A phase II study of agenT-797, botensilimab (BOT) and balstilimab (BAL) in PD-1 refractory gastroesophageal cancer (GEC)

Background: GEC is the second leading cause of cancer related mortality. Second line treatment with ramucirumab (RAM) and paclitaxel (PAC) has poor outcomes with an objective response rate (ORR) of 28% and median progression free survival (PFS) of 4.4 months (mo). Methods: This was an investigator initiated trial (NCT06251793) of patients with GEC who received one prior line of therapy evaluating BOT (Fc-enhanced anti-CTLA-4), BAL (anti-PD-1), and agenT-797 (allogeneic invariant natural killer T [iNKT] cells) with RAM/PAC. The primary endpoint was ORR. Secondary endpoints were safety, PFS, and overall survival (OS). Serial tissue and blood were collected throughout treatment for multiplex immunofluorescence (IF), flow cytometry, T cell receptor (TCR) sequencing, and cytokine analysis. Results: From Feb 2024 to Apr 2025 17 patients were enrolled: 2 received an induction cycle of agenT-797, 5 received induction with agenT-797/BOT/BAL, and 10 initiated all therapies simultaneously. All had received prior anti-PD-1/fluoropyrimidine/platinum containing regimens. Median follow up was 20.7mo (range 11.9 - 22.2mo). Of 15 patients with measurable disease, 0 (0%) had a complete or partial response, 11 (73%) had stable disease, and 4 (27%) had progressive disease as best response. Median PFS and OS were 4.4mo (95% CI 3.4 - 6.9mo) and 7.7mo (95% CI 4.9 - NR) respectively. Three patients had an OS ≥ 20mo, including one who remains progression free after 22.5mo. Patients treated with an induction cycle had superior PFS compared to those who started all therapies together (6.9 v 3.5mo, p=0.008). There was no difference in those with liver metastases versus those without (4.0 v 4.3mo, p=0.4). 17 (100%) patients had an adverse event (AE); 9 (53%) were grade (G) 3+. The most common were fatigue (82%), fever (59%), and diarrhea (47%). Nine (53%) patients had an immune related AE; 4 (24%) were G3+. The most common were colitis (24%), dermatitis (24%) and gastritis (18%). Analysis of on-treatment tissue (n=10) showed increased infiltration of proliferating CD8 + T cells (3.3 fold change [FC] p=0.037) and activated antigen presenting cells (5.7 FC p=0.006) compared to baseline samples. Peripheral blood analysis showed increased CD8+ effector memory T cells (p=0.022) and IFNgamma (p=0.001), a hallmark of iNKT cell activity. Conclusions: This is the first study to evaluate agenT-797/BOT/BAL in GEC. Although robust systemic and intratumoral immune activation was observed, ORR and median PFS/OS were not improved compared to historical controls. Long term survival in a subset of patients and superior PFS in the induction cohort suggest patient selection and therapeutic sequencing are key determinants of benefit. Ongoing mIF, flow, and TCR analyses of all 17 patients and comparisons of these subgroups are in process and may inform sequencing and patient selection, supporting further investigation of these therapies.