ESTRO 38 Abstract book

S692 ESTRO 38

individual tumor volume as input to predict individual V12 Gy. Regression using 16-patient training data yielded an R- square of 0.9498. This model was then applied to the whole 30 patients to predict individual V12Gy for each tumor, which was then summed within each patient to give predicted per-plan V12Gy. The predicted V12Gy using this method resulted an RMSE of 5.9 cm³, and on average over-predict by 3.4±4.8 cm³. (3) The geometric expansion method showed that a 5 mm expansion from each tumor’s equivalent sphere provided most stable prediction of actual V12Gy volumes with a fixed offset of 13.7 cm³ for nearly all tumor sizes. Using 5 mm expansion minus 13.7 cm³ as prediction, the predicted V12Gy showed an RMSE of 4.0 cm³. This result was not substantially improved over simple linear regression method.

91% [95% CI, 85%-95%], 88% [95% CI, 81%-93%] and 81% [95% CI, 70%-88%], respectively. Distant control rate at 1 year was 48% [95% CI, 81%-93%]. In multivariate analysis, primary NSCLC (p=0.007), the number of extra-cerebral metastatic sites (p=0.003) and planning target volumes (p=0.012) were associated with OS. There was no factor predictive of time to local progression. Median OS was 15.2 months [95% CI, 12.0-17.9 months] and the OS rate at 1 year was 58% [95% CI, 50%-65%]. Salvage radiotherapy was administered to 72 patients (45%), of which 49 received new HFSRT. Five patients underwent re- irradiation of the surgical bed by stereotactic radiation therapy. Ten (7%) patients presented late grade 2 and 4 (3%) patients late grade 3 toxicities. Thirteen (8.9%) patients developed radiation necrosis (all grades). Conclusion This large multicenter retrospective study shows that HFSRT allows for good local control of metastasectomy tumor beds and that this technique is well-tolerated by patients. EP-1258 Predicting Brain V12Gy for Single-Isocenter Multi-Target Stereotactic Radiosurgery (SRS) H. LIU 1 , T. Li 1 , H. Zhang 2 , W. Shi 1 1 Thomas Jefferson University Hospital, Radiation Oncology, Philadelphia, USA ; 2 Monte Vista Christian School, n/a, Watsonville, USA Purpose or Objective Stereotactic radiosurgery (SRS) for multiple brain metastasis has been shown to provide excellent local control with reduced cognitive impact. For this type of treatment, the volume of brain receiving 12Gy (V12Gy) has been shown to correlate with the occurrence of radionecrosis following SRS. This study aimed to determine a simple and efficient model for predicting V12Gy for patients undergoing stereotactic radiosurgery (SRS) for multiple brain tumors in a single session prior to treatment planning. Material and Methods 30 single-isocenter-multi-target SRS cases planned with Brainlab Elements™ treatment planning system were retrospectively analyzed. A total amount of 221 tumors were prescribed to 15-24 Gy. V12Gy for each individual tumor and for each plan were extracted. Three methods were used to model V12Gy based on anatomy and prescription: a linear regression model on per-plan V12Gy combining all targets, a linear regression model on per- target V12Gy followed by summation per plan, and 3D geometric per-target expansion model that assumes V12Gy occurs at a certain distance outside each target. Actual-vs-predicted graphs and root-mean-square errors for per plan V12Gy were used to assess the prediction accuracy of each model. Results (1) Linear correlation analysis showed improved correlation when using the prescription-weighted sum of tumor volumes per plan (R-square = 0.9618) as input instead of using only total tumor volume (R-square = 0.9135). Using this simple linear model, the predicted V12Gy resulted an RMSE of 3.8 cubic-cm.

Conclusion In this study we compared three empirical models to predict brain V12Gy using only tumor size and prescription information, which is readily available before planning. The results showed that the simple linear model performed reasonably well and offered very similar goodness of prediction compared to more complicated 2- step or 3D-based models. This linear model used prescription-weighted total tumor volume for a particular patient as input, and could help clinicians decide how many tumors and what prescription should be given without exceeding desired V12Gy limit even before treatment planning, therefore save clinical resources and improve patient throughput. EP-1259 Retrospective Analysis of radiosurgery for ≥ 4 brain metastases from oncogene-addicted NSCLC Y. Wang 1 , C. Calbat 1 , P. Allen 1 , N. Guha 1 , D. Gomez 1 , J. Li 1 1 University of Texas- MD Anderson Cancer Center, Radation Oncology, Houston, USA Purpose or Objective The prolonged survival in NSCLC patients with EGFR- mutation or ALK-rearrangement treated with target agents calls for treatment of their brain metastases (BMs) with minimal toxicity and less systemic interruption. Although stereotactic radiosurgery (SRS) without whole brain radiation therapy (WBRT) is the preferred initial strategy for patients with 1-3 BMs, the paradigm remains elude for those with more than 3 cerebral lesions, particularly given the CNS penetration of available target therapies. The current study aims to investigate the overall survival, CNS progression and the rate of salvage WBRT in patients of EGFR-mutated or ALK- rearranged NSCLC with 4 or more BMs treated with SRS. Material and Methods We retrospectively reviewed NSCLC patients with EGFR- mutation or ALK-rearrangement who received Gamma Knife (GK) radiosurgery for 4 or more BMs in a single radiosurgery session at MD Anderson cancer center between 2009 and 2018. Overall survival, intracranial progression and freedom from WBRT were estimated using Kaplan-Meier analysis. Results Thirty-two patients, including 25 patients with EGFR mutation and 7 with ALK re-arrangement were identified and included in the current study with a total of 405 lesions treated with SRS. The median follow-up time was

(2) The linear fitting model used prescription-weight