Key lessons from phase I trials in gastroesophageal cancer treated in Gustave Roussy Cancer Center.

321 Background: Gastroesophageal cancer (GEC) treatment has advanced in recent years with the introduction of targeted therapies and immunotherapies. Phase I trials are central to this progress, evaluating the safety and early efficacy of new agents and combinations. However, real-world data on patients (pts) outcomes, prognostic factors, and the proportion of therapies that progress beyond phase I remain limited. Methods: This retrospective study included pts with GEC treated in phase I trials at Gustave Roussy Cancer Center between 2010 and 2025. Pts and disease characteristics, investigational drug combinations, and their subsequent development were described. Clinical outcomes, including progression-free survival (PFS) and overall survival (OS), were evaluated. Results: A total of 210 pts were included. The median age at enrollment in phase I trials was 59.5 years. Most patients were male, with an ECOG performance status of 1 and an RMH score of 1. Overall, 137 (65.24%) pts had gastric adenocarcinoma, 51 (24.29%) had esophageal squamous cell carcinoma, and 22 (10.47%) had esophageal adenocarcinoma. The most frequent metastatic sites were lymph nodes (70.9%) and liver (48.6%). Patients had received a median of 2 prior treatment lines (range, 0–6). The median OS from the diagnosis of metastatic disease was 26.7 months (95% CI, 24.2–29.4). From trial inclusion, the median PFS and OS were 2.2 months (95% CI, 1.9–2.8) and 6.1 months (95% CI, 4.9–7.3), respectively. In multivariable Cox regression, both RMH score at inclusion and the number of previous lines were independently associated with shorter OS ( p<0.05). For PFS, RMH score at inclusion (p=0.028) and liver metastases(p=0.034) were significantly associated with worse outcomes. Severe toxicities leading to permanent treatment discontinuation occurred in 7 pts. Investigational therapies in phase I trials included 31 immunotherapy regimens (mainly immune checkpoint inhibitors (IO) as monotherapy, IO intravenous or intratumoral combinations, IO plus oncolytic virus, or IO plus radiotherapy), 7 bispecific antibodies, 8 antibody–drug conjugates, and 32 other targeted therapies (tyrosine kinase inhibitors and monoclonal antibodies as monotherapies or in combinations). Overall, 18 of these combinations were further evaluated in phase II and/or phase III trials. Conclusions: In this real-world cohort, RMH score and liver metastases emerged as key independent predictors of survival. Phase I trials were generally safe, and many treatments went on to later-phase studies, showing the important role of early trials in improving care for GEC.