Thresholds in chemical carcinogenesis.

It is common practice to base carcinogenic risk assessment on the view that there is no threshold for chemical carcinogenesis. In this context, the threshold is defined as a dose below which no effects are observed. Analysis of epidemiological and experimental data on chemical carcinogenesis confirms that no thresholds have been demonstrated in human or animal studies, including the ED01 study which used very large group sizes. Indeed, demonstrating a threshold by experiment is as difficult as proving a negative. Mechanistic data provide the justification for the assertion that thresholds do not exist in chemical carcinogeneses. This is commonly thought of as a dose-response relationship which is linear at low dose. It is noted that primary sites of action for other forms of toxicity--for example, inhibition of enzymes or occupation of receptors--also have a dose-response relationship which is linear at low dose. It is concluded that the use of differing methods of low-dose risk assessment for different toxicological endpoints is not justified. The same method should be used in order to provide symmetry in assessment of different risks; the influence of assumptions, such as the magnitude of safety factors or the mathematical model selected, should be clearly stated so that risk managers can make balanced judgments.