Radon-induced lung cancer in smokers and non-smokers: risk implications using a two-mutation carcinogenesis model

Three sets of data (population statistics in non-smokers, data from an investigation of the smoking habits of British doctors and a study of Colorado uranium miners) were used to analyse lung cancer in humans as a function of exposure to radon and smoking. One of the aims was to derive implications for radon risk estimates. The data were analysed using a two-mutation radiation carcinogenesis model and a stepwise determination of the model parameters. The basic model parameters for lung cancer were derived from the age dependence fit of the spontaneous lung cancer incidence in non-smokers. The effect of smoking was described by two additional parameters and, subsequently, the effect of radon by three other parameters; these five parameters define the dependence of the two mutation steps on smoking and exposure to radon. Using this approach, a consistent fit and comprehensive description of the three sets of data have been achieved, and the parameters could, at least partly, be related to cellular radiobiological data. The model results explain the different effect of radon on non-smokers and smokers as seen in epidemiological data. Although the analysis was only applied to a limited number of populations, lung cancer incidence as a result of radon exposure is estimated to be about ten times higher for people exposed at the age of about 15 than at about 50, although this effect is masked (especially for smokers) by the high lung cancer incidence from smoking. Using the model to calculate the lung cancer risks from lifetime exposure to radon, as is the case for indoor radon, higher risks were estimated than previously derived from epidemiological studies of the miners' data. The excess absolute risk per unit exposure of radon is about 1.7 times higher for smokers of 30 cigarettes per day than for non-smokers, even though, as a result of the low spontaneous tumour incidence in the non-smokers, the excess relative risk per unit exposure for the smokers is about 20 times lower than for the non-smokers. This prediction could have serious consequences for the transfer of risk estimates between populations. Although the solution of the model presented here is not unique but dependent on the model assumptions, the predictions and risk implications are sufficiently supported to justify a thorough investigation of the applicability of the model to other radon data sets.