Use of risk projection models for the comparison of mortality from radiation-induced breast cancer in various populations

The aim of our study is to compare lifetime risk estimates for radiation-induced breast cancer between various populations. Some epidemiological studies, mainly on Japanese A-bomb survivors, but also on North American populations irradiated for medical reasons, provide risk coefficients for excess mortality from breast cancer after exposure to ionizing radiation. For this comparative study, these risk coefficients have been transferred into the Japanese, American, and French populations, using demographic data, risk projection models, and assuming an individual acute exposure. The lifetime risk estimates are similar in the three populations when the additive projection model is used because of the similarity of the life tables of the three countries. However, using the multiplicative model, large differences appear due to the discrepancies in the baseline breast cancer rates between the Japanese and Occidental populations. It is for this reason that transfer of risk coefficients fitted on the Japanese population data must be considered with caution when applied to an Occidental population. Lifetime risk estimates for a given population, the French population for example, are largely dependent on the risk coefficients considered from various epidemiological studies. Nevertheless, for each source of data, they are higher with the multiplicative model than with the additive one. Moreover, it should be noted that data from medical irradiation lead to higher risk estimates when using the additive model, whereas studies of the A-bomb survivors lead to higher risk estimates with the multiplicative model. This comparative study points out the variations of the lifetime risk calculations according to the baseline breast cancer mortality rates of the three populations and with the use of different mathematical projection models.