BM-01 * DOSIMETRIC HOTSPOTS ARE THE MOST POWERFUL DOSIMETRIC PREDICTOR OF LOCAL CONTROL IN NSCLC BRAIN METASTASES

Abstract

BACKGROUND: There is scant data on actionable parameters that may improve local control (LC) after SRS for brain metastases. We evaluated the impact of various patient, tumor, and dosimetric variables on LC of NSCLC brain metastases with SRS. METHODS: Patients with NSCLC brain metastases treated with single fraction SRS without prior metastasectomy or SRS to the same lesion were identified. Brain MRI imaging was reviewed from time of SRS until death or last follow up. LC was scored for each lesion as stability or decrease in the size of the treated metastasis. Dose volume histograms (DVH) were collected for each lesion and evaluated with a custom DVH tool. V18-V50Gy, D10, D50, D90, dose mean, maximum, minimum, and median were extracted as were dose gradient index and RTOG conformity index. Univariate and multivariate Cox Regression were used to identify factors predictive of LC, and LC was estimated using the Kaplan-Meier method. RESULTS: 612 brain metastases in 299 patients were identified, yielding 122 local failures. On univariate analysis, lack of prior whole brain, lack of whole brain < 2 months, Perfexion machine model, increasing dose, increasing isodose, fewer shots, increasing V18-V40, decreasing volume, and increasing conformity index were associated with increased LC (p < 0.05). On multivariate analysis (MVA), decreasing volume (HR 1.359, 95% CI 1.197-1.543), increasing V32Gy (HR 0.069 95% CI 0.018-0.264), and increasing isodose (HR 0.953, 95% CI 0.911-0.996. were associated with increased LC (p < 0.05). CONCLUSIONS: Significant plan quality variation exists for a given prescription dose for SRS, and prescription dose is not predictive of LC. Optimizing the volume of tumor receiving high dose (such as V32) for a given prescription dose represents an actionable variable to drive treatment planning during SRS for brain metastases. Analysis of a larger number of lesions within a multi-institutional registry (CONDR-IMD) will further clarify such parameters.