Metastatic breast cancer subtypes and central nervous system metastases.

e11581 Background: The relapse pattern, survival and response to therapy are known to be different between breast cancer (BC) subtypes defined by combining hormone-receptor (HR) and HER2 status. Our aim was to study incidence and predictors of central nervous system metastases (CNS-M) and the outcome after CNS-M according to tumor subtype. Methods: 488 patients (pts) treated with at least one line of chemotherapy for metastatic BC were retrospectively evaluated. According to the combination of HR and HER2 status, tumors were grouped in: Luminal (Lum): HR+/HER2-, Luminal/HER2+ (Lum/HER2+): HR+/HER2+, pure HER2 positive (pHER2+): HR-/HER2+, and triple negative (TN): HR-/HER2-. All HER2+ patients received treatment with Lapatinib or Trastuzumab in addition to chemotherapy for metastatic disease. Median follow up was 34 months. Results: 133 pts (27%) developed CNS-M, with a median time to CNS progression of 43 months. The rate of CNS-M by subtype was: Lum 18%, Lum/HER2+ 37%, pHER2+ 49%, TN 25% (p <0.001). Multivariate analysis confirmed that, compared with Lum tumors, Lum/HER2+ ( HR 2.556, p<0.001), pHER2+ (HR 4.444, p<0.001) and TN (2.249, p=0.011) subtypes were at higher risk of CNS-M. Median overall survival (OS) CNS-M was 8.8 months in the whole series (IC 95% 6.6-11.0). Median OS in months by subtype was: Lum 9, Lum/HER2+ 18, pHER2+ 7, TN 7 (p<0.001). Multivariate analysis revealed that belonging to the Lum/HER2+ subtype (HR 0.528 compared with the Lum subtype, p<0.001) and having isolated CNS (HR 0.398, compared with CNS-M plus systemic progression, p<0.001) predicted significantly reduced risk of death. Conclusions: Among pts with a known increased risk of brain metastases, the Lum/HER2+ subtype appears associated with the longest OS after CNS-M, probably due to different biology and better extracranial disease control by chemotherapy, hormonal therapy and target agents. These results suggest that these patients may benefit from a more aggressive treatment of CNS-M and, possibly, from the screening for asymptomatic CNS lesions.