Background: CALGB 80303 evaluated gemcitabine (G) in combination with either B or placebo (P) in 602 patients (pts) with APC. No significant difference in overall survival (OS) between treatment arms was observed. Previously, plasma EDTA samples from CALGB 80303 were analyzed and potential predictive biomarkers for B were identified. This follow-up analysis evaluated citrated plasma samples for circulating proteins related to matrix remodeling and coagulation. Methods: Multiplex ELISA analysis was employed to assess the following circulating factors: E-selectin, matrix metalloprotease (MMP) 2, MMP9, von Willebrand factor (vWF), D-dimer, thrombospondin 1, and tissue factor (TF). Thrombin-antithrombin complex (TAT) was analyzed using a standard enzyme immunoassay. Prognostic characteristics were determined by associating baseline values with OS using a Cox model. Predictive markers were identified with inclusion of a treatment by marker interaction term. Results: Baseline citrated plasma samples from 109 pts were included in this analysis (59 pts G+P; 50 pts G+B). Negative prognostic factors for OS were identified, including E-selectin (p <0.01) for patients receiving G-based therapy, and E-selectin, TF, and vWF (p <0.05) in pts receiving G+P. Potential biomarkers predicting for OS benefit from B were also identified. Pts with low (< median) levels of TAT (interaction p=0.034) exhibited a median OS of 6.0 months (mo) vs. 8.3 mo (HR=0.49; 95%CI 0.28-0.88) when receiving G+P vs. G+B, respectively. Furthermore, pts with elevated levels of TF (>Q3) were found to benefit from B (interaction p=0.021), with median OS of 4.3 mo vs 4.8 mo (HR=0.44;95%CI 0.18-1.08), respectively. Longitudinal analysis of samples collected at baseline and at cycle 3 indicated that only vWF changed in pts receiving G+P, with an increase in vWF (p<0.05). Conclusions: In patients with APC, baseline levels of TAT and TF were identified as potential predictors of benefit from B. These data emphasize the importance of coagulation factors in modulating tumor angiogenesis. The potential predictive value of these factors warrants further validation.