Impact of Radionecrosis and Local Recurrence on Overall Survival After Stereotactic Radiosurgery for Brain Metastases.

PURPOSE: Radionecrosis (RN) is a dose-limiting toxicity after stereotactic radiosurgery (SRS) for brain metastases and is hard to differentiate from treated tumor progression. The outcomes of patients with RN following SRS are poorly described, particularly compared with those with local central nervous system progression. Therefore, we characterize outcomes of patients who developed RN versus those with treated tumor recurrence in the absence of out-of-field intracranial progression after SRS. METHODS AND MATERIALS: This cohort study included patients who completed an initial course of SRS between 2015 and 2020. The primary outcome was overall survival (OS) by the presence of RN or treated tumor local recurrence (LR), analyzed using the Kaplan-Meier method. Binomial logistic regression was used to test parameters associated with RN or LR. RESULTS: A total of 1383 patients were included; 758 (55%) were women, and the median age was 63.4 years. Common primaries were lung (n = 757, 55%), breast (203, 15%), and melanoma (118, 9%). At a median follow-up of 8.7 months, 143 patients (10%) had concern for intracranial progression or RN at irradiated sites. Following a multidisciplinary evaluation, 96 patients were ultimately diagnosed with RN and 47 with LR. Patients with LR were diagnosed earlier after SRS than those with RN (median 5.8 vs 8.3 months; P = .0087), and they had worse OS than those with RN (median 15.2 vs 40.4 months; P < .001). Patients symptomatic for either LR or RN presented earlier after SRS than asymptomatic patients and had worse OS. In binomial logistic regression, RN was associated with single-fraction SRS (odds ratio [OR], 0.38; P = .012), receipt of immunotherapy after SRS (OR, 0.32; P < .014), and a longer interval from SRS (OR, 0.94; P = .028). CONCLUSIONS: In this multi-institutional cohort study of brain metastasis patients treated with SRS, those with RN had a significantly better OS than those with LR. These data inform surveillance, diagnosis, and prognosis of these patients.

Duke Scholars

Author Trey Carlton Mullikin Radiation Oncology

Author John P. Kirkpatrick Radiation Oncology

Author Paul Wayne Sperduto Radiation Oncology

Author Scott Richard Floyd Radiation Oncology

Author Joseph Kamel Salama Radiation Oncology

Author Zachary James Reitman Radiation Oncology