Inter-patient and intra-tumor radiosensitivity heterogeneity

Purpose: To examine possible effects of pretreatment intra-tumor and inter-patient radiosensitivity heterogeneities, and their relation-ships to in vitro SF2 assays and gamma(50) values. Methods: Computer simulations of expected gamma(50) and in vitro SF2 assay results for two different tumor local control models were performed. Both models allow for variation of the intra-tumor radiosensitivity distribution between patients. The first model assumes that intra-tumor radiosensitivity is lognormally distributed, but with variations in the distribution between patients. The second model simulates effects of genomic instability on repair capacity. The most resistant clonogens are assumed to be those which are "genomically undegraded." Results: a) The lognormal model predicts that SF2 CVs of 40% (which are commonly observed) would always lead to very shallow dose-response curves (gamma(50) < 0.7) which are in clear disagreement with the clinically observed slopes (gamma(50) > 2) for some sites in question. The genomic instability model, under certain parameter ranges, predicts that intra-tumor heterogeneity would have a nearly minimal effect on dose-response. With this model, SF2 CVs of 40% could be consistent with steep dose-response curves, in broad agreement with clinical observations for the sites in question. Reported measurement errors in assay reproducibility are much smaller than reported SF, CVs and are unlikely to account for the results. Inclusion of other sources of inter- or intra-tumor heterogeneity only make it more difficult to reconcile modeling results with wide measured SF2 CVs and relatively steep local control curves. Conclusions: The simulation results suggest that reconciling the very wide reported SF2 distributions with fairly steep dose-response curves requires significant intra-tumor radiosensitivity heterogeneity, the extent of which varies between patients. Large variations in SF2 values would then be due to large inter-patient fluctuations in the numbers of the more sensitive clonogens, whereas steep local control curves would be due to smaller inter-patient fluctuations in the radiosensitivity of the most resistant clonogens.