Radiosurgery for brain metastases: relationship of dose and pattern of enhancement to local control

This study aimed to analyze dose, initial pattern of enhancement, and other factors associated with freedom from progression (FFP) of brain metastases after radiosurgery (RS).

All brain metastases treated with gamma-knife RS at the University of California, San Francisco, from 1991 to 1994 were reviewed. Evaluable lesions were those with follow-up magnetic resonance or computed tomographic imaging. Actuarial FFP was calculated using the Kaplan-Meier method, measuring FFP from the date of RS to the first imaging study showing tumor progression. Controlled lesions were censored at the time of the last imaging study. Multivariate analyses were performed using a stepwise Cox proportional hazards model.

Of 261 lesions treated in 119 patients, 219 lesions in 100 patients were evaluable. Major histologies included adenocarcinoma (86 lesions), melanoma (77), renal cell carcinoma (21), and carcinoma not otherwise specified (17). The median prescribed RS dose was 18.5 Gy (range, 10-22) and the median tumor volume was 1.3 ml (range, 0.02-30.9). The initial pattern of contrast enhancement was homogeneous in 68% of lesions, heterogeneous in 12%, and ring-enhancing in 19%. The actuarial FFP was 82% at 6 months and 77% at 1 year for all lesions, and 93 and 90%, respectively, for 145 lesions receiving > or = 18 Gy. Multivariate analysis showed that longer FFP was significantly associated with higher prescribed RS dose, a homogeneous pattern of contrast enhancement, and a longer interval between primary diagnosis and RS. Adjusted for these factors, adenocarcinomas had longer FFP than melanomas. No significant differences in FFP were noted among lesions undergoing RS for recurrence after prior radiotherapy (119 lesions), RS alone as initial treatment (45), or RS boost (55).

A minimum prescribed radiosurgical dose > or = 18 Gy yields excellent local control of brain metastases. The influence of pattern of enhancement on local control, a new finding in this retrospective analysis, needs to be confirmed.