A study of influencing factors and predictors for radiation encephalic necrosis after CyberKnife radiotherapy

Objective: To investigate the influencing factors and predictors for radiation encephalic necrosis after CyberKnife radiotherapy. Methods: Ninety-four patients (104 targets) with primary or metastatic intracranial tumors who were treated with CyberKnife radiotherapy from 2006 to 2011 were retrospectively analyzed. All surgeries adopted skull tracking modes with a dose of 12-45 Gy in 1-8 fractions prescribed to 60%-87% isodose line. Radiation encephalic necrosis was determined by imaging or pathological examination. Logistic regression analysis was used to analyze the relationship between radiation encephalic necrosis and factors including diabetes, cardio-cerebrovascular diseases, target volume, isodose line, prescribed dose, number of fractions, combination with whole-brain irradiation (WBI), and biologically equivalent dose (BED). Predictability and critical threshold of all influencing factors for radiation encephalic necrosis were determined by the receiver operating characteristic (ROC) curve. Results: Twelve targets (11.54%) had radiation encephalic necrosis. According to the result of logistic regression analysis, BED, combination with WBI, and number of fractions were influencing factors for radiation encephalic necrosis. In the ROC curves, the areas under curves for the above three factors were 0.892 #+-# 0.034, 0.650 #+-# 0.072, and 0.712 #+-# 0.064, respectively, indicating that only BED can well predict radiation encephalic necrosis after CyberKnife radiotherapy with a dose threshold of > 7410 cGy. Conclusions: BED, combination with WBI, and number of fractions are influencing factors for radiation encephalic necrosis. BED is the best predictor of radiation encephalic necrosis with a dose threshold of > 7410 cGy. (authors)