The effects of neutrons in Hiroshima. Implications for the risk estimates

Risk estimates for radiation-induced late effects are relevant to various considerations in radiation protection. Most of these considerations relate to small doses for which no excess risk can be seen even in extensive epidemiological studies. Risk coefficients for radiation protection must, therefore, be based on uncertain extrapolation of observations obtained at moderate or high doses. The extrapolation can not be replaced, as yet, by new, more direct information on processes such as radiation-induced genetic instability or adaptive response. While the new findings indicate complexities that may be highly relevant to the effectiveness- or lack of effectiveness- of radiation at low doses, they remain insufficiently understood to permit a decision as to whether dose-effect relations are linear, curvilinear, or have a threshold in dose. In view of these uncertainties radiation-protection regulations are, today, based on the conservative assumption of a linear dose dependence without threshold. This approach assures a sufficient degree of protection, but it may become unreasonably over-conservative, when the cautious hypothesis is treated as proven fact, and when-in addition-the assumed initial slope of the dose relation is not critically evaluated. A reliable evaluation needs to be based on the follow-up of the atom-bomb bomb survivors, and several major aspects of current interest are discussed here. a) Mortality from solid tumours in Hiroshima shows a statistically significant excess at a colon dose of 50 mGy; however, it is likely that this is the result of a bias in assigning causes of death. b) The solid cancer mortality data of the atom-bomb survivors are consistent with linearity in dose, but they can be shown to be equally consistent with a considerable degree of curvature. c) Even with the present dosimetry system, DS86, a substantial part of the effect at small doses in Hiroshima could be due to neutrons. If this is the case, the risk estimates for gamma-rays need to be accordingly decreased. d) Numerous neutron-activation measurements in Hiroshima indicate that the DS86 underestimates the neutron doses. The evidence is, up to now, based only on activation products of low energy neutrons, but efforts are currently underway to determine activation products of high energy neutrons. If these measurements should substantiate the present trend, the cancer data in Hiroshima would cease to be reliable proof of an effect of gamma-rays at low doses. Instead the dose dependence for gamma-rays could be purely quadratic, and any initial slope in the linear-quadratic dependence might well be attributable to neutrons only.