The current state of mutagenicity testing 209 and germ cells. Systems from micro-organisms, plants, insects, mammalian and human cells in vivo and in vitro have been used, and whilst the mutagenic effect of a chemical may be detectable in one system it may not be in another or even in different organs of the same system. There may be many interactions between a chemical and an organism, all of which may determine whether genetic damage is expressed- unlike ionising radiation, which is immediately active in terms of its mutagenic potential. With radiation extremely short-lived free radicals are generated randomly in cells and tissues very close to or within the genetic target molecules. There are many factors that determine whether a chemical compound reaches and reacts with the critical genetic targets. Among these factors in mammals are chemical structure of the compound, duration of treatment, route of administration, absorption, distribution, excretion, drug-protein binding, metabolic transformation, pharmacogenetic make-up (species, strain, sex) membrane barriers, numbers of SH groups, etc (19). The following examples will illustrate such factors: Vinyl chloride, a known carcinogen (20), requires metabolic activation to exert its effect (21). Vinyl chloride exposure causes mutations in micro-organisms (22) and in some test systems involving higher organisms. It causes chromosome damage in human peripheral lymphocytes (23), but no effect is seen in the dominant lethal assay in mice (24). This suggests that various tissues metabolise the substance to differing extents and maybe the testes metabolise the substance least. This is certainly the case with dimethylnitrosamine, another indirect mutagen, where there is less alkylation in the testes than in other organs (25). This compound also gives a negative result in the dominant lethal assay (26), and in a host-mediated assay with yeast cells, mutagenic activity is high in the liver, moderate in the lungs, but only very weak in the testes (27, 28). Other compounds such as methylnitrosoguanidine, a direct mutagen, are metabolised rapidly in mammals to non-mutagenic activity and 4-nitroquinoline-l-oxide has high mutagenic activity in yeast cells recovered from the lungs and slight activity in the liver but is not active in the testes (28). Another possibility is that such compounds may be detoxified by organs where little mutagenic activity is detected. Other factors determine whether the damage in the target molecules is expressed as genetic damage. Such factors are the innate susceptibility of the cell (e.g. repair capacity), numbers of susceptible cells, type of genetic target of the genetic target locus, germinal and somatic selection processes involved, mode of inheritance etc (19). It has been found with ethyl methanesulphonate (EMS) for example, a compound which can alkylate all tissues directly (29), that at doses below 100 mg/kg a dominant lethal effect is not observed (30) but at higher doses an effect is clearly manifest in CD-1 mice at 150 mg/kg body weight (31), confirmed by Anderson et al. (32, 24, 33). Sperm proteins and sperm DNA are alkylated by EMS at doses as low as 3 mg/kg body weight (34). Such discrepancies may occur because the numbers of animals screened in a dominant lethal test may be statistically too small to detect a weak effect at low doses, or unknown factors may affect the expression of damage in germ-cell development preventing the production of muta- tions (19). The factors described above as influencing a mutagenic response complicate muta- genicity testing. Many compounds that would require testing will not behave in the same way as model mutagens, which bind covalently with biologically important molecules. Many will react reversibly with receptors by mechanisms such as hydrogen bonding and electrostatic binding. Such compounds may not react with the genetic material unless they become metabolised to highly reactive forms. Other factors such as temperature, ageing, pH, hypoxia and viral infections can also produce genetic damage (19). Chemicals

210 Diana Anderson may, therefore, disturb normal bodily functions producing a change in one of these parameters to cause a genetic effect indirectly. TEST METHODS AND RECOMMENDATIONS FOR THEIR USE Geneticists have a large number of experimental methods at their disposal to assess the mutagenic action of chemicals. No single method, however, gives any conclusive in- formation about the genetic risk to a person who has been exposed to a mutagenic substance. Thus it would seem necessary for a number of test systems to be used. Bochkov et al. (17) have listed the effects, and given a useful pr6cis of the results of different types of chemicals such as cytostatics, antibiotics, pharmaceutical preparations, food additives and admixtures, chemosterilants. fungicities, insecticides, herbicides and industrial chemicals in the different test systems of micro-organisms, plants, fruit fly (Drosophila), mammalian cell cultures, bone marrow and dominant lethal systems of mammals and human cells in cultures and human lymphocytes. Various recommendations for combinations of test systems have been put forward (35-46, 17). All approaches have their advantages and disadvantages. In the more recent reports Bridges (43) proposed that assessment at each stage of testing should take into account the population exposed to the substance, its economic value and the possibility of substituting a non-mutagenic grouping for the mutagenic grouping in the structure of a substance. However, non-mutagenic substitutions are not always possible without altering the desired biological activity of a product. The testing scheme proposed by Flamm (45) includes tests for the detection of heritable translocations and specific locus mutations. Both of these tests require large numbers of animals, are not always economic- ally feasible, and in any case do not necessarily generate sufficient information to assess the genetic risk to man. Various governmental agencies, such as those of America, Japan, Britain and other EEC countries, are now preparing guidelines for testing methods. Some of the more recent ideas indicate the need for flexibility. This should permit changing to new and improved methods as soon as their usefulness has been substantiated. However, there is anxiety that any significant deviation from generally accepted guidelines may be questioned by regulatory agencies and that the guidelines that will satisfy the most demanding health authority may become the generally accepted procedure. With all the recommendations in the communications already in the literature (listed above), those put forward in this present communication will be brief, and it must be remembered that testing organisations, if allowed flexibility, will have their own ideas regarding scientific protocols, etc. The problem with the present testing methods is that not all systems are equally reliable and reproducible. Even the systems in which most compounds have been tested, such as the Salmonella typhimurium plate incorporation mutagenicity assay with meta- bolic activation (5), give different results for some compounds in different laboratories (47). Ideally, a system for assessing chemicals for mutagenic activity should: (a) detect the potential of the test substance to induce gene and chromosome mutations in somatic and germ cells, (b) provide quantitative data for extrapolation to man, (c) be capable of detecting metabolic products of the compound which have mutagenic potential,