Hypoxia-responsive CEA CAR-T cells therapy for relapsed or refractory non-small cell lung cancer: A single-arm, open-label, phase I trial.

8517 Background: Non-small cell lung cancer (NSCLC) is a leading cause of cancer-related mortality worldwide. Chimeric antigen receptor (CAR)-T cell therapy has achieved significant success in targeted tumor cells eradication. However, data on CAR-T therapies in NSCLC remain limited. This study evaluates the safety and efficacy of CEA CAR-T cell therapy in r/r NSCLC. Methods: Adult metastatic NSCLC patients with relapse after ≥2 lines of treatment were enrolled and administered intravenous CEA CAR-T cells at three dose levels (0.75×10 6 , 1.5×10 6 , and 3×10 6 CEA CAR-T cells/kg).Peripheral blood samples were collected on day 28 post-infusion and analyzed using 10x Genomics single-cell RNA sequencing (scRNA-seq).The primary endpoint was safety , and secondary endpoints including efficacy and pharmacological evaluation. Results: From August 1, 2023, to July 15, 2024, a total of 18 patients were screened, and 15 received CAR-T infusion. The median age was 60 years , with a median of 4 prior therapy lines (range 2-10 lines). Among 6 patients receiving the maximum dose (3×10 6 cells/kg), no dose-limiting toxicities, grade 4 cytokine release syndrome or ICANS were observed. No adverse events were observed during a three-month long-term safety evaluation. With a median follow-up of 5.7 months, 7 patients achieved PR, 6 had SD, and 2 experienced PD. The best DCR was 87%, and ORR was 47%. Notably, patients with ≥30% intense and complete CEA staining in tumor cells and no brain metastases showed better PFS (72.7% vs. 25.0%, p=0.03) and OS (90.9% vs. 0%, p=0.003). CEA CAR-T cells reached maximum concentration at a median of 10 days (range 7–60 days) post-infusion. At the third month, all 12/12 patients with available CAR-T cell copy number data maintained high levels of CAR-T cells, with a median of 8,236 gDNA copies/μg , and in the patient with a follow-up of 13 months, CEA CAR-T cells were still detectable at 50,760 gDNA copies/μg. The scRNA-seq was performed in 12 patients. Responders exhibited a higher percentage of NK cells in a less exhausted state, characterized by reduced activity of immunosuppressive pathways and lower expression of stress-associated genes. Further cell-cell communication analysis suggested HLA-DRB1 expression in NK cells might be influenced by interactions with the CD244 in CD8⁺ T cells. Conclusions: A single infusion of hypoxia-responsive CEA CAR-T demonstrated promising efficacy and manageable safety in r/r NSCLC. The findings highlight the potential role of specific NK cell states and immune interactions in the therapeutic effects of CEA CAR-T therapy. Clinical response at different doses. All dose level(n=15) Dose level 1(n=3) Dose level 2(n=6) Dose level 3(n=6) Disease control 13 (87%) 2 (66%) 6 (100%) 5 (84%) Overall response 7 (47%) 1 (33%) 4 (67%) 2 (33%) Partial response 7 (47%) 1 (33%) 4 (67%) 2 (33%) Disease stable 6 (40%) 1 (33%) 2 (33%) 3 (50%) Data are n (%).