Predicting realistic RBE values for clinically relevant radiotherapy schedules.

To consider the therapeutic potential of radiation effect modifiers it is necessary to balance the modification of the injury in tumours with that in different types of normal tissue. It is especially important to ensure that the effects that have been demonstrated in preclinical experiments are both qualitatively and quantitatively relevant for the radiation doses that will be used in clinical schedules. Most radiobiology studies are initially performed with large single doses or a few large fractions, and from those results predictions have sometimes been made of the potential clinical benefit from a radiation modifier. In the clinic they will be used with many repeated small fractions of about 2 Gy over a period of several weeks. The effects will be quantitatively different in these two dose ranges for a variety of reasons. No modifiers of radiation effect are truly dose-modifying over the whole dose spectrum. They all have a differential effect on the type of damage inflicted at high and low dose levels, i.e. those described by the linear and quadratic terms in the LQ model. This means that every modifier has a dose or dose per fraction dependence on the magnitude of the sensitization or protection. The details of that dose dependence will vary with the alpha/beta ratio of the tissue under consideration. Furthermore all tissues and tumours contain a mixture of cells, with different proliferative, redox and other characteristics that influence their sensitivity to radiation and their susceptibility to the radiomodifier. The influence of different subsets of cells changes as a fractionated treatment progresses and the sensitive cells are eradicated, leaving more resistant survivors. The overall response to a fractionated schedule then depends critically on whether there is re-assortment of cells from the resistant phase into more sensitive or modifiable phases before the next fraction in the series. In addition, the magnitude of dose modification depends totally on the standard curve against which the comparison is made. The reference standard is different in preclinical laboratory studies and in conventional clinical experience. Those differences must be considered when moving from the laboratory to the clinic and back again. The effect of these different factors is considered using the linear quadratic model to dissect the components. Examples are provided to demonstrate the clinical relevance.