Comparison of carcinogen hazard characterisation based on animal studies and epidemiology

Recently we have described a simple method for quantitative risk assessment of non-threshold carcinogens based on the dose descriptor T25. In the present report quantitative hazard estimates calculated with the T25 method have been compared with results obtained using quantitative methods based on epidemiological studies. "Known" and "Likely/Probably" human carcinogens were identified from the US EPA database IRIS. In cases were the hazard characterisation was performed on the basis of epidemiological studies, the IARC monographs were used to identify animal studies by oral or inhalation exposure suitable for hazard characterisation by the T25 method. Six agents were identified: benzene, benzidine, 1,3-butadiene, cadmium, nickel subsulfide and vinyl chloride for which US EPA had made their hazard estimation based on epidemiological data. Animal data suitable for hazard characterisation were also available. For comparing hazard characterizations based on epidemiological and animal data, it was pragmatically decided to do this by comparing the chronic doses expressed as those representing a lifetime cancer hazard of 10(-3). In all cases the difference between the chronic doses determined from animal studies by the T25 method differed from those determined from epidemiological studies by a factor of less than three. Although a limited number of carcinogens were studied, the results demonstrate a very good agreement between the hazard characterisation based on epidemiological data and animal experiments over a range of more than 10(4).