Abstract Background: Symptomatic BMinv occurs in a minor proportion of metastatic BC patients (pts). Few data exist regarding its clinical presentation, prognosis and optimal treatment (Tx). ZA has shown to interrupt metastatic “vicious cycle” in bone and to “clear” micrometastastic BMinv in early BC, but its role in symptomatic BMinv remains unclear. We reviewed our series of pts with proven symptomatic BMinv, focusing on Tx delivered and prognosis according to BC subtype and ZA use. Methods: Pts with histologic/cytologic evidence of BMinv from Jan2000 to Feb2012 and with any associated cytopenia were retrospectively identified from Pathology Department files. Results: Twenty-nine pts were identified. Ductal histology: 62%, lobular 24%; Luminal (Lum = HR+/HER2−):76%; HER2+:7%, Triple negative (TN):17%. Stage: II 28%, III 34.5%, IV 24%; median (M) disease free interval: 39.6 months (m) (1.3–113.4). Time BC relapse to BMinv (M): 11.5 m (0–127). At time of BMinv onset: M age 54.7 years (34.3–77.6); M systemic Tx for metastatic disease 2 (0–7). BMinv was present at time of metastatic relapse in 11 pts, including 3 without overt bone metastasis (BM1). Other involved sites at BMinv onset: bone 90% (55% prior to BMinv), nodes 28%, liver 24% and pleuropulmonar 21%. Anemia was the most prominent hematologic sign (97%) followed by thrombocytopenia (Th) (76%) and neutropenia (35%). Tx efficacy in terms of blood count improvement (BCI) and tumor response (TR) outside BM are summarized in table 1. Most pts received non-mielotoxic regimens (endocrine therapy, weekly chemotherapy or capecitabine). ZA was administered in 22 pts (with or without overt BM1). Overall survival (OS)(M) for the whole group was 5.6m (0.3–72.9); M OS in Lum, and TN groups were 9.3 (0.4–72.9) & 0.4m (0.3–20.9+), respectively, while in the 2 HER2+ pts OS was 5.6 & 44.9+ m (p = 0.023). BC subtype, ECOG (0–1 vs 2–3), ZA use after BMinv onset and Th grade (0–1 vs ≥2) were related with OS in the univariate analysis for the whole population. ZA use after BMinv onset, Th grade and presence of BMinv at time of metastatic relapse were also significantly associated with OS in the Lum subgroup. In multivariate analysis, Th remained as an independent factor for OS both in the whole group and Lum subset. ZA use after BMinv onset showed a strong prognostic trend in the Lum group (p = 0.07). Conclusions: BMinv has to be considered in BC pts with BM1 and otherwise unexplained cytopenia. Lum subtype correlates with 9 m M OS, while prognosis in TN subset appears to be particularly dismal. Severe th (<75000 platelets) independently predicts poor OS both in the whole population and in the Lum subtype. In this latter group, a strong trend to better outcome was seen by using ZA after BMinv diagnosis. Futher studies exploring the role of ZA use in pts with BC and symptomatic BMinv are warranted. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P6-13-03.