Abstract only 11045 Background: Enumeration of CTCs can be used to assess prognosis in MBC and to evaluate treatment response. Besides enumeration, molecular CTC characterization is a promising tool to develop a more personalized treatment approach. Here, we evaluated the association between mRNA expression of currently known CTC-specific genes and response to first-line AI in MBC patients with estrogen receptor (ER)+ primary tumors. Methods: CTCs were isolated and enumerated from blood of 25 MBC patients before first-line therapy with an AI. Fourteen patients received a non-steroidal AI (8 letrozole, 6 anastrozole) and 11 patients were treated with exemestane. mRNA expression levels of 96 genes were measured by quantitative RT-PCR as previously described (Sieuwerts et al. Clin Cancer Res. 17:3600-3618, 2011). Expression levels of these genes were studied for their association with time to progression (TTP) after start first-line AI. Results: Median TTP was 338 (range 14–1,239) days. Median baseline CTC count for the 25 patients was 14 (range 0–753). In this relatively small cohort, the clinically relevant cut-off level of ≥5 CTCs in association with TTP did not reach statistical significance (Hazard Ratio HR 4.76, 95% Confidence Interval CI: 0.59–38.22, P=0.14). For type of AI, when comparing steroidal with non-steroidal AI, the measures in Cox univariate regression analysis were HR 2.54 (95% CI: 0.67–9.64), P=0.17. A 10-gene CTC profile was constructed based on the Wald statistics of the contribution of the individual genes in univariate Cox regression analysis of TTP. To identify patients with good and poor outcome, the Wald corrected sum of the 10 genes was used to dichotomize the continuous 10-gene predictor (HR 12.87 95% CI: 1.60–103.56, P=0.016). In multivariate analysis, corrected for the clinically relevant variables type of AI and CTC count, only the 10-gene CTC profile was an independent factor associated with TTP (HR 12.46 95% CI: 1.29-120.08, P=0.029). Conclusions: A 10-gene CTC predictor was constructed which distinguishes good and poor outcome to first-line AI in MBC patients. This profile is currently being validated in an independent group of patients.