Some primary considerations in the interpretation of the dominant-lethal assay.

Among the various procedures proposed for use in assessing the mutagenic potential of drugs, the dominant-lethal (D-L) assay stands currently as one of the few tests for measuring mutagenic effects on germ cells. Early identification of the D-L assay as a possible member of a test battery relates strongly to its being a mammalian model. Many scientists within the pharmaceutical industry believe that only those tests which utilize a mammalian model should be considered for primary use in drug safety evaluation protocols. The reason for such belief is obvious when one considers that the entire process of drug safety evaluation is oriented on established concepts in pharmacology and toxicology. Mammalian pro-


Introduction
Among the various procedures proposed for use in assessing the mutagenic potential of drugs, the dominant-lethal (D-L) assay stands currently as one of the few tests for measuring mutagenic effects on germ cells. Early identification of the D-L assay as a possible member of a test battery relates strongly to its being a mammalian model. Many scientists within the pharmaceutical industry believe that only those tests which utilize a mammalian model should be considered for primary use in drug safety evaluation protocols. The reason for such belief is obvious when one considers that the entire process of drug safety evaluation is oriented on established concepts in pharmacology and toxicology. Mammalian processes of assimilation, absorption, distribution, metabolism and elimination must be permitted to work on the chemical under test in order to provide some basis for extrapolating mutagenicity test data to man (1). Dose levels tested in these models and routes of drug administration should both reflect human use. Also, differences in the qualitative pharmacologic action of drugs must be considered as an essential part of the criteria applied to dose selection. Test reproducibility and dose-effect relationships *Medical Research Laboratories, Pfizer Central Research, Groton, Connecticut 06340. must be emphasized in mutagenicity studies in order to identify those levels at which any mutagenic action is first detectable and to relate this level to the dose required for therapeutic efficacy.
In this presentation, data have been selected from a number of D-L studies which relate to these points and the difficulties encountered in interpreting D-L test results.

Methods
Random-bred CD-1 mice (Charles River), 8 weeks of age, were used in all experiments except where noted. Generally, 15 males were assigned to each test and control group, and 2 females were caged with each male. Pregnant females were identified by the presence of a mating plug. The number of total and dead implants/pregnant female were determined by autopsy at 12-14 days of pregnancy. Statistical analyses were computerized and all tests of significance were performed on arcsine transformed data. Weekly summations of test data were compared to a control regression computed across the entire 8 weeks of testing (2). These statistical models are discussed in the paper by Dr. David Salsburg (3).

Strain Characterization
Continuous surveillance of the mouse December 1973 Control data on 6820 pregnant females (CD-1 strain) are presented in Table 1. All data are expressed as a function of the week of mating following treatment of the male. The control males mated with these females had received physiological saline. This strain has consistently maintained an average level of total implants/pregnant female close to 12.50. The average number of dead implants/ pregnant female is 0.89 and the average of living implants/pregnant female is 11.61. When the number of dead implants is compared to the total implants an average value of 7.1% is obtained.
An example of a shift in the reproductive behavior of this strain is shown in Table 2. During the period of November 1, 1972 to March 1, 1973 the number of dead implants/pregnant female rose to a value of 1.02. This was accompanied by a reduction in the number of living implants/pregnant female to 11.43. Total implants/pregnant female was 12.38 and the percent dead implants/total implants was 8.2. Although such a shift may appear slight, this degree of fetal wastage can produce problems in the interpretation of test results and reduce the sensitivity of the test (4). The rapid rise observed in this period suggests the introduction of an infectious disease entity although no overt clinical disease was evident.
Occasionally, a genetically aberrant male is encountered which produces a D-L effect in several stages of spermatogenesis. Table  3 shows such a result with significant responses in weeks 1 through 7. The compound involved normally produces a D-L effect in weeks 5 and 6. Additional analyses revealed a single male had produced this response.

Test Reproducibility
A true mutagenic response in the D-L assay can be characterized by a statistically significant increase in dead implants/pregnant female accompanied by a statistically significant reduction in living implants/ pregnant female. Additionally, the compound involved should show a dose response relationship during a specific stage in the spermatogenic cycle. If a statistically significant response cannot be demonstrated reproducibly in the same stage of spermatogenesis, then a spurious positive result should be suspected. Table 4 demonstrates the typical response of the mutagen, ethyl methanesulfonate. In both experiments, the number of dead implants/pregnant female increases markedly during the first two weeks of mating. It should be noted that a significant decrease in the number of living implants per pregnant female occurs in the same two Environmental Health Perspectives weeks. A dose-response curve is shown in Figure 1 for the period 7-11 days following mating (5). Another example of a reproducible D-L effect is depicted in Table 5. The purine analog, 6-mercaptopurine has produced a consistent D-L effect during weeks 5 and 6 of the spermatogenic cycle (6). Again, the parameter of living implants/pregnant female showed a simultaneous and significant reduction.

Nonreproducible Results
In contrast to the reproducibility obtainedl  during the same stage of spermatogenesis weeks 2 and 4. It should be noted that the with a true mutagen, spurious or falseparameter of living implants/pregnant fepositive results do not repeat during the same male was not significantly reduced. stage of spermatogenesis. An example of this Tables 7-12 show the kind of inconsistenkind of results is shown in Table 6. The com-cies which may occur in the dominant-lethal pound produced effects on two separate assay with a nonmutagenic substance. The stages of spermatogenesis in the first two response at week 7 at a dose of 7.5 mg/kg experiments. A third experiment performed (Table 8) was not reproduced at a level of 75 at the same dose level was negative in both mg/kg (Table 10). Further, the response   C  T  C  T  C  T  C  T  C  T  C  T  1  26  33  319  2  43  38  535  3  39  43  460  4  34  52  435  5  41  34  498  6  28  24  360  7  44  28  563  8  48  37   December 1973 during week 3 at the dose level of 75 mg/kg is performed on a weekly basis and the parawas not reproduced at 150 mg/kg (Table 12). meters of total and living implants are ex-It should be noted that here again, no signifiamined simultaneously with dead implants/ cant reduction was observed in either the to-pregnant female. In addition, this compound tal or living implants/pregnant female at any was not active in any test for mutagenic podose level during the test. The investigator is tential including the host-mediated and in thus not mislead when the statistical analysis vivo cytogenetic assays.

Effect of Dose Level
The qualitative pharmacologic action of drugs must be considered when choosing dose levels for D-L experiments. Drugs such as anesthetics and tranquillizers have such pronounced pharmacologic activity that ex-cessive dose levels can produce marked temperature reductions and an inability to mate for several days following a single administration. An example of this kind of overdosage is shown in Figure 2 Environmental Health Perspectives observed at levels which were tested for mutagenic activity (7,8). Clearly, such reductions must reduce the overall metabolism of the test animal and therefore influence the metabolism of the drug. Levels of drug used in mutagenicity assessments should be chosen so as not to produce anorexia, sedation, or other exaggerated pharmacological effects (9).

Conclusions
In interpreting D-L data, the need for demonstrating a statistically significant and reproducible effect in the same stage of spermatogenesis cannot be over emphasized. In order to achieve consistent analyses, the degree of variability in important parameters of dead, living and total implants per pregnant female has to be firmly established for each strain of mouse employed. The statistical model utilized should include a transformation to reduce the effect of differing variances which occur in dead and total implants per pregnant female. Also, test results obtained during a specific stage of spermatogenesis should be compared to a control regression computed across the entire 8 weeks of testing. A dose response curve obtained during the active period of dominant lethality will provide additional evidence of compound activity. Data from D-L testing should be correlated and compared to other assessments of mutagenic potential such as the host-mediated and cytogenetic assays before applying the label of mutagen. Finally, the dosage regimen employed should not seriously alter the normal physiological processes of the test animal.