PURPOSE: Pancreatic ductal adenocarcinoma upregulates CD73, potentially contributing to immune surveillance evasion. Combining oleclumab (CD73 inhibitor) and durvalumab with chemotherapy may identify an effective treatment option. PATIENTS AND METHODS: We describe a multicenter phase Ib/II randomized clinical trial in patients with metastatic pancreatic ductal adenocarcinoma, untreated (cohort A) or previously received gemcitabine-based chemotherapy (cohort B; NCT03611556). During escalation, patients received oleclumab 1,500 or 3,000 mg, durvalumab 1,500 mg, and gemcitabine plus nab-paclitaxel (GnP; cohort A; n = 14) or modified FOLFOX (cohort B; n = 11). During expansion, cohort A patients (n = 170) were randomized to GnP (arm A1), oleclumab [recommended phase II dose (RP2D)] with GnP (arm A2), or oleclumab (RP2D) with durvalumab plus GnP (arm A3). Primary objectives were safety (escalation) and objective response rate (expansion). Secondary objectives included progression-free survival (PFS) and overall survival (OS). RESULTS: During escalation, 1/11 patients from cohort B (oleclumab 3,000 mg) experienced two dose-limiting toxicities. Oleclumab's RP2D was 3,000 mg. During expansion, grade ≥3 treatment-related adverse events occurred in 67.7% (42/62) of patients in A1, 73.7% (28/38) in A2, and 77.1% (54/70) in A3. The objective response rate was 29.0%, 21.1%, and 32.9% in A1, A2, and A3, respectively (A1 vs. A3; P = 0.650). PFS [HR = 0.72; 95% confidence interval (CI), 0.47, 1.11] and OS (HR = 0.75; 95% CI, 0.50-1.13) were similar for A3 versus A1. Patients with high CD73 expression had improved PFS and OS in A3 versus A1, although this should be interpreted with caution. CONCLUSIONS: Although the safety profile was acceptable, this study did not meet its primary efficacy endpoint.