Phase 1 study of LB1908, an autologous claudin 18.2-targeted CAR-T cell product, in subjects with advanced gastroesophageal adenocarcinoma.

374 Background: CLDN18.2 is a therapeutic target on gastric, gastroesophageal, and esophageal adenocarcinomas (GC/GEJC/EC) with normal expression limited to the gastric epithelium. We present interim results of a dose-escalation trial of LB1908, a CLDN18.2-targeted autologous chimeric antigen receptor (CAR)-T cell product in adults with advanced GC/GEJC/EC. Methods: LB1908-1001 is an ongoing, multicenter, phase 1 study of LB1908 in subjects with advanced GC/GEJC/EC, relapsed or refractory to ≥1 prior line of therapy (LOT), with ≥1+ CLDN18.2 expression in ≥50% of tumor cells. A 3+3 design evaluated 3 dose levels (DLs): 0.5, 1.5, and 3 × 10 6 CAR-T cells/kg. Subjects receive a single infusion of LB1908 after standard lymphodepletion (LD). The primary objective is to evaluate safety and dose-limiting toxicities (DLTs). Results: As of 28 Aug 2025,13 subjects (median 56 yrs) received LB1908 (6 at DL1; 3 at DL2; 4 at DL3): 6 with EC, 4 GC, and 3 GEJC (median 2 prior LOTs, range 1–8). Median tumor CLDN18.2 expression was 95% and H-score was 245. Eleven subjects received bridging therapy between apheresis and LB1908 infusion. The most common grade ≥3 TEAEs were hematologic and attributed to the LD regimen. The only non-hematologic LB1908-related TEAEs grade ≥3 occurring in ≥2 subjects were gastritis (n=3) and gastric mucosal lesion (n=2), including 1 DLT at DL1. After this DLT mitigation guidelines were introduced (prophylactic enteral beclomethasone and moderate-dose systemic steroids). In the 6 subjects treated since implementing prophylaxis, no DLTs have been observed and upper GI toxicity has been manageable. Ten subjects (76.9%) experienced low-grade cytokine release syndrome (CRS; 4 grade 1 and 6 grade 2), at a median 6 days post-infusion. Four subjects received tocilizumab; no steroids were required to treat CRS. There were no cases of ICANS or TEAE-related deaths. CAR-T cells expanded in all subjects (median C max , 1594 and 2211 copies/µg genomic DNA; median T max , 14 and 20 days at DL1 and DL2, respectively). There was no clear dose-exposure relationship based on DL1 and DL2 data. Of 10 response-evaluable subjects, 9 (90%) had tumor shrinkage (5/6 at DL1, 3/3 at DL2, 1/1 at DL3). Three subjects had RECIST-defined partial responses (1 at each DL). At DL2, all 3 subjects achieved disease control, which correlated inversely with CD4:CD8 ratio in the drug product (R 2 =0.8). Six subjects maintained disease control ≥50 days, including >10 months in the subject with the longest follow up. Response and PK data from DL3 are incomplete as of the data cut. Conclusions: LB1908 demonstrated expansion and anti-tumor activity across DLs in heavily pretreated subjects with advanced GC/GEJC/EC. Prophylaxis with beclomethasone and systemic steroids reduced the on-target/off-tumor toxicity of gastritis/gastric mucosal injury. Dosing at DL3 is ongoing to determine the optimal dose for expansion. Clinical trial information: NCT05539430 .