Local control after radiosurgery for brain metastases: predictive factors and implications for clinical decision

BACKGROUND To evaluate the local control of brain metastases (BM) in patients treated with stereotactic radiosurgery (SRS), correlate the outcome with treatment parameters and lesion characteristics, and define its implications for clinical decisions. METHODS Between 2007 and 2012, 305 BM in 141 consecutive patients were treated with SRS. After exclusions, 216 BM in 100 patients were analyzed. Doses were grouped as follows: ≤15 Gy, 16-20 Gy, and ≥21 Gy. Sizes were classified as ≤10 mm and >10 mm. Local control (LC) and overall survival (OS) were estimated using the Kaplan-Meier method. Log-rank statistics were used to identify the prognostic factors affecting LC and OS. For multivariate analyses, a Cox proportional model was applied including all potentially significant variables reached on univariate analyses. RESULTS Median age was 54 years (18-80). Median radiological follow-up of the lesions was 7 months (1-66). Median LC and the LC at 1 year were 22.3 months and 69.7%, respectively. On univariate analysis, tumor size, SRS dose, and previous whole brain irradiation (WBRT) were significant factors for LC. Patients with lesions >10 and ≤10 mm had an LC at 1 year of 58.6% and 79.1%, respectively (p = 0.008). In lesions receiving ≤15 Gy, 16-20 Gy, and ≥21 Gy, the 1-year LC rates were 39.6%, 71.7%, and 92.3%, respectively (p < 0.001). When WBRT was done previously, LC at 1 year was 57.9% compared with 78.4% for those who did not undergo WBRT (p = 0.004). On multivariate analysis, dose remained the single most powerful prognostic factor for LC. Median OS for all patients was 17 months, with no difference among the groups. CONCLUSIONS Dose is the most important predictive factor for LC of BM. Doses below 16 Gy correlated with poor LC. The SRS dose as salvage treatment after previous WBRT should not be reduced unless there is a pressing reason to do so.