Breast Cancer Incidence Among Atomic Bomb Survivors: Implications for Radiobiologic Risk at Low Doses

A theoretical form suggested by radiobiologic principles was used to investigate the extent to which linear interpolation may overestimate or underestimate breast cancer risk at low doses and the risk estimate obtained with the use of a simple linear model was not greater than estimates obtained with more general models that fit the data more closely.

Abstract

Supplemental dose-response analyses of breast cancer incidence data for Japanese A-bomb survivors during the period 1950–69 were presented. The first analysis led to an unpublished result cited in a report on risks associated with mammography in mass screening for breast cancer, namely, that the inference for a radiation among the A-bomb survivors did not depend solely on interpolation between values at 0 and high doses. In the second analysis, a theoretical form suggested by radiobiologic principles was used to investigate the extent to which linear interpolation may overestimate or underestimate breast cancer risk at low doses. The functional form, I(D) = (α0+α1D+α2D)e−β1D−β2D, where I(D) denoted breast cancer incidence at dose D and where all parameters were constrained to be positive, was treated as a basically linear form for description of the risk at low doses, with additional parameters to allow for upward curvature over the low-to-intermediate range and downward curvature at high doses. The analysis provided direct evidence of a radiation breast carcinogenesis effect at doses under 49 rads kerma (<40 rads to breast tissue). Evidence for an effect was contributed by the data from both Hiroshima and Nagasaki. The incidence data were too sparse to express all the subtleties of the above functional form for the dose response, and the form was simplified by the elimination of some parameters. The risk estimate obtained with the use of a simple linear model was not greater than estimates obtained with more general models that fit the data more closely.