Statistical Analysis of Heritable Mutagenesis Data

INTRODUCTION The objective of heritable mutagenesis studies is to evaluate some toxic chemical’s potential to induce genetic damage that leads to disorders in offspring of test subjects, i.e., transmissible genetic damage induced in the subjects’ germ cells (Maxwell and Newell, 1973; Russell and Kelly, 1982; Shelby, 1988; Shelby, et al., 1993). Typically, laboratory animals are employed as test subjects. For example, in a common design, male (or female) rodents are administered the agent of interest, mated with untreated partners, and their offspring are examined to determine whether or not chemically-related heritable genetic effects are observed. A number of different assays are available for studying such heritable damage, each of which centers its attention on a different type of genetic toxicity (Russell and Shelby, 1985). These assays include the dominant lethal test (Generoso, 1973; Generoso and Piegorsch, 1993; Green and Springer, 1973), the heritable translocation test (Bishop and Kodell, 1980; Léonard, 1975), and the specific locus test (Selby and Olson, 1981). In these assays, data are obtained on various stages of germ cell production. The data may involve embryonic or zygotic damage (Generoso, et al., 1991), effects on conceptal implantation, and/or fetal mortality. In all cases, however, conceptal damage is associated with some mutagenic event in the germ cells of the exposed parent (Shelby, et al., 1993). [This contrasts with the issue of in utero developmental damage and its associated statistical concerns, as seen, e.g., in developmental teratogenicity studies (Piegorsch and Haseman, 1991; Ryan, 1992).] There are several important statistical issues associated with the analysis of heritable mutagenesis data. These include: determination of the appropriate experimental unit, the selection of appropriate methodology for data analysis, and sample size/power considerations. Each of these issues will be discussed briefly below, using as a common paradigm the male dominant lethal assay in mice (Generoso and Piegorsch, 1993). For purposes of risk assessment, this assay is often conducted as a identifying screen, and then usually at only two treatment levels: a control and a single dose of the mutagenic stimulus. Thus, the discussion herein will center on two-group statistical comparisons for heritable mutagenesis data.