and colleagues (1966) mentioned above.

Because of the similarity of sulfur mustard and HN2 in producing pulmonary tumors in strain A mice, we would expect sulfur mustard to carry a potential ANL cancer risk. However, the magnitude of the effect may differ from HN2, because a different fraction of sulfur mustard inhaled would be transported to the bone marrow.

Dedrick and Morrison (1992) have also compared the carcinogenic potency of chlorambucil, cyclophosphamide, and melphalan in humans and rodents, but based on the integrated exposure of the ultimate carcinogenic chemicals. This procedure changes the relative positions of the chemicals, but not the correlation between human and rodent data. Thus, comparisons between effects data derived from external exposure are considered relevant, as pharmacokinetic effects of these compounds appear to be similar in humans and rodents.

Watson and colleagues (1989) have estimated the carcinogenic potency of sulfur mustard by the "rapid screening of hazard" (RASH) method. This approach compares exposures that produce documented toxic effects from an agent of interest to exposures of a reference chemical producing a similar effect (Jones et al., 1985, 1988). The results of these relative potency analyses are usually similar to those of the more traditional CAG (Carcinogen Assessment Group) and NTP (National Toxicology Program) carcinogenicity assessments in establishing exposure standards used by EPA, OSHA, and other regulatory agencies (Glass et al., 1991; Jones and Easterly, 1991; Jones et al., 1988; Owen and Jones, 1990). Watson and colleagues (1989) applied the RASH procedures to Heston's intravenous injection studies (1950) and subcutaneous injection experiments (1953a). Other available Heston data were not incorporated into the analysis because of Heston's incomplete characterization of exposure concentrations or high animal mortality induced by the experimental protocol. The experiments of McNamara and colleagues (1975) would have been desirable to use, but comparison exposures for primary or secondary standards were not available.

By considering all possible combinations of experiments and several reference compounds, sulfur mustard tumorigenicity was determined to be comparable with nitrogen mustard (HN2 and HN2-HCl) tumorigenicity in laboratory rodents. In the analysis of nitrogen mustard, data from the studies of Abell and colleagues (1965), Boyland and Horning (1949), Conklin and colleagues (1965), Heston (1949, 1950, 1953a), Schmähl and Osswald (1970), Shimkin and colleagues (1966), and Zackheim and Smuckler (1980) were used. Additional relative potency comparisons were made for the therapeutic nitrogen mustards melpha-