Comparisons of thresholds for carcinogenicity on linear and logarithmic dosage scales.

Comparisons on a linear and the Rozman logarithmic scale for dosage versus carcinogenicity in rodents are presented for methyl eugenol (ME), nitrosodiethylamine (NDEA), ethyl carbamate (EC) and 2-acetylaminofluorene (AAF). Each of these chemicals has been shown to be carcinogenic in experimental animals and, in addition, humans are regularly exposed to at least three of these compounds (ME, NDEA, EC) in foods. Although the source of adducts from AAF is not known, the aminofluorene (AF) adduct is present in humans. Plotted on the same graphs are either some doses from common foods (ME, NDEA, EC) or adducts (AF) on human haemoglobin, for perspective, with their thresholds for carcinogenesis in animals. Use of a linear scale when comparing doses administered to animals in studies of carcinogenicity with doses of those same chemicals to which humans are exposed does not provide useful, comparative information. On the other hand, the Rozman logarithmic scale for dose allows one to put these relative doses in perspective. It is also evident that forcing a linear extrapolation through the zero, zero origin does not agree with the experimental data. Further analyses for goodness of fit for these dose responses reveal that the dose response for three of these compounds (ME, NDEA, EC) appears to be linear with the logarithm of the dose. However, AAF appears to be linear with the logarithm of the dose for bladder, but not for liver. It is suggested that the high background incidence of tumours in the BALB/c StCrlfC3Hf/Nctr mouse liver may confound the interpretation of dose response from AAF carcinogenesis in mouse liver.