Tumor Bed Radiosurgery Following Resection and Prior Stereotactic Radiosurgery for Locally Persistent Brain Metastasis

PURPOSE Despite advances in multimodality management of brain metastases, local progression following stereotactic radiosurgery (SRS) can occur. Often, surgical resection is favored, as it frequently provides immediate symptom relief as well as pathological characterization of any residual tumor. Should the pathological specimen contain viable tumor cells, further radiation therapy is an option to sterilize the tumor bed. We evaluated the use of repeat SRS (rSRS) in lieu of whole-brain radiation therapy (WBRT) as a means of improving local control (LC) while minimizing potential toxicity and dose to the normal brain. MATERIALS/METHODS A retrospective review was performed to identify patients with brain metastases who underwent SRS and then surgical resection for locally recurrent or persistent disease. From 2004 to 2014, 13 consecutive patients or 15 lesions were treated with rSRS after resection, either post-operatively to the tumor bed (n = 10, 66.6%) or after a second local recurrence (n = 5, 33.3%). LC, distant brain failure (DBF), and radiation toxicity were determined using patient records, RECIST criteria v1.1, and CTCAE v4.03. RESULTS At a median follow-up interval of 9.0 months (range 1.8-54.9 months) from time of rSRS, five patients remain alive. Following rSRS, 13 of the 15 (86.6%) lesions were locally controlled with an estimated 100% LC at 6 months and 75% LC at 1 year. However, 11 of the 15 (73.3%) treated lesions developed DBF after rSRS with 3 of 13 patients proceeding to WBRT. Two of 15 (13.3%) resulted in either grade 2 radionecrosis with grade 3 seizures or grade 3 radionecrosis. CONCLUSION Repeat SRS represents a potential salvage therapy for patients with locally recurrent brain metastases, providing additional tumor control with acceptable toxicity, even in the setting of prior SRS and surgical resection. rSRS may be reasonable to use as an alternative to WBRT in this setting.