REGULATORY CONSIDERATIONS: MUTAGENESIS 729 My comments on those questions reflect current issues and important considerations, but some of my judgments will undoubtedly change as the data base improves. 1. SENSITIVITY The Ames test is extremely sensitive. It will detect the majority of chemical carcinogens presently identified as such in our current data base, except for hormones, metals, and carcinogens dependent on physicochemical effect, such as asbestos. So far there is a strong statistical association with known carcinogens (3, 4). The test system has been greatly modified to improve its sensitivity to known carcinogens, such as the activation systems and the modification of the bacterial cell wall. It must be recognized that the incidence of cancer in test animals or in man can be af- fected by many other factors that have not been (nor should be) identified as carcinogens. Some of these factors affect susceptibility to cancer, such as nutritional and metabolic factors important to the mammalian body's defense against toxicity, including cancer. Other modifying factors are more appropriately classified as tumor promoters. Examples are hormonally active substances which can stress target receptor cells they may lead directly to adenocarcinomas or to certain epithelial changes, more properly called metaplasia, in which glandular or transitional epithelium is transformed to squamous, and a squamous cell or undifferentiated type of tumor can form. These factors are associated with overall patterns of cancer induction, but are not necessarily involved in causation or mechanism latent viruses and their genomes may play a substantial part in the etiology of some cancers in humans. Thus reducing the number of mutacarcinogens in our environment will generally help to control cancer but will not solve the entire problem. The public cannot be protected against all or even most cancers by using a very sensitive test system to identify all "positive" compounds and then trying to reduce all human exposures to these positives to infinitesimal levels. Other Federal laws also affect regulatory decisions. A relevant statute known as the "Delaney Clause" prohibits the intentional addition to foods of a substance legally classified as a food additive if that substance induces cancer in animals or man or has been shown by appropriate tests to induce cancer. The question of whether the Ames test is an appropriate device to support the Delaney Clause should be considered in the context of statistical association with identified carcinogens. We should likewise ques- tion the relevance of in vivo long-term feeding experiments in rodents to the human situation. 2. SPECIFICITY So far, the test system has been fairly specific for the substances tested in that muta- genic effects have been detected for many carcinogens. Thus the present designation of true positives is acceptable, but the question of false positives has not yet been answered satisfactorily. Most of the substances tested so far have been carcinogens relatively few so-called noncarcinogens have yet been tested. Many more substances commonly accepted as safe for human exposure or consumption should be tested before a judgment is made. I did not say "commonly accepted as being noncarcino- genic," because, under our present in vivo bioassay system and current dosing regimens, some substances will register positive in an animal feeding experiment even though they are not carcinogens. Any factor--carcinogen, promoter, or modifier--can

730 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS register as positive in a feeding experiment if the right convergence of factors is present under the particular experimental circumstances. If the Ames test is less specific in actuality than some of its proponents now believe, a number of unexpected positives are likely to be identified in the near future. What do these positives mean? Are they more truly positives than bioassay positives, or are they less potent carcinogens for the intact mammal? And how should we extrapolate these results to humans? Some advocates propose that the Ames test should be used as a screening test and that all compounds found positive should then be tested by appropriate exposure to intact mammals in vivo. Others propose that compounds found positive by the Ames test should be withdrawn from use until they have undergone bioassay testing in mammals and have been demonstrated negative in two rodent species. Still others maintain that a compound which is Ames-positive and bioassay-negative merely proves that carcinogenic bioassay is insensitive either because of species-specific metabolism or the fi error problem inherent in using practical sample sizes of 50 male and 50 fe- male rodents per dose and controls. Obviously the term "screening test" can have many connotations, especially in regulatory judgments. What a manufacturer chooses to do after voluntarily testing for mutagenesis and find- ing a positive result is still largely a matter of free choice. He can try to eliminate the positive component from his product, he can shift to an alternative compound, or he can undertake lifetime bioassays which are expensive and whose outcome is un- predictable. But in a strict regulatory approach, the number of false positives must be kept to an acceptable level. Undoubtedly, false positives occur in bioassays that problem, however, was identified only after the carcinogenic bioassay system was ac- cepted by most of the scientific community. Even then, some scientists had reserva- tions about whether the bioassay protocols are a relevant model for human exposures that occur at substantially lower levels than those to which test animals are exposed. In summary, the question of how false positives relate to the specificity of the Ames test can be resolved only by expanding the number of chemicals tested however, it is essential not to select only those chemicals known to have the characteristics of carcinogens. The question of relation of false negatives to the specificity of the Ames test does not appear to be a significant problem. Many other criteria can be used to help determine whether a substance should be tested for carcinogenic activity. 3. QUANTITATIVE MEASUREMENT OF THE PARAMETERS OF CONCERN In one sense, the ability of the Ames test to measure the strength of biological activity of the parameter of concern (reverse mutations) over an exceedingly broad range ap- pears better than the ability of the intact animal assay to measure carcinogenesis. However, each test measures different--although probably related--parameters. The carcinogenesis bioassay measures the capability of a substance in a limited system to statistically affect the incidence of tumors in test animals. For true positives, it measures the ability of the substance to express its carcinogenic activity in the intact mammal. Thus penetration through defense mechanisms is involved, and the critical convergence of intrinsic capability (penetration to the target of opportunity and sur- vival in activatable form) has resulted in a carcinogenic hit. In my judgment, dosage must be considered, as well as many parameters of classical toxicology that show the ability of a chemical to penetrate to the target after having survived various defense