, and computation
formula, see Fleiss (1981).

2.4. Long-Term Medical Follow-up

People need to be identified uniformly on various files in
order to monitor human populations for delayed effects caused by
exposure to environmental agents. Epidemiological studies are
greatly facilitated by the existence in many countries of
centralized, computerized, accessible national registries relating
to health outcomes.

The kinds of data required are concerned with establishing
statistical associations, which may serve as pointers to possible
"troublespots", or with investigating suspected exposed groups. An
investigation involves defining the population cohort for study,
the end-points of interest, and the appropriate study methods for
carrying out individual long-term follow-up efficiently (WHO,
1983).

A number of existing population-based files, concerned with
health events, are available as end-point records that can be used
to measure various health effects among a cohort under study
(Bloom, 1981). Normally, follow-up will consist simply of using a
starting-point record, in its machine-readable form, to search by
computer for an end-point record relating to the same individual or
family. Procedures have been developed for doing this and for
analysing the results, without loss of personal privacy (Smith &
Newcombe, 1980).

2.4.1. Relevance and limitations

Lack of adequate recorded information and of organization and
use of historical data in a consistent fashion are two major
stumbling blocks in long-term epidemiological or genetic research.
The term "record linkage" has been used to describe the process
whereby two or more records relating to the same individual,
family, or event are brought together. The success of this
procedure depends on the quality, quantity, and discriminating
power of the items of personal identification on the machine-
readable records that are to be brought together.

Existing historical data should be more readily accessible
for statistical analysis, but with built-in safeguards so that
confidentiality of health records and personal privacy are not
compromised. Examples of relevant data are: a) mortality data;
b) birth defect monitoring programmes; c) health surveillance
programmes; d) vital records, such as marriages and births; e)
morbidity records; and f) cancer registries. In addition, where
new data are being collected (e.g., on suspected high-risk groups,
or on individuals who exhibit "early indicator" conditions), this
should be done in such a fashion that it is possible to make
international comparisons and to pool data.

2.4.2. Personal identification

Each record needs to contain enough information to indicate
unambiguously the particular individual and/or family to whom it
refers (Table 2). If a universal numbering system were available
for each individual in the population, and were in general use on
all medical and vital health records, the problem of matching would
almost disappear. In the absence of such a universal number, the
record should ideally contain full birth names, birth date,
birthplace, sex, mother's maiden surname, and place of residence.
Other useful items are full current surname and current address.
Since records being linked will seldom have been generated at
exactly the same time, and, since some of the items are subject to

change with time, it is desirable that the date of the event and
the type of event be known. For example, the date of event (e.g.,
hospital visit) may infer a "last-known alive date", which could
substantially reduce the amount of scanning where a mortality
search is being carried out.

To identify the family involved, the marital status of the
individual, and, if applicable, the spouse's birth surname,
forenames, and birthplace should be recorded. In the case of
birth, marriage, death, cancer registries, and especially genetic
registries, it is desirable to have parental birth names,
birthplaces, and birth dates (or ages).

Table 2. A list of items to be included on starting- and end-point
records to facilitate follow-up studies^a
-------------------------------------------------------------------
* 1. Surname
* 2. Previous surname (if any)
* 3. First given name
* 4. Second and other given names
* 5. Usual name (or nickname)
* 6. Sex
* 7. Birth date (year, month, day)
* 8. Birth province or country
* 9. Birth city or place
* 10. Father's surname
* 11. Father's first given name
* 12. Father's second given name
+ 13. Father's birth province or country
14. Father's birth city or place
+ 15. Father's birth date (or age)
* 16. Mother's maiden name
+ 17. Mother's first given name
+ 18. Mother's second given name
+ 19. Mother's birth province or country
20. Mother's birth city or place
+ 21. Mother's birth date (or age)
* 22. Marital status
* 23. Spouse's birth surname
* 24. Spouse's first given name
25. Spouse's second given name
26. Spouse's birth province or country
27. Spouse's birth city or place
28. Social Security Number or equivalent
29. Health Insurance Number
* 30. Place of residence - province or country
31. Place of residence - complete address including postal
code
* 32. Date of event
* 33a. Last known alive date (e.g., date of last contact)
33b. Date hired by company
33c. Date left company
34. Principal lifetime occupation - type of work, type of
business, length of time worked
-------------------------------------------------------------------

Table 2. (contd.)
-------------------------------------------------------------------
35. Other items where applicable (e.g., birth order of child,
status of birth, religion, race, etc.)
* 36. Control code to indicate the kind of event
* 37. A control code digit to indicate whether alternate entries
for the same event are being recorded (e.g., where an
individual may have alternate spellings for surname)
* 38. A unique number (where no other suitable number is
available)
39. Where applicable, an indicator to denote whether the
individual is known to be dead,the date of death, and the
province or country of death
-------------------------------------------------------------------
^a From: Smith (1977).
*,+ Top priority would be given to collecting the items identified
with asterisks. For genetic studies, additional parental
variables will be required, particularly those indicated with
+. Information relating to diagnostic procedures, work
histories, exposure histories, etc., plus updates, are added to
the basic record. For marriages, the record should identify
the groom, bride, groom's parents and bride's parents.

Records in files can contain erroneous information and
omissions. Thus,in order to reduce the frequency of matching
errors, it is necessary to introduce a measure of "redundancy"
into the identifying information. Alternative entries can also be
created, where the surname of an individual may have an alternative
spelling (e.g., adoptions). It may also be useful to record
nicknames, race, principal lifetime occupation, and birth order.

2.4.3. Starting-point records

Where information is sought concerning the likely health
effects of possible harmful agents in the environment, and of
various social and economic circumstances, the chief limiting
factor is a shortage of starting-point records in a form suitable
for searching death records, cancer registries, etc., for evidence
of harm to the exposed individual or his/her offspring.

Collaboration and cooperation is required among a variety
of organizations to help resolve this problem (e.g., Regulatory
Agencies, Departments of Labour and Health, and research workers).
Various laboratory tests and studies are being carried out to
predict whether particular chemicals would be likely to cause
mutations or cancers in human beings exposed over long periods to
low or moderate doses. Human data relating to subpopulations can
serve as a potential starting point for a variety of studies, but
the records need to be uniformly stored and readily accessible.
Studies involving large cohorts can be carried out at a national
level. International cooperation is necessary in cases where the
numbers are small, and pooling of data is required. An example
would be to follow up people who show "early indicator" conditions.
For instance, individuals who show chromosomal breakage could be
identified, and followed up to see whether they subsequently

developed cancer or had reproductive problems (e.g., stillbirths)
or defective children. Similarly, children found by survey to have
protein variants could be followed up to see whether they developed
special health problems or died early.

Starting-point records can comprise a wide assortment of
administrative and other kinds of microdata files. These may
include ad hoc local files (e.g., nominal rolls created using
personnel, pay, and pension records), specialized centralized
registers (e.g., health surveillance registers), records of
exposure to certain agents, records of employment, special
survey records, vital registrations, or other sources.

2.4.4. End-point records

A number of health events are routinely documented in forms
that are centralized to a certain extent and are in a machine-
readable form that lends itself to studies in which a measure of
the health impact on a particular cohort is required. Included
among such sources are death registrations (Kinlen, 1980;
Patterson, 1980; Smith & Newcombe 1980), and special registers of
cancer, congenital anomalies, and other handicapping conditions.
Secondary sources that might be helpful include marriage records,
to give changes of name for follow-up and family composition, and
birth records to indicate offspring at potential risk. Both birth
and stillbirth records are of particular value, where fertility is
of interest, for indicating a potential health effect. Hospital
and medical insurance records, may be useful to help convert
locally-based registers into population-based ones, and to give
a measure of the social burden of disease.

2.4.5. Methods required for integrating source files

Computer methods have been developed to derive family
statistics relating to hereditary and congenital disease and terms
of other ill health (Newcombe, 1967, 1969, 1977; Smith, 1977,
1980). For example, combining records from the vital registration
system, certain specialized disease registers, and records from a
universal hospital insurance scheme made multi-generation and
cousin studies possible (Table 3 and 4). However, individual
follow-up is in much greater demand, primarily for detecting
delayed cancers.

For linking an individual with health and other relevant
records, a unique identity number assigned at birth would be
extremely useful. These are not available in many countries.
Where no unique number is available, the searching procedures
to be used in bringing records together may involve probabilistic
matching techniques. A generalized record linkage system and
probabilistic matching procedures have been developed and used for
such studies (Smith et al., 1980; Howe & Lindsay, 1981; Smith &
Newcombe, 1982).

Linkage projects have been developed or are being planned in a
number of international centres (Mi, 1967; Acheson, 1968; Skolnick,
1980; Beebee, 1981; Fox & Goldblatt, 1982; Smith, 1982), which
facilitate the establishment of: (a) mortality data bases; (b)
cancer incidence reporting systems; (c) death clearance of cancer
registers; (d) procedures for supplementing cancer registers with
hospital and medicare data; and (e) procedures for constructing
better handicap registers.
Table 3. Identifying information available in the magnetic tape files of vital and health records example from British Columbia^a
--------------------------------------------------------------------------------------------------------------------------------------
Event and its date and place Father Mother Child
Event Date City & Sur- Initials Age Province Maiden Initials Age Province Sur- Given Sex Birth Full City Province Other^b
(or Province name of birth surname of birth name names order birth of of birth
year*) date birth
--------------------------------------------------------------------------------------------------------------------------------------
Marriage X X X X X X X X X X (1)
Livebirth X X X X X X X X X X X X X X X X X (2)
Stillbirth X X X X X X X X X X X X X X X X X (2)
Death X X X X X X X X X X X X X X X (3)
Handicap X* X X X X X (4)
Congenital X* X X X X X (4)
malformation
surveillance
Hospital
Separation X X X X X X (5)
--------------------------------------------------------------------------------------------------------------------------------------
^a From: Newcombe (1977).
^b (1) Marital status of groom and bride.
(2) Legitimate or illegitimate status; singleton, twin, triplet, etc.
(3) If married women, own maiden name surname instead of mother's; marital status.
(4) Coded address representing city, school district and provincial health unit area; whether born in home province.
(5) Hospital code and admission number: maiden surname if different; marital status; address as city, street, and number; religion;
three initials of head of family, or spouse, or guardian; plus codes for relationship and whether address is the same.

Table 4. Example of the file organization for a sibship history of disease (fictitious)^a,b
---------------------------------------------------------------------------------------------------------------------------------
Event (birth Year Father Mother Child
order in Surname Initials Province Age Maiden Initials Province Age Given Sex Birth Disease
parentheses) of birth^c surname of birth^c names year month day
---------------------------------------------------------------------------------------------------------------------------------
Marriage 1948 Cox JW 07 25 Bell MA 09 21
Live birth (1) 1950 Cox JW 07 27 Bell MA 09 23 Annie May F 50 01 09
Live birth (2) 1952 Cox JW 07 29 Bell MA 09 25 Brian John M 52 04 14
Reg. handicap 1952 Cox - - - - - - - Brian John M 52 04 14 Spina bifida
Hospitalization 1955 Cox - - - - - - - Brian John M 52 04 14 Spina bifida
Death 1956 Cox JW 07 - Bell MA 09 - Brian John M 52 04 14 Spina bifida
Stillbirth (3) 1954 Cox JW 07 31 Bell MA 09 27 Mary Jane F 54 07 26 Anencephaly
---------------------------------------------------------------------------------------------------------------------------------
^a From: Newcombe (1977).
^b Obtained by merging the individual histories of birth, ill health and death, with the sibship histories of parental marriages
and births.
^c Coded from a 2-digit geographical code for all Canadian provinces and all major countries; 07 = Saskatchewan 09 = British
Columbia.

3. MUTATIONS IN SOMATIC CELLS

3.1. General Considerations

General reasons for studying somatic mutation in human
populations include:

(a) to determine if an unsuspected introduction of new (or
increase in already present) mutagens has occurred;

(b) to evaluate mutation frequencies in populations known (or
suspected) to be exposed to mutagens;

(d) to identify groups in the general population with high
frequencies of somatic mutations, so that studies can be
undertaken to attempt to detect the responsible factors;
and

(e) to define the heterogeneity of susceptibility to genotoxic
agents within the population.

3.1.1. End-points

Genetic damage can be assayed at the molecular, functional
gene, and chromosomal levels. At the molecular level, DNA damage
can be assessed in terms of adduct formation, strand breakage and
repair, or base sequence alterations. At present, the measurement
of adduct formation requires either prior knowledge of the nature
of the adduct or limited population sizes, and is, therefore, not
currently applicable as an assay for monitoring large populations.
DNA strand breakage and repair are usually transient phenomena
initiated at the time of insult and completed shortly afterwards,
making them less than ideal for population monitoring purposes.

Recent research developments have made it feasible to use
restriction enzyme mapping techniques to analyse alterations in
DNA base sequence (Botstein et al., 1980; Skolnick & Francke,
1982; Southern, 1982). This approach is being used to define gene
structure and identify polymorphisms for linkage analysis. The
same approach could equally well be used to screen for mutational
changes, but its application must await the availability of
appropriate batteries of DNA probes.

With current techniques, changes in gene function can be
detected by the acquisition of resistance to selective agents or
by the alteration or loss of cellular constituents detected by
immunological methods. Although these immunological approaches
hold considerable promise, the available data are insufficient
for application to routine monitoring. On the other hand,
methods for the detection of thioguanine-resistant peripheral
blood T-lymphocytes have been established and are being refined.
A detailed description of the principles and methods of this
approach is presented in section 3.3.

At the chromosomal level, genetic damage is observed as an
alteration either in chromosome number or in chromosome structure.
Although numerical changes represent a significant proportion of
human heritable genetic diseases, their consequences in somatic
cells are less well characterized. Moreover, estimates of
numerical changes in somatic cells are heavily influenced by
technical artifacts. Alterations in chromosome structure are more
accurately assessed in somatic cells and are observed cytologically
as chromosomal aberrations or sister chromatid exchanges (SCEs).
These are currently the most readily-available and widely-used end-
points for evaluating somatic mutations in human beings. There is
extensive literature concerning these methods (Office of Technology
Assessment, US Congress, 1983), which are discussed in section 3.2.
Micronuclei, which can result from structural or numerical
chromosome aberrations, have been used as an indicator of such
damage in human studies. Micronuclei have been enumerated in
cultured lymphocytes to detect the effects of ionizing radiation
(Krepinski & Heddle, 1983) and in the exfoliated epithelial cells
of persons exposed to chemical carcinogens (Stich et al., 1982).
However, this method, though showing potential for monitoring
genetic damage in human beings, has not yet been accepted or used
widely and so is not considered in detail.

3.1.2. Study requirements

Any study attempting to evaluate human populations for somatic
cell mutations should satisfy the following requirements:

(a) the study should include exposed groups and appropriate
concurrent controls;

(b) specimens from exposed and control individuals should be
gathered, handled, transported, and evaluated at the same
time and in the same way, using for example, the same
culture medium, serum, and reagents;

(c) samples should be coded and evaluated in such a way that
the the origin of the sample is not known, at the time of
evaluation;

(d) attempts should be made to quantify levels of exposure and
to determine dose-response relationships.

However, in some instances, concurrent controls may not be
necessary, for example, in longitudinal studies to determine
temporal changes in mutation frequencies. In these cases, any
possible confounding variables, such as changes in batch or source
of media, sera, or reagents used in the study should be minimized.
Reagents should be stored and, when there is a batch change,
comparative studies of selected individuals (or stored samples)
with new and old reagents should be undertaken to determine if
there are differences. It is recognized that it may be extremely
difficult to find appropriate controls in all instances, because of
various confounding factors that can influence somatic mutation
frequencies (e.g., age, sex, exposure to ionizing radiation,

medication, recent viral disease, smoking, etc). An appropriate
control group is one that is matched as closely as possible with
the exposed group for these confounding factors.

If it is impossible to find appropriate matched controls for
everyone in an exposed group, but it is necessary to study all
those exposed, the following alternatives are suggested:

(a) to obtain the best controls available and avoid the most
significant confounding factors;

(b) to stratify the cases, prior to analysis, into those with
well-matched controls and those with less well-matched
controls;

(c) to give the greatest weight in the analysis and
interpretation of results to the group with the well-
matched controls.

In emergency situations, when appropriate matched controls
are not available, it is important to obtain some control specimens
that will be sampled, transported, and cultured at the same time
as those from exposed individuals. Sources of such specimens
are members of study teams, assuming that they have not been
contaminated. Each batch of specimens, collected in emergency
circumstances and transported to the laboratory, should include a
specimen from at least one unexposed individual.

The questions arise as to when, and for how long after exposure
to a mutagen, individuals can usefully be studied. It may be
important to follow populations with elevated mutant frequencies
sequentially, to determine if such frequencies are changing.
Because of the long life span of the circulating peripheral
lymphocyte, some types of mutation can be detected many years after
the exposure. However, the persistence of these events cannot be
predicted with certainty and will vary with exposure duration,
agent, and other factors. Thus, for chromosomal aberration and
sister chromatid exchange measurements, samples should be taken as
soon as possible after exposure. For example, persons receiving
chemotherapy can have high frequencies of chromosomal aberration,
sister chromatid exchange (SCE), and gene mutation in the
lymphocytes. However, in the months following cessation of
treatment, the chromosomal aberration frequency decreases, often to
a level that may not be significantly different from that in the
controls. The frequency of SCE also diminishes in time. For gene
mutations, there may be a time interval between the mutational
event and the expression of resulting mutants. This interval will
vary from days to weeks and may vary with the markers. Thus,
following exposure, somatic mutant frequencies may remain the same,
decrease, or increase as a result of cell division, selection, or
migration in vivo. Similarly, persons therapeutically exposed to
ionizing radiation show elevated frequencies of chromosomal
aberration and gene mutation, but not SCE. In these patients, the
frequency of aberrations may remain elevated for many years after
exposure. At present, there are few data relevant to the

persistence of specific locus somatic mutants. It is recommended
that more than one end-point for somatic mutational change be
considered in any population study, because each end-point has
limitations with regard to the conclusions that may be drawn.
For example, negative findings for chromosomal aberrations do not
preclude the possibility of increases in somatic mutations of other
types. In a population presumably exposed to ionizing radiation,
chromosomal aberration studies are most appropriate, but SCE
could also be analysed as an indicator of exposure to confounding
chemical agents. Conversely, in a population exposed to chemicals,
SCE studies are more appropriate in terms of assay sensitivity, but
chromosomal aberration studies provide additional important
information.

Considerable effort can be expended in subject identification
and sample collection; therefore, sufficient blood should be
collected at one time to assay several somatic cell mutational end-
points. Even though it may be necessary to evaluate a number of
end-points in a particular study, it is worthwhile setting
priorities for their analysis. For example, slides could be
prepared for the determination of both chromosomal aberrations and
SCE, but only one end-point scored initially. This choice should
be based on the type of end-point most likely to be elevated. The
remaining slides can be scored later, according to the needs of the
study.

In addition, blood collected from mutagen-exposed and control
individuals should be sufficient to allow cryopreservation of
cells and other components for later study. This will make
more extensive evaluations possible as well as contribute to the
creation of a repository of material from mutagen-exposed human
beings for subsequent research.

3.2. Chromosomal End-Points

3.2.1. Structural aberrations

Chromosomal aberrations can be studied in any cycling cell
population, or in any non-cycling cell population that can be
stimulated by a mitogenic agent to enter the cell cycle. In
animals, there are several cell types that fit these criteria,
but for human studies, to all intents and purposes, there are only
two cell types that are practically suitable. These are the bone
marrow cells, which are a cycling population, and the peripheral
blood lymphocytes, which are normally non-cycling, but can be
stimulated to enter the cell cycle by in vitro culturing with a
mitogen such as phytohaemagglutinin (PHA). However, because of
the ease of obtaining blood samples, in contrast to bone marrow
samples, lymphocyte assays have been used in the majority of
studies on the induction of chromosomal aberrations in human
beings.

Since the observation by Moorehead et al. (1960) that
peripheral lymphocytes could be stimulated by PHA to enter the cell
cycle and be observed at metaphase, an enormous amount of data has
been obtained on the induction of chromosomal alterations by
radiation and chemical agents using this lymphocyte assay system
(Preston et al., 1981). It is often assumed that this is a simple
and informative assay for human population monitoring, providing
information on potential clastogenic (or mutagenic) exposures.
However, it should be emphasized that such a view should be
regarded with caution in the case of supposed chemical exposures,
and the obvious distinction from its use for radiation exposures
should be appreciated.

3.2.1.1. Methods of culture

There are numerous methods for culturing human lymphocytes.
Many components of the methods are optional, and depend on
individual preference. Those that are less variable are described
here. More complete methods can be found in Evans & O'Riordan
(1975), Bloom (1981), and Preston et al. (1981).

It appears to be advantageous to set up cultures from fresh
blood samples. However, this is not always possible, usually
because samples are taken at some distance from the laboratory and
have to be transported. It is still possible to achieve good
growth from samples taken several days before culturing. Although
samples can be maintained at 4 °C during shipping and storage, it
appears to be better to maintain them at room temperature whenever
possible (Dzik & Neckers, 1983).

Cultures can be established from whole blood (approximately
0.5 ml per culture), buffy coats, or purified lymphocytes.
However, if a large number of samples are to be handled at one
time, it is clearly advantageous to use whole blood. For buffy-
coat cultures, it is preferable to use the white-cell layer from
about 3 ml of centrifuged whole blood, thus larger total blood
samples are needed. One of the advantages of buffy-coat cultures
is that the cell fixation procedure is not hampered by the large
volume of red cells, present in whole-blood cultures, which are
often difficult to disperse. A possible disadvantage of whole-
blood cultures is that a small quantity of the agent under study,
or another contaminating agent, could be carried over with the
donor's serum. However, there seems little probability that this
would influence the results, and thus, it would not negate the use
of whole-blood cultures. It is also important to establish
cultures in duplicate, as a precautionary measure against culture
failure or sparseness of analysable cells.

A large number of different tissue culture media have been
successfully used, and the choice is a matter of personal
preference. This is also true for the serum type used, with the
added proviso that it should be virus-free. Several different
T-cell mitogens are also suitable, PHA being the most frequently
used. One additional point concerns the use of tissue culture
media containing a low concentration of folic acid. It has been

noted that the growth of human lymphocytes in such media (notably
TC 199) can result in the appearance of specific chromosome breaks
at so-called "fragile sites" (Jacky et al., 1983), or in an
increase in aberrations at possible fragile sites (Reidy et al.,
1983). It is recommended that, for population-monitoring studies,
low folate media should be avoided, though it is suggested that
parallel cultures with low folate media could still be used to
provide additional or different types of information.

While it is appropriate for each laboratory to use a medium and
serum that provides good lymphocyte growth conditions, it is most
important that each laboratory should determine the rate of
progression of mitogenically-stimulated cells through their first
and subsequent cell cycles with their specific culture conditions,
using blood samples from several individuals. The rate of cell
progression can be variable, depending on the mitogen, culture
medium, serum, and temperature (usually 37 °C in a 5% C0₂/95% air
environment).

The progression of cells can be measured by culturing in the
presence of bromodeoxyuridine (BrdU), sampling cells over a range
of times (for example, every 2 h, from 42 h to 54 h after mitogenic
stimulation), and staining the fixed preparations using a technique
for obtaining differentially-stained chromosomes (Goto et al.,
1978). In this way, the cells that have replicated their DNA once,
in the presence of BrdU, will contain evenly- and darkly-stained
chromatids. Those that have replicated twice, in the presence of
BrdU, will contain differentially-stained chromatids, one light
blue and one dark blue, when Giemsa stain is used. Cells that
progress through 3 or more cell cycles, in the presence of BrdU,
will contain some differentially-stained and some evenly but
lightly-stained chromosomes. In studies on chromosome aberrations,
it is important to analyse cells in their first metaphase after
mitogenic stimulation; thus, a fixation time should be chosen when
a high proportion of analysable cells are at the first division
stage. Generally, it is not feasible to use a fixation time when
all the cells are at this stage, because this requires very early
fixation (approximately 42 h after stimulation), when the number of
mitotic cells is too low. A compromise time (for example, 48 h) is
selected, when, for the average individual, about 90% of the cells
are at the first division stage. There is considerable variation
from individual to individual in the percentage of cells at the
first or subsequent division, even 48 h after culture initiation,
and it is good practice to check the proportion of first mitoses
in a series of cultures containing BrdU, separate from the series
established for aberration analysis. In this way, the possible
effects on aberration frequencies of analysing different
proportions of first division cells from different samples can be
ascertained. It is also possible to analyse chromosome aberrations
from cultures grown with BrdU, where preparations have been
differentially stained: cells showing no differentiation, and
cells that are clearly first metaphase (M1) can then be analysed.

The decision on selecting or rejecting samples with a high
proportion of 2nd divisions is optional, but should, at least, be
consistent. If resampling is possible, then this should be done
using an earlier fixation time than that for the first sample.
There is also a possibility that, at the standard fixation time for
any particular laboratory, there may be insufficient analysable
mitotic cells, as a result of an unusually long cell cycle, either
inherent or induced. The only solution to this problem is to
resample using a later fixation time.

3.2.1.2. Fixation and slide preparation

Many different methods are available for obtaining
conventional, banded or harlequin-stained metaphase preparations,
and essentially any one that produces well-spread complete
metaphases is acceptable (MacGregor & Varley, 1983). There is
little likelihood of this step of the assay influencing the
results.

3.2.1.3. Analysis of cells

There are many schemes available for the classification of
chromosome aberration types, and comprehensive descriptions can be
found in Savage (1975) and Bloom (1981). The following section
includes descriptions of the abberations most commonly observed in
control samples or samples from individuals exposed to radiation or
chemical agents.

(A) Chromosome-type aberrations

(1) Terminal and interstitial deletions

It is not possible to distinguish between chromosome-type
terminal deletions and non-sister union isochromatid deletions
(Fig. 2D), and so, in cases of radiation exposure, when induced
aberrations are of the chromosome-type, it is appropriate to
classify all paired acentric fragments as terminal deletions
(Fig. 1A).

The small interstitial deletions appearing as paired dots are
classified as "minutes" (Fig. 1A). The larger interstitial
deletions in which there is a clear space in the centre of the ring
are usually classified as acentric rings. The distinction is not
particularly clear-cut, and, in general, is merely an indication of
the different sizes of interstitial deletions. Acentric fragments
associated with inter- or intrachanges are not classified as
terminal deletions.

(2) Asymmetrical interchanges (usually dicentrics)

It is assumed that a dicentric, analysed at the first in
vitro metaphase will be accompanied by an acentric fragment
(Fig. 1B) and a tricentric by 2 acentric fragments (Fig. 1C). A
cell with a dicentric and two acentric fragments is, by convention,
classified as a dicentric with its accompanying fragment and a
terminal deletion. The two acentric fragments could be the
result of incomplete rejoining during the formation of the
dicentric. However, although this cannot be ascertained, it has
been shown experimentally that its probability of occurrence is
rather low (< 10%) (Schmid & Bauchinger, 1980).

Asymmetrical interchanges, i.e., dicentrics, can be analysed
with greater efficiency than any other aberration type (> 95%),
and it is their frequency that is generally used for estimations of
radiation dose. For such determinations, a tricentric, for
example, is assumed to be equivalent to 2 dicentrics.

(3) Asymmetrical intrachange (centric ring)

In asymmetrical intrachanges as in interchanges, a centric ring
is accompanied by an acentric fragment, and the same classification
scheme applies to these as for dicentrics ((2), above) (Fig. 1D).

(4) Symmetrical interchanges (reciprocal translocations)

Symmetrical interchanges (Fig. 1E) are particularly difficult
to observe in conventionally-stained preparations, unless the
exchanged pieces produce 2 chromosomes, very different from the
normal karyotype. However, it is suggested that obvious
symmetrical interchanges should be recorded, but giving less weight
to their frequency than to that of most other aberration types.

It is not usually possible to observe these symmetrical
intrachanges (Figs. 1F, 1G) in conventionally-stained preparations,
unless a pericentric inversion produces a chromosome that is
distinctly different from the normal karyotype. Any obvious
symmetrical intrachanges should be recorded, but less weight given
to their frequency than to that of most other aberration types.

(5) Symmetrical intrachanges (peri- and paracentric
inversions)

(B) Chromatid-type aberrations

Chromatid-type aberrations are generally classified in the
same way as chromosome-type aberrations; the apparent unit of
involvement in a chromatid-type aberration is, in most cases, the
single chromatid, and not the whole chromosome, as seen for
chromosome-type aberrations.

(1) Terminal deletions

A terminal deletion is a distinct displacement of the chromatid
fragment distal to the lesion, or, if there is no displacement, the
width of the non-staining region between the centric and acentric
regions is greater than the width of a chromatid (Fig. 2A). The
latter definition is used to distinguish between terminal deletions
and achromatic lesions or "gaps" (section (2)).

(2) Interstitial deletions

Chromatid-type interstitial deletions (Fig. 2B) are not as
readily observable as their chromosome-type counterpart, partly
because the small deleted fragment is often separated from the
deleted chromosome, and is not observed.

(3) Achromatic lesions ("gaps")

Achromatic lesions or gaps are non-staining or very lightly
stained regions of chromosomes, present in one chromatid (single)
or, in both sister chromatids, at apparently identical loci
(double). If the non-staining region is of a width less than that
of a chromatid, the event is recorded as an achromatic lesion (Fig.
2C). This is clearly only a working definition. It is generally
suggested that achromatic lesions should be recorded, but always
separately from chromatid deletions. Their frequency should not
be included in the totals for aberrations per cell, since their
significance and relationship to other "true" aberration types is
not clear at present.

(4) Isochromatid deletions

Isochromatid deletions appear as exceptions to the class of
chromatid-type aberrations, since they involve both chromatids,
apparently with "breaks" at the same position on both. However, in
suitable material they can be shown to be induced by radiation in
the S and G₂ phases of the cell cycle, as is the case for other
chromatid-type aberrations.

There are several possible types (Fig. 2D), depending on
the nature of the sister unions. If sister union occurs, it is
possible to distinguish isochromatid aberrations from chromosome-
type terminal deletions.

In mammalian cells, however, sister union is a rare event, and most
of the isochromatid deletions are of the non-union proximal and
distal type. The acentric fragment is most often not associated
with the deleted centric part of the chromosome. The convention
for analysis (discussed for chromosome-type terminal deletions,
above) is that, for radiation exposures, paired acentric fragments
are recorded as chromosome-type terminal deletions, but, for
chemical exposures or control samples, they are most appropriately
recorded as isochromatid deletions.

(5) Asymmetrical interchanges (interarm interchanges,
asymmetrical chromatid exchanges)

Asymmetrical interchanges are the chromatid-type equivalent to
chromosome-type dicentrics (Fig. 2E).

(6) Symmetrical interchanges (symmetrical chromatid
exchanges)

Symmetrical interchanges are the chromatid-type equivalent of
chromosome-type reciprocal translocations. In the case of
chromatid-type symmetrical exchanges, somatic pairing maintains an
association between the chromosomes involved in the exchange, and
thus they can be readily observed in the absence of any chromosome-
banding procedures (Fig. 2F).

(7) Asymmetrical and symmetrical intrachanges (interarm
intrachanges)

There are 2 forms of symmetrical and asymmetrical interarm
intfachanges but, when analysing metaphase cells, only one of each
is distinguishable (Fig. 2G). The symmetrical intrachange can be
observed in somatic pairing.

(8) Triradials

A triradial (3-armed configuration) can be described as the
interaction between one chromosome having an isochromatid deletion
and a second having a chromatid deletion. Many types of triradials

can be formed but, in mammalian cells, where the frequency of
sister union is low, the 2 illustrated are the most common (Fig.
2H).

It is important to note that the different types of chromosome
aberration should be recorded separately. The results for any
sample should be presented as the aberration frequency per cell
for each aberration type, or frequency of aberrant cells for each
aberration type. It is not appropriate to express the results as
total aberrant cells or total breaks per cell, as this method of
presentation can hide important facets of the data. For the
statistical analysis of data, it is legitimate to combine
aberration classes into the general categories of chromatid-type
deletions, chromosome-type deletions, chromatid exchanges, and
chromosome exchanges. Occasionally, cells are seen with multiple
aberrations, and an attempt should be made to analyse such cells in
detail, rather than to record them simply in a category "multiple
aberration", since this classification will mean a loss of
information. If it proves impossible to analyse a cell, it should
be recorded in a category of "too many aberrations to analyse", and
not included in the total of cells scored.

The selection of cells to be analysed is particularly
important, to avoid observer bias. Such bias can result in either
a higher or lower measure of the unbiased aberration frequency,
depending on the nature of the selection. The generally accepted
method of cell selection is to scan slides at low magnification;
cells that appear to be suitable are then analysed at high
magnification. Once a cell is observed at high magnification,
it should only be rejected if: (a) the cell cannot be analysed
because of the number or complexity of aberrations; in this case,
it should be appropriately recorded, but not included in total
cells analysed; (b) the cell is not sufficiently well spread
when seen at high magnification, and the overlap of chromosomes
prohibits an accurate analysis; (c) non-chromosomal material, such
as dirt or stain crystals, not discernible at low magnification,
prevent complete analysis; (d) the cell contains fewer centromeres
than the observer's acceptable cut-off point; this may be the
analysis of cells with 46 ± 2 centromeres or only cells with 46
centromeres, either cut-off point is acceptable, provided that the
criterion is consistently applied.

It is particularly important in population-monitoring studies
for the samples to be analysed under code, and for the slides from
exposed and control individuals to be randomized in such a way that

the observer is unaware of the identity of the sample. There are
strong arguments in favour of more than one laboratory, with at
least 2 individuals in each laboratory, being responsible for the
analysis, in order to remove possible laboratory or individual
scoring bias, which could influence the results. Microscope
co-ordinates of every suspected aberrant cell should be recorded.
In order to provide further consistency, the classification of any
cell as aberrant by one individual should be confirmed by a second,
and it is advantageous to keep a photographic record of all
confirmed aberrant or otherwise suspect cells.

The number of cells to be analysed from any sample cannot be
fixed with any certainty, because it can be argued that this number
depends on the aberration frequency, and thus is retrospective. A
general suggestion would be to analyse 200 cells per sample.

The pros and cons of analysing banded preparations have been
discussed frequently, and do not warrant repetition. The amount of
time expended on the preparation and analysis of banded samples
does not seem to be justified by the increase in sensitivity of
analysis for aberration types detectable by banding. However, the
choice of scoring conventially-stained or banded preparations
should be left to the discretion of each laboratory.

3.2.1.4. Data analysis

Many statistical methods are available for the analysis of
data obtained from human population-monitoring studies. The
appropriateness of any particular method will depend, to some
extent, on the design of the study and the nature of the data
collected. Two examples of general approaches are given in Brewen
& Gengozian (1971) and Evans & O'Riordan (1975).

3.2.1.5. Radiation-induced aberrations and estimates of exposure

For a variety of reasons, it is sensible to discuss the use of
the lymphocyte assay for estimating exposure to ionizing radiation
separately from cases of chemical exposure. In some ways, it was
fortunate, but, in retrospect, perhaps also unfortunate, that all
the original studies on the analysis of chromosomal aberrations in
the lymphocytes of exposed or possibly-exposed persons applied to
radiation exposure. The fact that, in these cases, estimates of
exposure could be made with some accuracy has often led to the
assumption that similar exposure-estimating procedures could be
applied to persons, groups, or populations exposed environmentally
or occupationally to chemical agents. A clear note of caution in
accepting this assumption is made here, and will be discussed
further.

Ionizing radiation and a small number of chemical agents (e.g.,
streptonigrin, bleomycin, neocarzinosatin, cytosine arabinoside,
and 8-methoxycaffeine) are able to produce aberrations in all
stages of the cell cycle including chromosome-type aberrations in
G₁ and chromatid-type aberrations in S and G₂.

The peripheral lymphocytes, a subpopulation of which is
mitogenically stimulated when blood samples are placed in culture,
are essentially non-cycling, and are in the G₀ stage of the cell
cycle, the term usually applied to non-cycling G₁ cells. Thus,
following radiation exposure, chromosome-type aberrations will be
induced in these non-cycling G₁ lymphocytes, and can be observed at
the first metaphase following mitogenic stimulation in culture.
The fact that aberrations can be produced in G₁ cells means that
their frequency can be directly related to the dose received (i.e.,
aberration frequency is proportional to dose). In addition, it has
been shown for many species (Brewen & Gengozian, 1971; Preston et
al., 1972; Clemenger & Scott, 1973), including man (Brewen et al.,
1972), that the frequency of aberrations induced in vitro is the
same as the frequency induced by the same dose delivered in vivo.
This means that a standard dose-response curve, for any particular
radiation type, can be obtained for in vitro exposures, and can
then be used for estimating doses received by individuals as a
result of radiation accidents during medical or other environmental
exposures. The frequency of aberrations (usually dicentrics) is
measured in cultured blood samples from exposed individuals, and
then converted into a dose estimate from the standard dose-response
curve. There is extensive literature on the use of the lymphocyte
assay as a biological measurement of dose for many different types
of radiation (Bender & Gooch, 1966; Lloyd et al., 1983). However,
there are still some limiting factors in the use of the assay.
These include the length of time of blood sampling after exposure,
partial-body exposure, and non-homogeneous exposures.

It is also apparent that the lymphocyte assay can be used for
estimating doses following long-term radiation exposure (Evans et
al., 1979). Again, this is possible because the aberrations are
induced in G₀ cells, and their frequency will be directly related
to exposure. Cells containing radiation-induced aberrations as
well as normal cells, will gradually be lost from the peripheral
lymphocyte pool, as a result of lymphocyte turnover. However, the
cells repopulating the peripheral pool will be derived mainly from
normal precursor cells, and not from those containing aberrations.
Many chromosome-type aberrations are lethal to the cell as the
result of loss of acentric fragments at division, or because of the
mechanical interference of the aberration with division. Thus,
during long-term exposures (months or years), the aberration
frequency will not be additive with time (or dose), but will reach
an equilibrium where new aberrations are formed and existing ones
are lost from the sampled population.

A similar argument holds for determining the length of time
after short-term exposure when samples can provide a measure of
the maximum aberration frequency or true induced frequency.
The lymphocyte turnover will result in loss of cells from the
peripheral pool, both normal and abnormal, and repopulation largely
by normal precursor cells. The aberration frequency appears to
stay constant for about 6 weeks after exposure (Brewen et al.,
1972).

Samples have been taken many years after exposure, for
example, with radiation-treated ankylosing spondylitis patients
(Buckton et al., 1978) and human beings exposed to the atomic bombs

in Hiroshima and Nagasaki (Awa et al., 1978). The fact that some
lymphocytes are very long-lived, in excess of 20 years (Awa et
al., 1978; Buckton et al., 1978), means that radiation-induced
aberrations can still be observed in cells that were present as
peripheral lymphocytes at the time of exposure. If it is assumed
that the aberration frequency declines exponentially with time, an
approximate estimate of the dose received can be made; the word
approximate is emphasized.

3.2.1.6. Chemically-induced aberrations and estimates of exposure

On the basis of the success of the lymphocyte assay for
estimating radiation exposures, it seemed appropriate to analyse
chromosome aberrations in blood samples from individuals,
occupationally exposed to single chemicals or complex mixtures
(Office of Technology Assessment, US Congress, 1983), to determine
whether there had been exposure to a clastogen, and to relate
aberration frequencies to dose level. However, the ability to
estimate dose levels from aberrations in populations exposed to
chemical mutagens is not reliable at present. More research is
necessary to make the assay more reliable and applicable, and to
make the results more readily interpretable.

The present low sensitivity of the assay for measuring the
frequency of chromosome aberrations following chemical exposure is
due to the mechanisms of induction of aberrations by chemical
agents. These mechanisms may also result in induced aberration
frequencies being only indirectly related to exposure dose, in
contrast to radiation-induced aberrations.

Chromosome aberrations induced by chemical treatments are
almost exclusively produced during the S-phase, irrespective of the
cell-cycle stage treated. Thus, the majority of aberrations will
be of the chromatid-type, although there are exceptions to this
general hypothesis (Evans & Vijayalaxmi, 1980; Preston & Gooch,
1981). Therefore, it is more appropriate to consider that the
probability of producing aberrations in G₁ or G₂ cells following
chemical treatments is low, but is considerably increased when
treated cells are in, or pass through, the S-phase.

Chemically-induced DNA damage in non-cycling lymphocytes will
not generally be converted into aberrations until the cells are
stimulated to re-enter the cell cycle in vitro, and undergo DNA
replication. Since repair of DNA damage can take place in G₀ cells
as well as during the long first in vitro G₁ stage, the aberration
frequency will not necessarily be proportional to the amount of
induced DNA damage but rather to the amount of DNA damage remaining
at the time of DNA replication. It is clear that the amount of DNA
damage, present at the time of replication, which has the potential
to be converted into aberrations, will depend on several factors
including: (a) the dose received; (b) the induced amount of the
particular types of DNA damage that can result in aberrations (a
value that can vary with the agent); (c) the amount of DNA repair
in G₀ cells before sampling (i.e., time between exposure and
sampling); and (d) the amount of repair in G₁ cells from mitogenic
stimulation to the first in vitro S-phase. Many of these will also

be subject to individual variation. The outcome is that, for most
chemical agents, only a proportion of induced DNA damage can be
converted into aberrations, at the time of replication. These DNA
repair factors all work to reduce the sensitivity of the lymphocyte
assay for measuring chemical exposure, and will result in the
aberration frequency being, at best, only indirectly proportional
to exposure.

If an increase in chromosome aberrations is observed in a
potentially exposed group when compared with a matched control
group, it is reasonable to conclude that there has been exposure to
a clastogenic agent, but no estimate of exposure level or of
subsequent adverse health effects (genetic or somatic) can be made.
On the other hand, if there is no difference in aberration
frequency between the possibly-exposed group and a matched control
group, it is not possible to rule out an exposure. However, in
this case, it may be possible, on the basis of previous experience,
to deduce a maximum exposure level that is not associated with a
detectable difference in aberration frequency.

3.2.1.7. Conclusions

Future research should enhance the sensitivity of the assay
system. In the meantime, it is suggested that care be taken in
the choice of groups to be considered for monitoring studies, with
particular regard to the usefulness of the information that might
be obtained, and to the likelihood of being able to draw definitive
conclusions. Cytogenetic monitoring can be an appropriate method
for the detection of chemical exposure, but, clearly, it cannot, at
present, be considered universally decisive. It should be used in
conjunction with other end-points.

3.2.2. Sister-chromatid exchange

3.2.2.1. Formation of sister-chromatid exchange

SCE results from the breakage and rejoining of DNA at
apparently homologous sites on the 2 chromatids of a single
chromosome. They were first visualized by Taylor et al. (1957)
who used tritiated thymidine to differentially label the DNA of
replicating cells and autoradiography to distinguish the silver
grain pattern on the 2 sister chromatids. Advances in
cytochemistry over the subsequent 15 years led to greatly
simplified procedures for visualizing these exchanges. Latt
(1973) demonstrated that bromodeoxy-uridine (BrdU), an analogue
of thymidine, when incorporated into DNA, could quench the
fluorescence of the fluorochrome Hoechst 33258. Perry & Wolff
(1974) found that incorporated BrdU also diminished the uptake
of Giemsa stain into the chromatin. In principle, SCEs can be
observed in any cell that has completed two, or the first of two,
replication cycles in the presence of BrdU. The most common
procedure for human lymphocytes is to grow the cells in BrdU for
two replication cycles.

SCEs are most efficiently induced by substances that form
covalent adducts with the DNA, or interfere with DNA precursor
metabolism or repair (Perry & Evans, 1975; Wolff, 1977; Perry,
1980; Carrano & Thompson, 1981; Latt, 1981). It is well documented
that SCEs are produced during DNA replication (Wolff et al., 1974)
and that the polarity of DNA is maintained in the process of
exchange (Taylor, 1958; Wolff & Perry, 1975). The molecular
mechanism by which the exchange is formed is not known but, since
there is an absolute requirement for DNA synthesis, hypotheses have
implicated the mechanics of this process in SCE formation (Bender
et al., 1974; Painter, 1980; Cleaver, 1981, Shafer, 1982).
Baseline SCE frequencies are increased in some human diseases
such as Bloom's syndrome (Chaganti et al., 1974) and multiple
sclerosis (Sutherland et al., 1980; Vijayalaxmi et al., 1983a).
Cells from patients with the inherited disease Xeroderma
pigmentosum are hypersensitive to the induction of SCEs by
ultraviolet radiation and alkylating agents (Wolff et al., 1975,
Wolff, 1977). There is no clear relation, however, between the
repair capacity of these cells and their sensitivity to SCE
induction (de Weerd-Kastelein et al., 1977).

3.2.2.2. Relevance of sister-chromatid exchange

An SCE represents the breakage of 4 strands of DNA (2 double
helices), a switch of the strands from one to the other arm of the
same chromosome, and the rejoining of the strands in their new
locations. The question is whether the breakage and rejoining
events occur without producing any modification in the genetic
code. There is evidence suggesting that this exchange process
might not always be error-free. In particular, a considerable
amount of information has been gathered concerning the relationship
between induced SCEs and other genetic effects in vitro and in
vivo.

Because the technique for enumerating SCEs is relatively
simple, this assay has been used quite extensively in genetic
toxicology studies. It has been amply demonstrated that the
frequency of SCE is dramatically increased when cells, or
animals, including human beings, are exposed to known mutagens
and carcinogens (Perry & Evans, 1975; Latt et al., 1981; Lambert
et al., 1982). Moreover, in early experiments using Chinese
hamster cells in culture, it was found that the induction of SCEs
was linearly related to the increase in single-gene mutations (HPRT
locus), when the cells were exposed to several substances each of
which differed in the type of lesion produced in DNA (Carrano et
al., 1978; Carrano & Thompson, 1982). The ratio of induced SCE to
induced mutation differed for each of the agents, suggesting that
each type of lesion formed is processed differently by the cell
favouring either the formation of SCEs or mutations. It is
possible that the same lesion is capable of producing both an SCE
and a mutation and/or that the lesions that form mutations are a
subset of those that form SCEs. The induction of mutations has
been shown recently to be correlated with the induction of SCEs in
mice for one chemical, ethyl nitrosourea (Jones et al., in press).
In this study, the induction of SCEs in mouse splenocytes,
following intraperitoneal administration of the chemical, was

linearly proportional to the increase in mutations at the HPRT
locus in the same splenocyte population, suggesting that the
mutation-SCE relation is not unique to in vitro studies.

Other investigations compared the induction of SCEs to either
the induction of transformation in vitro or induction of tumours
in vivo. In a series of experiments, the induction of SCEs was
shown to be linearly and positively correlated with the induction
of transformation of Syrian hamster embryo cells in vitro (Popescu
et al., 1981). This relationship held for exposure to 5 chemicals,
but not for X-irradiation. The ratio of induced SCEs to induced
transformation frequency varied for each chemical. This is
analogous with the results found for the SCE-to-mutation ratio
described above. Cheng et al. (1981) examined the induction of
SCEs in mouse bone marrow, regenerating liver, and alveolar
macrophages following intraperitoneal injection of ethyl carbamate
and related chemicals. The relative potencies in inducing SCE
parallelled the reported activities for the induction of lung
adenomas in mice. For ethyl carbamate, the doses for the
significant induction of SCEs in alveolar macrophages and lung
adenomas were similar. Further, the slopes of the dose responses
showed that the two end-points were of comparable sensitivity.

The results of early studies on rabbits demonstrated that it
was possible to measure an increase in SCEs in peripheral blood
lymphocytes for many months following repeated low-level exposure
to mitomycin C in vivo (Stetka et al., 1978). Increased SCEs in
human beings have been observed in cigarette smokers (Lambert
et al., 1978; Carrano, 1982) and in individuals undergoing
chemotherapy (Lambert et al., 1978; Musilova et al., 1979; Gillian
Clare et al., 1982). The results of these studies suggest that a
subpopulation of lymphocytes with high SCE frequencies may be
present for long periods after termination of exposure. These
cells may represent persistent damage in long-lived lymphocytes
or a sensitive subpopulation (Carrano & Moore, 1982). If this
is true, they may offer a more appropriate approach to the
quantification of persistent damage in vivo.

3.2.2.3. Factors potentially influencing SCE frequency

The baseline SCE frequency in human peripheral lymphocytes
averages about 7 - 10 per cell, but has been reported to range from
about 2 to 45 per cell in unexposed individuals (Alhadeff & Cohen,
1976; Lambert et al., 1976; Crossen et al., 1977; Galloway, 1977;
Carrano et al., 1980). Even within the same laboratory, the
baseline frequency may vary by a factor of two or more. Several
potential sources of variation have been identified and they
generally fall into 2 categories: a) culture factors associated
with the in vitro growth of the lymphocytes; and b) biological
factors associated with the genotype, lifestyle, or general health
of the individual. The biological factors usually cannot be
mitigated and must be accounted for in the selection of appropriate
control cohorts or ultimately in the statistical analysis.

(a) Culture-associated factors

A major source of variation can be attributed to the
concentration of BrdU relative to the number of lymphocytes in the
culture. The frequency of SCE has been shown to increase by as
much as 50%, when the BrdU concentration is raised 10-fold (Carrano
et al., 1980). The ability of BrdU to induce SCE has been well
documented in human and other cell systems (Wolff & Perry, 1974;
Lambert et al., 1976; Latt & Juergens, 1977; Mazrimas & Stetka,
1978) and it has been further shown that the SCE frequency can
increase if the BrdU concentration is held constant, but the
cell number decreased (Stetka & Carrano, 1977). The number of
lymphocytes that respond to mitogen and eventually incorporate
the analogue is generally not controllable, and, therefore, the
effect of BrdU can best be minimized by standardizing both the
concentration of BrdU and white blood cells at culture initiation.
Since the SCE increase above baseline following most long-term
exposures may be small, careful consideration of the influence of
BrdU is warranted.

Other culture-associated factors have been examined to
determine whether they influence the SCE frequency. One concern
has been the optimal time to harvest cells. In at least three
studies, it has been concluded that the baseline SCE frequency
does not depend on the time of harvest. Carrano et al. (1980)
found that, for any individual, the SCE frequency could vary by
as much as 30% as a function of culture time, but that the pattern
was not consistent among donors. Similarly, Beek & Obe (1979),
Becher et al. (1979), and Morimoto & Wolff (1980) did not observe
any significant alteration in the baseline SCE frequency as a
function of culture time. In a study by Snope & Rary (1979), for
4 individuals, the baseline SCE frequency observed in 58-h cultures
was about 30% less than in 70-h cultures. The authors interpreted
this to indicate that there were at least 2 subpopulations of
lymphocytes that differed in SCE frequency. Studies by Lindblad &
Lambert (1981) and Das & Sharma (1984) also revealed that the more
slowly cycling lymphocytes had a higher SCE frequency. The
possibility that this effect might be due to T- and B-lymphocyte
subpopulations has not been completely resolved. Santesson et al.
(1979) found that B-lymphocytes had a significantly lower SCE
frequency than T-lymphocytes, but Lindblad & Lambert (1981) could
not confirm this. In order to optimize an experimental design to
account for such an effect, all samples should, where possible, be
harvested at the same time.

Two other factors have been shown to influence the baseline SCE
frequency, i.e., the serum used and the culture temperature. Using
sera from 4 different sources, McFee & Sherrill (1981) showed that
the SCE frequency could increase by 50% as the serum concentration
increased. Kato & Sandberg (1977) also demonstrated an effect of
sera on the production of SCEs. The frequency of SCE in amphibian
cells increased almost 5-fold with increasing culture temperature
(Speit, 1980). In Chinese hamster cells, both lower (33 °C) and
higher (40 °C) temperatures increased the frequency of SCE by
approximately 60%. Because the alteration in temperature was
only effective in altering SCE frequencies while BrdU was being

incorporated, the author concluded that SCEs might arise via a
temperature-dependent disturbance of DNA replication. In a study
by Das & Sharma (1984), the frequency of SCEs in human lymphocytes
increased as a function of culture temperature and reached a
maximum at 40 °C. The authors concluded that temperature-dependent
DNA replication enzymes might be responsible for this effect.

In many human population studies, the cohort to be sampled
may be far away from the laboratory performing the cytogenetic
analysis, and it will often be necessary to ship the samples.
If blood is shipped commercially, recording thermometers in the
shipping container will provide some quality control. The effect
of blood storage on SCE frequency has been examined (Carrano et
al., 1980). No consistent alteration in SCE frequency was observed
for blood held at either 4° or 22 °C for up to 48 h. This was
confirmed by Sharma & Das (1984) for blood stored at 5 °C for up to
7 days, at 22 °C for 3 days, and at 37 °C for 2 days. However, it
should be noted that such storage can lead to an overall decrease
in lymphocyte numbers and modification of lymphocyte subclasses
(Dzik & Neckers, 1983). Thus, the samples should be placed in
culture as quickly as possible.

Finally, Deknudt & Kamra (1983) evaluated 4 different
mitogens for their effects on SCE frequency. They found that
PHA, concanavalin A (Con A), Wistaria floribunda (WFA), and lentil
lectin (LcH-A) extracts did not result in different baseline SCE
frequencies in the same donors. However, significant differences
among mitogens were noted for lymphocytes that had been exposed
to cyclophosphamide or mitomycin C. The use of a single T-cell
mitogen, such as PHA, for all studies is therefore recommended.
The experience accumulated from the use of lymphocyte cultures
suggests that there are many culture-associated factors that could
influence the SCE frequency. In order to minimize variation and
maximize sensitivity, the use of standardized procedures is
essential.

(b) Biological factors

Factors, unique to the individual being sampled, include such
confounding variables as sex, age, diet, genotype, medication, and
smoking. Each of these potentially plays a role in the induction
or expression of SCEs. Although some information is available on
the influence of these factors, the consequences are not always
clear-cut. Moreover, it is likely that they represent only a small
number of possible confounding factors.

Several investigators have compared the baseline frequency of
SCEs in male versus female donors and have not observed significant
differences (Galloway & Evans, 1975; Alhadeff & Cohen, 1976;
Crossen et al., 1977; Latt & Juergens, 1977; Morgan & Crossen,
1977; Cheng et al., 1979; de Arce, 1981; Waksvik et al., 1981;
Carrano & Moore, 1982; Livingston et al., 1983). The absence
of clear documentation of a sex difference for baseline SCE
frequencies suggests that this factor may not have to be matched in
a population study. However, that there is evidence for increased
SCE frequencies in women taking oral contraceptives (see below).

No studies relating SCEs to cyclic hormonal variation in the female
have been reported.

Another variable of concern in population studies is the age of
the donor, since it is not always possible to obtain accurate age
matching. The independence of age and baseline SCE frequency in
lymphocytes in adults has been observed (Galloway & Evans, 1975;
Morgan & Crossen, 1977; Hollander et al., 1978; Cheng et al., 1979;
Lambert & Lindblad, 1980; Carrano & Moore, 1982; Livingston et al.,
1983). Lower SCE frequencies have been found in infants (Seshadri
et al., 1982), in cord blood (Ardito et al., 1980), and in children
with a mean age of 1.5 years (Husgafvel-Pursiainen et al., 1980).
De Arce (1981) reported that people in the age range of 30 - 40
years had a higher number of SCEs per cell than those between the
ages of 0 - 10 or 60 - 70 years. Schmidt & Sanger (1981) found a
significant increase in SCE with age when comparing age groups of
1 - 18 months, 10 - 13 years, 26 - 32 years, and 63 - 85 years.
However, this might be due to differential BrdU incorporation
between cultures. With the exception of lower SCE frequencies in
new-born infants, the weight of evidence suggests that baseline SCE
frequencies do not change with age (at least up to 60 years).

It has been suggested that genetic factors play a role in the
baseline SCE frequency. When Cohen et al. (1982) compared SCE
frequencies, both within and among 12 pedigrees (2 generations),
they found significant differences between, but not within, the
families, suggesting a genetic contribution to baseline SCEs.
Pedersen et al. (1979) examined the cultured lymphocytes from 11
monozygotic and 9 dizygotic twins of the same sex. They did not
find any significant differences in the SCE frequency in
monozygotic versus dizygotic twins and concluded that genetic
factors do not play a role in baseline SCE frequency. Waksvik et
al. (1981) also did not find any genetic contribution to SCE
frequencies in their studies on twins. In order to determine
whether baseline SCE frequencies were related to ethnic origin,
SCEs were measured in adults of both sexes with no interracial
family backgrounds from Caucasian, American black, oriental, and
native American races (Butler, 1981). There was no significant
difference in the average frequency of SCEs in the 4 races.
Further research is needed to resolve the issue of genetic factors.

The potential influence of other factors on the frequencies of
human SCEs are worth considering. Because a large number of women
use oral contraceptives, care must be taken that this does not
confound the consequences of exposure to other agents. A study of
oral contraceptive users demonstrated a significant increase in
lymphocyte SCE associated with the daily intake of d-norgestrel or
ethenyl estradiol over a period of 6 - 24 months (Balakrishna
Murthy & Prema, 1979). Additional studies are needed to confirm
and extend these findings.

Despite knowledge of natural mutagens in food and the role of
food processing in creating mutagens (Nagao & Sugimura, 1981; Ames,
1983), little attention has been given to the role of diet in
causing baseline SCE variation. It has been reported that severe
protein calorie malnutrition in children is associated with

increased lymphocyte SCE frequencies (Balakrishna Murthy et al.,
1980). Following nutritional rehabilitation, SCE frequencies were
seen to decrease. Increased SCEs have also been reported in
chronic alcoholics (Butler & Sanger, 1981). For most human
studies, however, the nutrition factors are more likely to involve
long-term, low-level intake of caffeine, saccharin, other
additives, or well-cooked foods. Each of these factors has been
shown to be genotoxic in microbial, mammalian in vitro, or animal
assays, but there are no data on how these might influence SCE
frequency in human beings. Dietary information should be collected
in the initial questionnaire. Such information may be useful in
interpreting a population difference or can be used in large
composite data bases to search for diet-related factors. The use
of prescription and non-prescription medication, or vitamins,
should also be identified and appropriate information should be
collected before blood sampling, so that matched controls can be
employed.

Cigarette smoking is, perhaps, the most important confounding
factor in the interpretation of human SCE frequencies. The
preponderance of evidence indicates that individuals who smoke
have elevated SCE frequencies (Lambert et al., 1978; Ardito et al.,
1980; Hopkin & Evans, 1980; Lambert & Lindblad, 1980; Carrano,
1982; Husum et al., 1982; Livingston et al., 1983). However, this
effect has not been observed in all studies (Hollander et al.,
1978; Hedner et al., 1983). Since the elevation can be so marked,
it is important to control for cigarette smoking in any population
study.

Taken together, the culture and biological factors that
potentially confound the baseline SCE frequency in human beings
are numerous and contribute to individual variation.

3.2.2.4. Methods for sister-chromatid exchange analysis

The procedures for SCE analysis in human populations will
depend, to some extent, on the specific circumstances of the study.
The following generalized outline of a procedure may have to be
modified to fit special situations.

In order to identify any of the factors that may confound the
analysis of SCEs, it is important to obtain relevant information on
the background of each individual in the study population. This
information is best obtained through a questionnaire. The
questionnaire should attempt to determine the obvious exposure and
potential confounding factors. In addition to the questions
related to personal and family cancer incidence, birth defects,
known genetic disorders, and current health, there are specific
questions related to the interpretation of the SCE results. These
include: a) smoking history; b) medication and drug use; c) known
or suspected exposures to physical or chemical agents occurring
at work, at home, or during the pursuit of various hobbies or
recreation; and d) nutrition, particularly the adherence to special
diets or the use of artificial sweeteners or caffeine. Because a
complete list of agents that may directly or indirectly affect SCE
frequencies has not been established, it is better to collect too
much information than too little.

When the blood is collected, it should be mixed with heparin,
acid-citrate-dextrose, or another appropriate anticoagulant. As
already stated, the cultures should be established as soon as
possible after the blood is withdrawn, preferably within 24 h.
Just prior to establishing the culture, a total white blood count
and a slide for a differential should be made from the blood
sample. These parameters can serve as indicators of abnormally
high or low lymphocyte counts, factors that can have a bearing on
the observed frequency of SCEs.

It is important that a standard procedure be employed and,
although the exact procedure may differ among laboratories, there
are some common requirements. Because BrdU can induce SCEs in
human lymphocytes, both the BrdU concentration and the number
of lymphocytes potentially incorporating the drug, should be
stabilized. In order to ensure this, it is important that each
laboratory uses a fixed BrdU concentration and a consistent number
of lymphocytes per culture. Selection of culture medium, serum
type and concentration, whole blood versus purified lymphocyte
cultures, and duration of cultures is somewhat arbitrary and should
be based on experience. The serum lot should be pre-tested for its
potential to support lymphocyte growth and the reproducibility of
SCE frequencies in duplicate samples from the same donor. It is
recommended that the same serum lot be used for each population
study. The duration of culture should be such that it yields a
high proportion of second division cells so that the SCE frequency
is representative of a major fraction of the lymphocytes. Exposure
of the cultures to other than yellow or red light should be
avoided. Replicate cultures should be established for each
individual, and, when a large number of individuals is processed,
the inclusion of blood cultures from historical controls is useful.

At the appropriate culture time, the cells can be arrested in
mitosis with Colcemid^(R), colchicine, or other suitable agents.
Several slides should be prepared from each culture and the
remaining cell pellet can be stored in absolute methanol:glacial
acetic acid (3:1) at a temperature of 4 °C or lower, for future
recall. The slides can be stained for analysis by fluorescence
alone (Latt, 1973), fluorescence plus Giemsa (Perry & Wolff, 1974),
or a suitable alternative. Once a staining procedure has been
standardized and adopted, it should be used throughout the whole
course of the study.

Slides should be coded and scored blind. The coded slides can
be scanned under low magnification (100 - 200X) and selected for
scoring on the basis of good staining and chromosome number. Only
well-differentiated metaphases should be accepted for scoring. It
is not essential for all the chromosomes in a cell to be counted.
Overlapping chromosomes can be disregarded and the remaining
chromosomes scored. At least 43 chromosomes should be analysed per
cell. The SCEs are counted for each chromosome in the cell and
expressed as SCE per chromosome or converted to the number of SCEs
per diploid cell. Although the precise number of cells to be
scored per individual will ultimately depend on the desired
statistical sensitivity, a minimum of 50 cells should be scored,
with 25 cells from each replicate culture.

3.2.2.5. Data processing and presentation

The standard statistical descriptors such as mean SCE per cell
or per chromosome, standard deviation or standard error, and number
of cells scored should be determined for each individual. For
heteroploid cells, the SCE frequency should be normalized to a
diploid cell. Population means and their errors should be
presented similarly and information on the number of cells scored
per sample as well as the number of people per group should be
given.

Student's t-test is commonly used for the statistical analysis
of population data. When comparing population groups, however, the
sample size is the number of people in each group, not the number
of cells scored for each group. Caution is advised with the use of
t-tests. Individual and/or population means may not be normally
distributed and hence the t-test may be inappropriate. Analysis of
variance or appropriate non-parametric statistics may be more
rigorous for the particular population under study (Carrano &
Moore, 1982). Data transformation can be useful in the statistical
analysis, if the mean and the variance are independent (DuFrain &
Garrand, 1981). No single statistical procedure can be recommended
for universal application. The investigators must carefully
examine the data and apply the most appropriate test for the
questions being addressed.

3.2.2.6. Conclusions

There are many examples of populations exposed to mutagens in
which large increases in SCE have been demonstrated. These studies
show that the end-point can be a very sensitive indicator of
chemical exposure. In contrast, SCEs are not a useful indicator of
human radiation exposure. From the many reported studies, it is
evident that further quantitative information on variations in
SCE level between people, and the factors that influence these
frequencies, are required. For this reason, marginal increases
above a baseline SCE frequency must be interpreted with caution.
Significantly increased levels are an indication that the
population has been exposed to a genotoxic agent; on the other
hand, failure to observe increased levels does not necessarily
indicate the absence of exposure. Any increase in SCE frequency
for an individual or group cannot be interpreted as an indication
that the individual or group is likely to suffer adverse health
consequences.

3.3. Gene Mutations

New procedures are being developed to detect mutations that
lead to the production of abnormal haemoglobins (Bigbee et al.,
1981), the loss of cell surface proteins (Bigbee et al., 1983), or
the loss or modification of enzymes used in the synthesis of DNA in
human somatic cells. The last of these procedures has been used
for human population studies and is considered in detail.

Resistance to 6-thioguanine is a marker that has been used
widely to detect somatic mutations occurring in vitro in cultured

mammalian cells (Hsie et al., 1979). Since similiar somatic
mutations occur in vivo, quantification of 6-thioguanine resistant
(TG^r) cells in peripheral blood provides a measure of specific-
locus somatic mutations occurring in vivo.

In human beings, a naturally-occurring germinal mutation
of the X-chromosomal gene for hypoxanthine-guanine
phosphoribosyltransferase (HPRT) (EC 2.4.2.8) provides a prototype
for 6-thioguanine resistance at the somatic level (DeMars, 1971).
HPRT normally converts hypoxanthine and guanine to inosine
monophosphate, and it phosphorylates some purine analogues (e.g.,
6-thioguanine) to produce cytotoxic products (Caskey & Kruh, 1979).
Mutant cells, deficient in this enzyme, are unable to use exogenous
hypoxanthine as a purine source and are resistant to killing by
these purine analogues.

TG^r somatic cells arising in vivo can be shown to be mutants
by the stability of their phenotype and deficiency of HPRT activity
(Albertini, 1982; Albertini et al., 1982a; Morley et al., 1983b).

Also, molecular studies have demonstrated changes in the HPRT locus
at the DNA level (Albertini et al., in press).

3.3.1. Principles and basis for the methods

TG^r T-lymphocytes (T-lys), present in human blood, can be
measured either by autoradiography to determine TG^r T-ly variant
frequencies (Vfs), or by cloning to determine TG^r T-ly mutant
frequencies (Mfs). Autoradiography is the earlier method (Strauss
& Albertini, 1977, 1979) and has been used for some human
population studies (Strauss & Albertini, 1979; Strauss et al.,
1979; Cole et al., 1982; Lange & Pranter, 1982; Morley et al.,
1982; Vijayalaxmi et al., 1983b; Albertini, 1983; Albertini et al.,
1983a); it is simple, inexpensive, and can be automated (Anmeus et
al., 1982; Zetterberg et al., 1982). Cloning is a more recent
development by which it is possible to recover the TG^r cells for
characterization of the nature of the mutations (Albertini, 1982;
Albertini et al., 1982; Strauss, 1982; Morley et al., 1983a;
Vijayalaxmi & Evans, 1984). This characterization may be
necessary in order to quantify the different types of somatic
mutations responsible for a collection of mutants.

3.3.1.1. Autoradiographic method

The autoradiographic detection of rare TG^r T-lys is based on
the ability of these cells to incorporate tritiated thymidine
(HTdR) in vitro following short-term lectin stimulation in the
presence of cytotoxic concentrations of 6-thioguanine. The vast
majority of human peripheral blood T-lys are in the G₀ stage of the
cell cycle in vivo. When stimulated in vitro with lectins, such
as phytohaemagglutinin (PHA), T-lys are activated to early G₁,
which is characterized by the acquisition of T-cell growth factor
(TCGF) receptors on their surfaces (Maizel et al., 1981). These
receptors are also called interleukin-2 (IL-2) receptors or Tac
antigens (Deeper et al., 1983). A series of cellular events
results in the production of interleukin-1, (IL-1) and IL-2, and
other factors in short-term peripheral blood mononuclear cell
cultures, with IL-2 driving the activated T-lys through G₁ to DNA
synthesis and subsequent events in the cell cycle.

The initial round of lectin-stimulated T-ly DNA synthesis
in vitro is inhibited by 6-thioguanine. Inhibition probably is
at the activation step in 6-thioguanine sensitive cells and
occurs without incorporation of the analogue into DNA. The
interval of lectin stimulation before the addition of HTdR in
the autoradiographic assay is short, so that cell division in
vitro does not occur (Strauss & Albertini, 1979). Under proper
conditions, the T-lys that incorporate this label in the presence
of 6-thioguanine are TG^r variants. These cells can be enumerated
by autoradiography and the TG^r variant frequency (Vf) calculated.

3.3.1.2. Cloning method

Human T-lys are easily propagated for extended periods in vitro
by supplying TCGF to cells that have been activated with lectin or
antigen (Paul et al., 1981). Thus, mass populations derived from

peripheral blood can be developed and maintained. Moreover,
freshly-obtained human T-lys may be cloned directly in vitro, in
the presence or absence of 6-thioguanine. In the first case, large
cell innocula must be used, to ensure the presence of rare TG^r
cells (Albertini, 1982; Albertini et al., 1982a; Strauss, 1982;
Morley et al., 1983a,b; Vijayalaxmi & Evans, 1984). Activating
intervals are short so that cell division does not occur in vitro
prior to exposure to cytotoxic concentrations of 6-thioguanine,
thus ensuring that the TG^r cells recovered in vitro actually arose
in vivo. The number of TG^r T-lys determined by cloning is used to
derive in vivo TG^r T-ly mutant frequencies (Mfs).

3.3.2. Relevance and limitations

3.3.2.1. Relevance

The ability to measure TG^r T-lys arising in vivo