SURG-12. A COMPARISON BETWEEN PROGNOSTIC SCORES OF BREAST CANCER BRAIN METASTASES APPLIED TO A NEUROSURGICAL POPULATION

INTRODUCTION: Prognostic scores have been developed to predict overall survival (OS) in patients with brain metastasis from breast cancer, comprising different combinations of prognostic factors. A new prognostic score including the number of brain metastases has been proposed. We aimed to evaluate the use of these prognostic scores on a neurosurgical population. METHODS: Retrospective study with consecutive patients with brain metastasis from breast cancer treated in the neurosurgery department. Clinical end point is overall survival estimated by the Kaplan Meier method. Univariate and multicovariate Cox proportional hazard models are applied to estimate the effect of covariates of interest on OS. We employ Bootstrap validation method to estimate the bias-corrected or over fitting-corrected predictive accuracy of Cox models, which is presented by concordance index(C-index). RESULTS: 315 consecutive patients with brain metastasis from breast cancer. Median OS was 14 months(95% CI 11–16.9), KPS, number of brain metastases, biological subtypes, age and presence of extracranial metastases were significantly associated with improved OS on the univariate analysis. Multivariate analysis showed that KPS, biological subtype, age and number of brain metastases were statistically significant for OS. The recursive partitioning analysis(RPA) classes, the graded prognostic assessment(GPA) score, the diagnostic specific GPA(DS-GPA) and the modified DS-GPA identified individual subgroups with different OS. RPA and DS-GPA had OS statistical significant between all groups with a C-index of 0.561 and 0.586 respectively. DS-GPA had a c-index of 0.639 and modified DS-GPA had an c-index of 0.637. DS-GPA and modified DS-GPA had a better performance in terms of discrimination when compared to RPA(p< 0.001) and GPA(p=0.01). CONCLUSIONS: DS-GPA and modified DS-GPA were able to better discriminate subgroups OS, which most likely reflects the use of biological subtype in the score calculation. The incorporation of number of BM by the modified DS-GPA improved the distinction between the higher score and the lower score group.