Dose-response modelling of stereotactic radiosurgery for brain metastases: Preliminary prospective data from the SAFER study (IRAS 276567).

Abstract

The number of people living with brain metastases (BMs) has increased in the last decade. Despite the large number of patients treated with stereotactic radiosurgery (SRS), there are very few prospective studies of the dose-response relationship to guide practice. We present preliminary dose-response data from a prospective observational study of post-SRS outcomes in patients with BMs (SAFER, IRAS 276567). The first 91 patients who did not have prior radiotherapy or neuro-surgery, to reach 9-months post-SRS follow up in the SAFER study were included. Local control (LC) probability was calculated using Kaplan Meier methodology. Prescription doses were converted to Biologically Effective Doses (BED) using the linear quadratic cubic (LQC) equation, and an alpha-beta ratio of 10. Logistic regression was used for dose-response modelling on a per-lesion basis. The overall 9-month LC rate was 92.01 %. The dose-response curve demonstrated a significant relationship between LQC model BED₁₀ and LC (odds ratio 1.145, slope 2.479, p = 0.0132). Based on the regression model, 71.50 % of those receiving LQC model BED₁₀ 35 Gy were predicted to have LC at 9 months. This increased to 97.4 % in those receiving LQC model BED₁₀ 55 Gy. This is a large prospective dataset on the dose-response relationship of BMs treated with SRS. Our results indicate an LQC model BED₁₀ 45 Gy is required to achieve 9-month LC of > 90 %. This is equivalent to single fraction-SRS doses > 20 Gy, or 3-fraction SRS doses > 27 Gy.