Optimizing anisotropic margins in single-isocenter multiple brain metastases radiosurgery using regressor strategies:...

To explore a comprehensive method for assigning anisotropic margins in single-isocenter multiple-metastasis radiosurgery (SRS), utilizing regressor strategies with geometric, dosimetric, and mechanical information. Such margins may reduce the volume of the irradiated brain without compromising the dose to lesions, by considering additional treatment variables.

First, the impact of slice thickness on the margin was investigated on an anthropomorphic phantom for different lesion volumes. In addition, for clinical cases, the anisotropic margins were assigned using two methods: by geometric margin criterion (GMC) based on clinical protocols and optimized margin criterion (OMC) based on regressor strategies. To evaluate the plans (456 metastases in total), we reported the Paddick conformity, gradient, and efficiency indices, monitor units, the volume of brain receiving at least 12 Gy (V12), and risk of radionecrosis by two radiobiological models (Flickinger and normal tissue control probability models).

An overestimation of volumes was related (absolute difference of 0.13 ± 0.18 cm³) to a larger slice thickness and smaller target volumes. The OMC demonstrated an overall reduction (20%) of the target volumes. OMC also showed improvement in gradient index and monitor units compared to GMC. V12 showed a statistically significant decrease when OMC was used compared to GMC, ranging from 5% to 41% across all three institutions (p = 0.045). The radiobiological models pointed out that the risk of radionecrosis was lower than 10%, and there was no statistically significant difference between GMC and OMC (p = 0.151).

While the feasibility and potential dosimetric benefits of anisotropic margins in SRS-such as improvements in the gradient index and V12-were observed, the magnitude of these differences may be modest in terms of clinical impact. The results suggest that anisotropic margins can be integrated into SRS workflows and may offer advantages in specific scenarios, particularly when minimizing brain dose is a priority.