Risk Factors for Radiation Necrosis After Radiosurgery for Brain Metastases

Abstract

To determine the incidence, time-course, and risk factors associated with radiation necrosis after single-fraction stereotactic radiosurgery (SRS) for brain metastases (BM). All patients treated with SRS for BM September 1998 through 2009 were reviewed. Patients/lesions were excluded for death ≤2 months after SRS or insufficient follow-up imaging (<3 months of imaging follow-up, >8-month imaging gap in the first year, or inadequate imaging availability). Dates of necrosis, failure, or both failure and necrosis were recorded. An additional category, failure versus necrosis, was scored in the event of >25% lesion increase without confirmation of failure and/or necrosis based on surgical pathology or serial MRI ± perfusion MRI. Freedom from necrosis after SRS (allowing calculation of actuarial necrosis risk) was calculated using the Kaplan-Meier method with censoring at last imaging, excluding all cases of failure versus necrosis to give best-case necrosis risk or including all cases of failure versus necrosis for worst-case necrosis risk. Of 900 patients and 4093 BMs, 436 patients and 2057 BMs were evaluable; 93 patients died ≤2 months after SRS and 371 patients had insufficient follow-up imaging. The median survival time in evaluable patients was 17.4 months after SRS, with median follow-up 36.3 months in living patients and median imaging follow-up of 9.9 months in censored lesions. The appearance of necrosis was noted to be variable and frequently indistinguishable from tumor progression at single points in time; review of serial imaging was key, along with pathology when available. Overall, 136 BM (6.6%) developed necrosis at a median of 7.6 months after SRS (range, 0.9-52.0 months; 85% between 3 and 18 months); another 86 (4.2%) were indeterminate for failure versus necrosis. Best- and worst-case 1-year necrosis probabilities were 6-9% for SRS alone, 7-15% after prior radiation therapy, 12-17% for SRS concurrent with radiation therapy, and 35-43% for repeat SRS after prior SRS (p < 0.001). Excluding 86 BM treated with repeat SRS, best- and worst-case 1-year necrosis probabilities were 3-6% for breast, 6-9% for melanoma, 10-16% for lung, and 15-23% for kidney BM (p < 0.001), and 0%, 2-4%, 8-14%, 17-24%, 22-32%, and 25-30% for BM ≤0.5, 0.6-1.0, 1.1-1.5, 1.6-2.0, 2.1-3.0, and >3.0 cm maximum diameter (p < 0.001), respectively. Evaluation of concurrent chemotherapy as a risk factor for necrosis is in progress. The risk of necrosis after SRS varies significantly by primary site, BM diameter, and other radiation treatment (prior SRS > concurrent radiation therapy > prior radiation therapy).