Abstract Background RANK (receptor activator of nuclear factor kappa B) is a tumor necrosis factor receptor family protein that is expressed on the surface of osteoclasts and critical to bone turnover. RANK is also expressed on breast cancer (BC) cells, especially in the metastatic setting. Pre-clinical mouse models have shown that the ligand to RANK (RANKL) can induce migration of RANK-expressed tumor cells to bone and lung, and that treatment with a RANKL inhibitor can reduce BC metastases. ISPY is a multicenter neoadjuvant study with well annotated gene expression data. We evaluated RANK expression in primary tumor tissues at the time of diagnosis (dx) in patients (pts) with BC and correlated expression with risk of subsequent metastases. Methods: We evaluated expression of genes in the RANK/RANKL pathway in core biopsies at dx, and correlated this data with risk of recurrence as well as risk of site specific metastases using the student t-test. Results: A total 221 pts were enrolled in this clinical trial, all had locally advanced BC (LABC) and received neo-adjuvant chemotherapy (96.6% pts received adriamycin + cyclophosphamide-based regimens), tumor samples were collected at the time of dx by core biopsy, and underwent gene array study. Total 38508 genes were tested and 149 pts had available gene array data. Average age was 48.0 yr, pt characteristics are listed in the Table. Median follow up was 1260.3 days, range from 175 to 2520 days. Thirty-six pts developed recurrent disease; 13 pts developed bone-dominant metastases (BDM), 15 pts developed non-bone metastases (liver, lung, brain, lymph node and contralateral side breast) (NBDM), and 8 pts were not evaluable. Median RFS for those with BDM was 804 days, and for NBDM was 550 days, not significantly different (p=0.07). When comparing gene expression in the RANK/RANKL pathway, pts who developed BDM had a significantly higher expression of RANK (p=0.05) in primary tumor tissues than those with NBDM, while there was no difference in RANKL expression between the 2 groups. In addition, RANK was also significantly more highly expressed in the ER- group than ER+ group (p=0.006) and in the basal group compared to the luminal A/B group (p= 0.002). Analysis of variance (ANOVA) was then used to adjust clinical variability, after adjusting the difference of ER status, pathology subtypes, and menopausal status, RANK expression is even more significantly correlated with the bone-dominant group (p=0.002). Conclusion: Higher RANK expression in primary tumor tissues correlates with increased risk of subsequent BDM in pts enrolled in iSPY-1 trial. The sample size is small (n=149) and all pts had LABC. Our study suggests that targeting RANK/RANKL pathway is a promising strategy in preventing BDM in pts with high risk BC, further analysis in larger data sets is ongoing. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P5-14-04.