Generalised microdosimetric calculations of cell-to-cell variance

The cell-to-cell variance of an endpoint gives extra information, which is highly model dependent and thus potentially useful for distinguishing between possible mechanisms. Some years ago variances of dicentric chromosome aberrations were calculated using two assumptions: first, that dicentrics are produced independently by different radiation tracks, and second, that the dicentric yield due to a single track is Poisson distributed. It is shown that the formalism can be generalised so that the Poisson assumption is dropped. This affects predictions of neutron induced variances quite markedly. For lesions formed directly, such as immediate, irremediable chromosomal breaks, the formalism is general. For lesions formed by pairwise damage interaction, such as exchange-type chromosomal aberrations, the first assumption (single track action) limits the applicability of the results to high LET or low dose situations. In general, therefore, estimating variances requires the use of spatially inhomogeneous models. (author)