P4-17-09: Comparative Performances of Prognostic Indexes for Breast Cancer Patients Presenting with Brain Metastases.

Abstract

Abstract Introduction: Incidence of brain metastases (BM) from breast cancer has increased over the past decade, especially for HER2(+) tumours. Several scores and prognosis systems have been developed in the brain metastases (BM) setting in order to help physicians tailor treatment options depending on patient prognosis and to stratify patients enrolled in clinical studies. The aim of our study was to compare the clinical relevance of the major classifications and existing prognostic scores in a population of breast cancer patients affected by BM. Methods and Materials: In this retrospective study conducted in Montpellier and Nice Cancer Centres, we retrospectively reviewed the clinical and biological data of 250 patients diagnosed with breast cancer BM between 1995 and 2010. Prognostic value and accuracy of recursive partitioning analysis (RPA), graded prognostic assessment (GPA), basic score for brain metastases (BS-BM) and a clinico-biological score (BS) developed in phase I study and validated in the BM setting were compared. Results: Median age at BM diagnosis was 55 years-old. Most patients (74%) had good performance status (0-1). Brain was the first metastatic site in one third of patients (33.6%). In 12.4% of cases, no extracerebral metastases were detected at BM diagnosis. Among the 250 patients’ tumors, 43.6% overexpressed HER2 receptor and 25.6% were triple negative (negative estrogen receptor, progesterone receptor and HER2 status). After a median follow up of 4.5 years, median overall survival from BM diagnosis was 8.9 months (CI 95%, 6.9−10.3 months). The four scores were able to discriminate patients according to their survival. Chi2 tests showed a correlation between the different scoring systems. In multivariate analysis, the elimination model identified RPA (p = 0.031; hazard ratio, 0.77; 95% confidence interval, 0.64−0.91) and BS (p = 0.043; hazard ratio, 0.5; 95% confidence interval, 0.31−0.80) as the only two independent predictors of survival. RPA was the most accurate score in order to identify patients with high life expectancy, while RPA and BS were the most accurate scores to classify patients with short life expectancy. Conclusions: In a general unselected population of breast cancer affected by BM, RPA seems to be the most useful score. Integration of biological parameters in addition of existing clinico-radiological scores is promising in order to improve the prognostic determination accuracy. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P4-17-09.