Shape of the dose-response curve for radiation carcinogenesis. Extrapolation to low doses

It follows from microdosimetric principles that biological effects which require for their production the interaction of two independent sublesions will have both a linear and a quadratic term in their dose-response relationship. Thus the incidence, I, of these effects will be related to the dose, D, of radiation by the equation: I = αD + βD2, where α and β are constants. Part of the justification for this equation is that there is a high-LET component to all low-LET radiation. It therefore follows that if the initial part of the dose-response curve for cancer induction by high-LET radiation is linear and is independent of dose rate, then the same must be true for low-LET radiation. This initial linear portion of the curve will underestimate the true response by a factor of 2 at a radiation dose equal to α/β, and by progressively more at higher doses. A review of the values of α/β from dose-response curves for chromosome aberrations and mutations following low-LET radiation shows that these values are related to the amount of DNA in the genome, and that for mammalian cells this dose is approximately 100 rad. (The geometric mean of 16 values is 127 rad with a 95 percent confidence interval from 85 to 190 rad.) This suggests that if the initial radiation lesion triggering a human cancer is a mutational or cytogenetic event, risk estimates derived from doses of 100 rad will overestimate the cancer risk at low doses and/or low dose rates by a factor of 2 or less. An analysis of risk estimates for radiation-induced cancers in humans supports this prediction