EXTRAPOLATING RADIATION-INDUCED CANCER RISKS FROM LOW DOSES TO VERY LOW DOSES

Abstract

There is strong evidence that ionizing radiation increases cancer risks at high doses (e.g., >or=1 Gy), and persuasive, if controversial, epidemiological evidence that cancer risks are increased at low doses ( approximately 10 mGy). Discussed here are the issues related to extrapolating radiation risks from low radiation doses to very low doses (<or=1 mGy) - for which purpose we are forced to rely on radiobiological evidence and biophysical arguments. At high doses, cells are typically hit by many tracks of radiation, while at low doses most cells are typically hit by a single track of radiation; at very low doses proportionately fewer cells are hit, again only by a single track of radiation. Thus, in comparing low doses to very low doses, the damage to hit cells remains essentially the same (a single radiation track passing through a cell), but what changes is the number of cells that are subjected to this same damage, which decreases linearly as the dose decreases. This is the argument for a linear no-threshold (LNT) model. It is important to emphasize that this LNT argument only applies to the extrapolation from low doses to very low doses, not from high to low doses. Of course there are caveats to this argument, such as the potential effects of phenomena such as inter-cellular communication and immunosurveillance, and the possibility of different radiobiological processes at very low doses, compared to low doses. However, there is little conclusive experimental evidence about the significance of these phenomena at very low doses, and comparative mechanistic studies at high doses vs. low doses will not be informative in this context. At present, we do not know whether such radiobiological phenomena would produce small or large perturbations, or even whether they would increase or decrease cancer risks at very low doses, compared with the prediction of a linear extrapolation from low doses.