Background: In vivo preclinical models, transforming growth factor-beta (TGFβ) inhibitors enhanced gemcitabine activity. In this study, patients (pts) with Stage II to IV unresectable PC were treated with GG or GP to evaluate their efficacy and safety. Methods: Pts were randomized 2:1 (stratified for ECOG PS, disease stage and previous gemcitabine use) to GG or GP. Oral galunisertib 150 mg BID was given as intermittent dosing (14 days on/14 days off per cycle); gemcitabine was administered per label. The primary endpoint was overall survival (OS). Using a Bayesian augmented control (BAC) design assuming a hazard ratio (HR) of 0.7 for OS, the study had 90% power to identify if the probability of HR < 1 was > 0.85. Secondary endpoints included progression-free survival (PFS), overall response rate (ORR), biomarkers and safety (CTCAE v4). Results: 156 pts were randomized to GG (N = 104) or GP (N = 52). Baseline characteristics were balanced between groups. HR based on BAC was 0.8 with 95% credible interval (0.60, 1.09) and PR (HR < 1) = 0.92, indicating a positive study. Efficacy data are summarized in the table. For the subgroup of pts with baseline TGFβ1 levels ≤ 4224 pg/mL (n = 117), median OS was 10.9 vs 7.2 months (GG vs GP) with unadjusted HR (95% CI) = 0.68 (0.44, 1.04), logrank p = 0.076). The most frequent CTC grade 3/4 adverse events possibly related to study treatment (GG vs GP) were anemia (7.8% vs 13.5%), neutropenia (32.0% vs 26.9%) and thrombocytopenia (7.8% vs 9.6%). 46% vs 43% of pts had > 50% decrease in CA19-9 and 43% vs 41% had > 50% decrease in TGFβ1 (GG vs GP). For both biomarkers a reduction in concentration correlates with improved OS. Conclusions: GG resulted in improvement of OS and PFS in pts with PC, with a manageable toxicity profile as compared to GP. TGFβ1 and CA19-9 reduction correlated with OS. Pts with lower levels of TGFβ1 may have greater benefit from treatment with galunisertib. Clinical trial information: NCT01373164