Safety and feasibility of Claudin18.2 (CLDN18.2)-specific CAR-NK cells in advanced gastrointestinal cancers: a phase 1 trial

Technical Summary: Safety and Feasibility of CLDN18.2-Specific CAR-NK Cells in Advanced Gastrointestinal Cancers

Problem Statement
Despite advances in immunotherapy, outcomes for patients with advanced solid tumors remain suboptimal. While Chimeric Antigen Receptor (CAR) T-cell therapy has achieved success in hematological malignancies, its application in solid tumors faces significant hurdles, including lengthy autologous production timelines, immunosuppressive tumor microenvironments (TME), antigen heterogeneity, and T-cell exhaustion. Furthermore, CAR-T therapies are often associated with severe toxicities such as Cytokine Release Syndrome (CRS) and Immune Effector Cell-Associated Neurotoxicity Syndrome (ICANS). Consequently, there is an unmet clinical need for "off-the-shelf" allogeneic cellular therapies that offer a favorable safety profile and potent anti-tumor activity. While CAR-NK cells have emerged as a promising alternative due to their lower incidence of CRS, lack of MHC restriction, and potential for allogeneic use, their clinical application in advanced gastrointestinal cancers, specifically targeting Claudin 18.2 (CLDN18.2), had not been previously reported.

Methodology
This study was a phase 1, single-arm, open-label trial conducted at Zhejiang Provincial People's Hospital and the Second Affiliated Hospital of Zhejiang University School of Medicine (ClinicalTrials.gov: NCT06464965).

• Cohort: Nine patients with refractory advanced gastric cancer (GC) or pancreatic adenocarcinoma (PC) were enrolled. All patients were CLDN18.2-positive (confirmed via IHC), had an ECOG performance status of 0 or 1, and had received at least one prior line of systemic therapy.
• Intervention: Patients received lymphodepleting chemotherapy (cyclophosphamide and fludarabine) followed by three intravenous infusions of CLDN18.2-BBz CAR-NK cells on days 0, 3, and 7. The CAR-NK product was engineered from umbilical cord blood-derived CD3- mononuclear cells, transduced with a lentiviral vector encoding a second-generation CAR containing a CLDN18.2-specific scFv, a 4-1BB co-stimulatory domain, and IL-15.
• Dosing: Dose escalation was performed across three cohorts: 2 × 10⁶, 8 × 10⁶, and 1.6 × 10⁷ CAR-positive NK cells per kg.
• Endpoints: The primary endpoints were safety and tolerability (Maximum Tolerated Dose and Dose-Limiting Toxicities). Secondary endpoints included efficacy (ORR, DCR, PFS, OS), pharmacokinetics (CAR-NK expansion and persistence), and pharmacodynamics (serum cytokines).
• Mechanistic Analysis: To understand immune remodeling, the study employed single-cell RNA sequencing (scRNA-seq) on peripheral blood mononuclear cells (PBMCs) from three representative patients (pre- and post-infusion) and utilized CellChat analysis to map cell-cell communication networks. Digital droplet PCR (ddPCR) was used to quantify CAR transgene persistence.

Key Results

• Safety: The treatment demonstrated a favorable safety profile. No dose-limiting toxicities (DLTs), CRS, ICANS, or graft-versus-host disease (GvHD) were observed. The most common grade ≥3 adverse events were myelosuppression (leukopenia, neutropenia) attributed to the lymphodepleting conditioning regimen. No grade ≥3 gastrointestinal toxicities occurred.
• Efficacy: No objective tumor responses (partial or complete responses) were observed. However, the Disease Control Rate (DCR) was 88.9% (8/9 patients), with all responders achieving Stable Disease (SD). One pancreatic cancer patient (Patient 3) maintained durable SD for 11 months despite receiving the lowest dose. The median Progression-Free Survival (PFS) was 2.0 months, and the median Overall Survival (OS) was 11.0 months.
• Pharmacokinetics: CAR-NK cells exhibited transient in vivo expansion, peaking at a median of 3 days post-infusion. While robust expansion (>30%) was observed in two patients, persistence was generally limited, with CAR+ cells becoming undetectable (<0.1%) by day 14 in most patients. However, ddPCR confirmed low-level, long-term CAR transgene persistence in patients with durable disease control (e.g., detectable at day 270 in Patient 3).
• Immunological Profiling: scRNA-seq revealed profound remodeling of the peripheral immune landscape post-infusion. There was a marked expansion of monocytes/macrophages and effector memory CD8+ T cells (CD8_TEM), alongside a reduction in T cells. While CD8+ T cells showed increased activation and cytotoxicity signatures, they also exhibited signs of exhaustion. CellChat analysis indicated that CD8_TEM cells established new interactions with monocytes, which recruited them into the network but simultaneously delivered inhibitory signals.

Significance and Claims
The authors claim this study represents the first-in-human clinical investigation of CLDN18.2-targeted CAR-NK cell therapy for advanced gastrointestinal malignancies. The primary significance lies in demonstrating that allogeneic CAR-NK cells can be administered safely to heavily pretreated patients without the severe toxicities (CRS, ICANS, GvHD) often associated with CAR-T therapies.

While the study did not achieve objective tumor regression, the authors posit that the high rate of disease stabilization (88.9% DCR) and the durable SD observed in specific patients indicate tangible biological activity. The paper suggests that the modest efficacy compared to CLDN18.2 CAR-T trials may be attributed to lower proportions of high CLDN18.2 expression in the cohort and the limited in vivo persistence of CAR-NK cells. Furthermore, the single-cell analysis provides mechanistic insight, suggesting that while CAR-NK therapy induces immune activation, it is counteracted by myeloid-derived inhibitory signals and subsequent T-cell exhaustion. The authors conclude that while CLDN18.2 CAR-NK cells are a promising off-the-shelf alternative with a unique safety landscape, future strategies must focus on enhancing CAR-NK persistence and overcoming TME-mediated immunosuppression to improve long-term efficacy.