Cancer risk modelling and radiological protection.

Statistical models describing how the radiation-related risks of particular types of cancer vary with the doses of radiation received by specific tissues are derived from data gathered in epidemiological studies of exposed groups of people, guided by an incomplete understanding of radiobiological mechanisms gleaned from experimental studies. Cancer risk models have been developed for a dozen or so different types of cancer, and take account of the effect of important risk modifying factors such as age at exposure and time since exposure. Of primary importance in the development of cancer risk models is the experience of the Japanese atomic bomb survivors, but other exposed groups contribute information, including those exposed to radiation from internally deposited radioactive material, such as inhaled radon. Cancer risk models predict that at low doses or low dose rates the excess risk of cancer is directly proportional to the dose of radiation received, with no threshold dose--the linear no threshold (LNT) dose-response model--and the inferred summary estimate of the overall average lifetime excess risk of developing a serious cancer is ∼ 5%/Sv. It is these cancer risk models and this inferred nominal risk estimate that provide the technical basis of radiological protection. Although it is difficult to definitively test the LNT model at low doses or low dose rates, because the predicted excess risk is small compared with fluctuations in the baseline risk, evidence exists that a small risk of cancer results from low-level exposure to radiation and that the excess risk is around that predicted by current risk models.