Critique of dose response in carcinogenesis.

A few landmarks in the development of dose response in toxicology are presented, with an explanation of why dose should only be considered on a logarithmic scale. Examples are shown, illustrating that the current practice of labeling dose-response curves for carcinogenesis as supralinear, linear or sublinear, is meaningless unless the dose-response scales are defined. Since many reports labeling such curves as supralinear, linear, or sublinear are carried out with dose on a linear scale, the scientific significance of the shape of the curve is obscured. Examples of dose-response curves for carcinogenesis from 2-acetylaminofluorene, N-nitrosodiethylamine, aflatoxins, and radium are shown. In addition, more than 500 National Toxicology Program Technical Reports (NTP-TR) on carcinogenicity were examined; from this database, three groups of studies were selected. The first group consisted of those studies in which the lowest dose produced no tumors and the study had a positive dose response. The second group consisted of those studies with three or more doses, with a positive dose response producing tumors, but in which there were no tumors in the control group. The third group of more than 50 studies was from NTP-TR-00 to NTP-TR-52 that had only two data points with a positive dose response. These studies were all evaluated on the Rozman et al. scale, since it conforms to the laws of nature and allows evaluation of all doses. It was observed that virtually all of these NTP-TR carcinogenicity studies show a linear response when dose is on this logarithmic scale; a clear threshold for carcinogenicity is typical for nearly all of these chemicals. An exponential dose-response curve was a better fit for a few, but experimental error could account for this deviation from linearity. It is pointed out that there is strong experimental evidence that the mere presence of DNA adducts does not necessarily lead to tumor production. Hormesis probably applies to carcinogenesis and proof of this will require abandoning the no threshold concept. Experiments showing that cumulative dose is a better metric than daily dose may require reevaluating almost all carcinogenicity studies.