Epidemiological studies on the effects of low-level ionizing radiation on cancer risk

The health effects of low-level ionizing radiation are yet unclear. As pointed out by Upton in his review (Upton, 1989), low-level ionizing radiation seems to have different biological effects from what high-level radiation has. If so, the hazard identification of ionizing radiation should he conducted separately for low- and high-level ionizing radiation; the hazard identification of low-level radiation is yet to be completed. What makes hazard identification of ionizing radiation difficult, particularly in the case of carcinogenic effect, is the difficulty in distinguishing radiation-induced cancer from other cancers with respect to clinicopathological features and molecular biological characteristics. Actually, it is suspected that radiation-induced carcinogenesis involves mechanisms not specific for radiation, such as oxidative stress. Excess risk per dose in medium-high dose ranges can be extrapolated to a low-dose range if dose-response can be described by the linear-non-threshold model. The cancer risk data of atomic-bomb survivors describes leukemia risk with a linear-quadratic (LQ) model and solid-cancer risk with linear non-threshold (LNT) model. The LQ model for leukemia and the LNT model for solid cancer correspond to the two-hit model and the one-hit model, respectively. Although the one-hit model is an unlikely dose-response for carcinogenesis, there is no convincing epidemiological evidence supporting the LQ model or non-threshold model for solid cancer. It should be pointed out, however, even if the true dose response is non-linear various noises involved in epidemiological data may mask the truth. In this paper, the potential contribution of epidemiological studies on nuclear workers and residents in high background radiation areas will be discussed. (author)