Normal tissue complication probabilities: Dependence on choice of biological model and dose-volume histogram reduction scheme

Purpose: To evaluate the impact of dose-volume histogram (DVH) reduction sc hemes and models of normal tissue complication probability (NTCP) on rankin g of radiation treatment plans. Methods and Materials: Data for liver complications in humans and for spina l cord in rats were used to derive input parameters of four different NTCP models. DVH reduction was performed using two schemes: "effective volume" a nd "preferred Lyman". DVHs for competing treatment plans were derived from a sample DVH by varying dose uniformity in a high dose region so that the o btained cumulative DVHs intersected. Treatment plans were ranked according to the calculated NTCP values. Results: Whenever the preferred Lyman scheme was used to reduce the DVH, co mpeting plans were indistinguishable as long as the mean dose was constant. The effective volume DVH reduction scheme did allow us to distinguish betw een these competing treatment plans. However, plan ranking depended on the radiobiological model used and its input parameters. Conclusions: Dose escalation will be a significant part of radiation treatm ent planning using new technologies, such as 3-D conformal radiotherapy and tomotherapy. Such dose escalation will depend on how the dose distribution s in organs at risk are interpreted in terms of expected complication proba bilities. The present study indicates considerable variability in predicted NTCP values because of the methods used for DVH reduction and radiobiologi cal models and their input parameters. Animal studies and collection of sta ndardized clinical data are needed to ascertain the effects of non-uniform dose distributions and to test the validity of the models currently in use. (C) 2000 Elsevier Science Inc.