Irradiated volume as a predictor of brain radionecrosis after linear accelerator stereotactic radiosurgery.

PURPOSE: To investigate the correlation between volume of brain irradiated by stereotactic radiosurgery (SRS) and the incidence of symptomatic and asymptomatic brain radionecrosis (RN).

METHODS AND MATERIALS: A retrospective analysis was performed of patients treated with single-fraction SRS for brain metastases at our institution. Patients with at least 6-month imaging follow-up were included and diagnosed with RN according to a combination of criteria, including appearance on serial imaging and histology. Univariate and multivariate analyses were performed to determine the predictive value of multiple variables, including volume of brain receiving a specific dose (V8 Gy-V18 Gy).

RESULTS: Sixty-three patients were reviewed, with a total of 173 lesions. Most patients (63%) had received previous whole-brain irradiation. Mean prescribed SRS dose was 18 Gy. Symptomatic RN was observed in 10% and asymptomatic RN in 4% of lesions treated. Multivariate regression analysis showed V8 Gy-V16 Gy to be most predictive of symptomatic RN (p < 0.0001). Threshold volumes for significant rise in RN rates occurred between the 75th and 90th percentiles, with a midpoint volume of 10.45 cm(3) for V10 Gy and 7.85 cm(3) for V12 Gy.

CONCLUSIONS: Analysis of patient and treatment variables revealed V8 Gy-V16 Gy to be the best predictors for RN using linear accelerator-based single-fraction SRS for brain metastases. We propose that patients with V10 Gy >10.5 cm(3) or V12 Gy >7.9 cm(3) be considered for hypofractionated rather than single-fraction treatment, to minimize the risk of symptomatic RN.