2097 Background: Survival after the onset of metastases in the central nervous system is very short. However, some variables could indicate subsets of worse prognosis. Our aim was to determine the value of clinicopathological characteristics and prognostic scores in the post-SNC recurrence survival. Methods: We evaluated a retrospective cohort of 2597 breast cancer patients treated at the Instituto Nacional de Enfermedades Neoplasicas (Lima-Peru) between 2000-2005. Clinicopathological data was retrieved, RPA and GPA brain metastases prognostic scores were constructed and phenotypes were categorized according to the IHC expression in [HR+,HER2-], [Any HR, HER2+] and Triple Negative. Survival was calculated according to the Kaplan Meier methodology and cases were stratified by variables evaluated. The log-rank or Breslow tests were used when appropriate and multivariate analysis was done by the cox regression. A P<0.05 was considered statistically significant. Results: One hundred and fifty seven cases developed CNS metastasis, from which 23 developed leptomeningeal metastases. The post recurrence CNS survival was 0.405 years. There were not differences according to phenotype (P=0.102), histological grade (P=0.647), number of brain metastases (P=0.695) and metastases volume (P=0.155). We found statistic differences in regard to leptomeningeal carcinomatosis (present, 0.249ys vs absent 0.436ys; P=0.033); CSF infiltration (present, 0.115ys vs absent, 1.044ys; P=0.022); status of primary tumor (controlled, 0.501ys vs uncontrolled, 0.263ys; P<0.001); ECOG performance status (<2, 0.504ys vs ≥2, 0.288ys; P=0.030); and time from BC diagnosis to SNC metastases (<8 moths, 0.115 vs ≥8 months, 0.425ys; P=0.023). Cox regression identifies to CSF infiltration as statistically significant (HR=9.77; P=0.025). In regard to Prognostic scores, we found differences when cases were stratified according to RPA score (Class I, 0.564ys vs Class II, 0.455ys vs Class III, 0.288ys; P=0.049) and GPA score (0-1, 0.26ys vs 1.5-3, 0.455ys vs 3.5-5, 0,564; 0.048). Conclusions: RPA and GPA scores are more accurate to identify poor survival subsets in this group of patients than other tumor features (phenotype or histology).
149P Predicting prognosis of breast cancer patients with brain metastases in the BMBC registry: Comparison of three different prognostic scores
