3514 Background: Chimeric antigen receptor (CAR) T cell therapy has been successful in hematologic malignancies but not in solid tumors. Homing capacity and tumor microenvironment are key issues to limit CAR-T efficacy to solid tumors. Our previous study has shown that hypoxia-responsive CAR-T cells exhibit reduced exhaustion and enhanced efficacy in solid tumors (Cancer Res 2024). Here, we aimed to optimize the administration route by comparing safety and efficacy of our newly established CEA CAR-T through intraperitoneal (I.P.) and intravenous (I.V.) infusions in solid tumor patients who failed to standard therapy. Methods: In this phase I, open-label, dose-escalation clinical trial, eligible patients were CEA positive and relapsed or refractory to at least second-line treatment. CAR-T cells was administered via I.P. or I.V. with dose escalation including low dose of 1-3 x106 CAR+/kg and high dose of 4-6 x106 CAR+/kg. Adverse events were graded by CTCAE 5.0. Tumor responses were evaluated using RECIST 1.1. Results: Between June 2022 to Jan 2024,40 pts were enrolled and infused CAR-T cells, including 35 CRC pts, 3 GC pts, 1 NSCLC pts and 1 BTC pts. There were 16 pts in the I.P. group and 24 pts in the I.V. group. CRS was observed in 62.5% of the pts at grade 1-2, and no ICANS or treatment related death was reported. Grade 1-2 mucositis was reported in 25% of the patients. Immune-mediated diarrhea and colitis of all grades occurred in 32.5% of the patients, including 17.5% grade 3 events. And 100% of patients experienced grade 3-4 hematologic toxicity. In terms of efficacy, the I.P. group displayed a higher objective response rate (ORR) of 25% (4/16), compared to 8% (2/24) in the I.V. group, alongside a superior disease control rate (DCR) of 88% (14/16) in I.P. compared to 67% (16/24) in I.V. group. Notably, within the high-dose group receiving I.P. infusion, the ORR increased to 28.5%. Moreover, the responded patients in the high-dose group all had sustained tumor remission for more than 5 months, including one patient who had a 76% target lesions regression at six months follow up. CAR-T cells demonstrated robust expansion in both the I.P. and I.V. groups, while no significant difference was found in maximum concentration. Conclusions: The CEA CAR-T cell therapy had acceptable toxicity via I.P. or I.V. infusion. We have observed a promising antitumor potential by I.P. infusion and a prolonged tumor remission in high-dose group. Further investigation is needed to highlight these findings and the corresponding mechanisms. Clinical trial information: NCT05396300 . [Table: see text]