The individual risk of symptomatic radionecrosis after brain metastasis radiosurgery is predicted by a continuous function of the V12Gy()

INTRODUCTION: Brain metastases are frequently treated with stereotactic radiosurgery (SRS). Radionecrosis (RN) is the late side effect in up to 24% of patients, being symptomatic in 8–10%. Fixed values of the radiosurgical volume receiving 12 Gy or more (V12Gy) are used to roughly predict the global risk. The aim of this retrospective study is to fine-tune the model of individual risk prediction for symptomatic radionecrosis and identify modulating factors. MATERIALS AND METHODS: Data of patients treated with SRS for ≤3 BM of solid tumours at CHU-UCL-Namur were retrospectively reviewed. Doses ranging from 15 to 24 Gy were prescribed to the 70% isodose in function of the lesion diameter. Treatment was administered with a stereotactic linear accelerator. Follow-up magnetic resonance imaging was performed 3-monthly for 18 months and 6-monthly thereafter. RN was prospectively diagnosed and confirmed by the tumour board. V12Gy, previous or salvage whole-brain radiotherapy (WBRT), smoking history, diabetes, postoperative SRS, diagnosis-specific graded prognostic assessment score, cerebral lobe location and relative location (superficial versus deep) were retrieved. Univariate and multivariate analyses were performed to assess their predictive values and derive a model. RESULTS: 128 patients with 220 lesions were analysed. The risk of RN was predicted by a continuous function of the V12Gy (p = 0.005). No other factor had a significant impact, particularly WBRT that did not increase the risk. CONCLUSION: The risk of symptomatic RN is predicted on an individual basis by a model in function of the V12Gy and must be confirmed in a prospective study.