Cancer risk estimation of glycidol based on rodent carcinogenicity studies, a multiplicative risk model and in vivo dosimetry.

Here we evaluate a multiplicative (relative) risk model for improved cancer risk estimation of genotoxic compounds. According to this model, cancer risk is proportional to the background tumor incidence and to the internal dose of the genotoxic compound. Furthermore, the relative risk coefficient per internal dose is considered to be approximately the same across tumor sites, sex, and species. In the present study, we demonstrate that the relative risk model is valid for cancer risk estimation of glycidol, a common food contaminant. Published tumor data from glycidol carcinogenicity studies in mice and rats were evaluated in combination with internal dose estimates from hemoglobin adduct measurements in blood from mice and rats treated with glycidol in short-term studies. A good agreement between predicted and observed tumor incidence in responding sites was demonstrated in the animals, supporting a relative risk coefficient that is independent of tumor site, sex, and species. There was no significant difference between the risk coefficients for mice (5.1% per mMh) and rats (5.4% per mMh) when considering internal doses of glycidol. Altogether, this mechanism-based risk model gives a reliable risk coefficient, which then was extrapolated to humans considering internal dose, and background cancer incidence.