Radiotoxicity in robotic radiosurgery: proposing a new quality index for optimizing the treatment planning of brain metastases

BACKGROUND As irradiated brain volume at 12 Gy (V12) is a predictor for radionecrosis, the purpose of the study was to develop a model for Cyberknife (CK) plans that is able to predict the lowest achievable V12 at a given tumor size and prescription dose (PD), and to suggest a new quality index regarding V12 for optimizing the treatment planning of brain metastases. METHOD In our model V12 was approximated as a spherical shell around the tumor volume. The radial distance between tumor surface and the 12 Gy isodose line was calculated using an approximation of the mean dose gradient in that area. Assuming a radially symmetrical irradiation from the upper half space, the dose distribution is given by the superposition of single fields. The dose profiles of a single field were derived by the measured off-center ratios (OCR) of the CK system. Using the calculated gradients of the sum dose profiles, minimal-V12 was estimated for different tumor sizes. The model calculation was tested using a phantom dataset and retrospectively applied on clinical cases. RESULTS Our model allows the prediction of a best-case scenario for V12 at a given tumor size and PD which was confirmed by the results of the isocentric phantom plans. The results of the non-isocentric phantom plans showed that an optimization of coverage caused an increase in V12. This was in accordance with the results of the retrospective analysis. V12 s of the clinical cases were on average twice that of the predicted model calculation. A good agreement was achieved for plans with an optimal conformity index (nCI). Re-planning of cases with high V12 showed that lower values could be reached by selecting smaller collimators and by allowing a larger number of total MU and more MU per beam. CONCLUSIONS V12 is a main parameter for assessing plan quality in terms of radiotoxicity. The index f12 defined as the ratio of V12 from the actual plan with the evaluated V12 from our model describes the conformity of an optimally possible V12 and thus can be used as a new quality index for optimizing treatment plans.