Background: In 2010, Temel et al. reported that early palliative care (PC) (EPC) led to significant improvements in quality of life and, surprisingly, on overall survival (OS) over standard oncologic care (SOC) for patients (pts) with metastatic lung cancer. The American Society of Clinical Oncology then recommended “combined standard oncology care and PC earlier in the course of the illness for any patient with metastatic cancer”. The median OS of unselected pts with metastatic upper gastrointestinal cancers (mUGIC) does not exceed 11 m. Whether EPC benefits also apply to pts with mUGIC was unknown. Methods: EPIC (NCT02853474) is a randomized 1:1, open-label, multicenter, phase 3 trial aiming to evaluate the OS benefit of EPC compared to SOC in pts with mUGIC (esophageal, HER2 negative esogastric junctional and gastric, pancreatic, and biliary tract cancers), with ECOG PS ≤ 2, to be treated by first-line chemotherapy (CT). EPC included 5 PC visits scheduled every month, starting within 3 weeks of randomization. Pts allocated to SOC could access to PC visits upon request in case of worsening of health status. The primary endpoint was OS. Other endpoints included quality of life, and depression and anxiety assessment. A total of 381 deaths was required (1 year OS 40% vs. 50.3%, HR = 0.75, 2-sided alpha 5%, 80% power). Randomization was balanced on center, ECOG-PS and tumor site. Results: 473 pts were randomized from 10/2016 to 12/2021 in 19 French centers. Modified intention-to-treat population was 470. Median age = 67.5 (range, 24-90); 273 males; tumor site: esophagus 39, esogastric 48, pancreas 297, biliary tract 86; ECOG PS 0/1/2: 104/278/88. 74/233 pts (32%) in EPC arm had less than the planned EPC visits (major deviation): 3 related to treatment, 17 due to pt decision, 45 due to the clinical team decision, 2 other reasons, 7 unknown. 57/237 pts in the SOC arm got at least 1 PC visit, including 3 major deviation (performed < 6 months of treatment and set for a reason other than worsening of the health condition). Median follow-up was 46 m, and 35/470 pts were still alive. We did not observe any OS difference between arms (EPC, median OS: 7.0 m [95%CI: 6.1-8.8 m] and 1-year OS: 33%; SOC, median OS: 8.6 m [6.8-9.8 m] and 1-year OS: 32%; stratified HR = 1.04 [0.86-1.26], p = 0.68). Results were similar on the per-protocol dataset (excluding major violations). Results were homogeneous across the different tumor locations, and ECOG-PS groups. Conclusions: OS was lower than expected in both arms, which may reflect the difference between pharma-sponsored and academic trials with less stringent exclusion criteria. Also, only 68% of the pts got the planned number of PC visits in the experimental arm. EPC was not found to prolong OS over SOC in pts with mUGIC in this EPIC trial. Clinical trial information: NCT02853474.