Abstract Background Prediction of survival for metastatic breast cancer (MBC) patients (pts) remains a major clinical challenge. Most studies proposed prognostic scores based on clinical criteria that are often subjective. We already demonstrated that low lymphocyte count is a prognostic factor in various pts populations, and we recently showed that low CD4 lymphocyte count is a prognostic factor for overall survival (OS) in MBC pts. In this study, we evaluate the plasma levels of various cytokines and chemokines to improve our immune-based prognostic model. Methods: The first cohort (A) consisted of 39 pts with MBC treated with first line chemotherapy between Sep. 04 and Oct. 07. The second cohort (B) consisted of 114 pts with MBC who relapsed after at least one line of chemotherapy between Dec. 2000 and Nov. 05 and who received further chemotherapy. In the cohort A the blood samples were drawn before administration of any chemotherapy. In the cohort B the samples were drawn after white blood cell recovery. Fresh cells have been used for extensive phenotypic analyses. Plasma cytokines levels have been measured using commercially available Luminex-based multiplex kits. Cytokines were clustered into 3 groups based on biological pathways, group 1: Th1 response and T-cell proliferation (IL-2, IL-7, IL-15, IFNγ, IL-12p40, IFNa2, GM-CSF); group 2: Th2 response (IL-10, IL-13, CCL22), and group 3: inflammatory response (IL-1b, IL6, IL-17, TNFα). Results: In the cohortA: CD4+ T cell levels <450/μL were associated with worse OS (HR=2.46 [CI95%=1.21−5.01), p=0.013). In the cohort B, 48% of pts had received one previous line of chemotherapy; 52% had received more than one. CD4+ T cell levels <450/μL were also associated with worse OS (HR=1.70 [CI95%=1.04−2.78], p=0.036). Analyzing cytokines by clusters, elevation in group 1 (Th1 response) was associated with poor OS for both cohorts A (HR=1.055 [CI95%=1.002−1.111), p=0.041) and B (HR=1.12 [CI95%=1.01−1.24), p=0.032). Conversely, the group 2 (Th2 response) was not associated with OS for both cohorts. For the group 3 (inflammatory response), the increase of these cytokines values was a prognostic factor of OS only for the cohort B (HR=1.10 [CI95%=1.02−1.18), p=0.009). In multivariate analysis: in the cohort A, CD4+ T cell levels <450/μL (HR=2.45 [CI95%=1.20−5.03), p=0.014) and group 1 of cytokines (HR=1.055 [CI95%=1.004−1.109), p=0.034) remains independent prognostic factors; in the cohort B, poor Performance Status (HR=3.10 [CI95%=1.99−4.86), p<0.0001) and group 3 of cytokines (HR=1.09 [CI95%=1.01−1.17), p=0.020) were shown to be independent prognostic factors. Conclusions: Combination of PS and biological covariates such as lymphocyte CD4+ count or cluster of cytokines is an effective strategy to predict survival of pts with MBC receiving first-line chemotherapy or subsequent lines. The validation of our immune-based prognostic score (combining cytokine levels and immune cell phenotypes) will be initiated in a prospective study. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P4-09-10.