Relationship between dose and risk reduction: statistical evaluation of a combination experiment with three hepatocarcinogenic N-nitrosamines in rats.

Data from an experiment on the single and combination effects of very low doses of N-nitrosodiethylamine (NDEA), N-nitrosopyrrolidine (NPYR) and N-nitrosodiethanolamine (NDELA) in 1800 male Sprague-Dawley rats were analysed for age-specific incidence (time to death with liver tumour) by Cox's proportional hazards model. The model revealed a linear relationship between daily exposure to low levels of N-nitrosamine and time to death with liver tumour within the dose range investigated: an increase in individual dose resulted in a proportional decrease in liver tumour-free survival. The finding was established for the three individual carcinogens as well as for their combination. NDEA was 40 times more active than NDELA. Extrapolation to N-nitrosamine exposure levels lower than those used in the experiment revealed only a minor reduction in age-specific liver tumour incidence compared to that achieved by an equivalent reduction within the experimental dose range. In rats at advanced age a further reduction in carcinogen-induced liver tumour incidence did not contribute to longer overall survival, due to competitive, probably independent causes of death. The data thus support the idea of a quasi-threshold in terms of a 'no observed effect level'.