The claudin-18.2-CD3 bispecific antibody IBI389 exerts an antitumor effect on malignant tumors with positive claudin-18.2 expression by promoting T-cell infiltration and remodeling the tumor immune microenvironment

Introduction

Gastric cancer remains one of the leading causes of cancer-related mortality worldwide, ranking third in terms of cancer-associated death and posing a major clinical challenge due to its aggressive nature and limited treatment options (1,2). Despite advances in targeted therapies such as trastuzumab for HER2-positive disease and the recent integration of immune checkpoint inhibitors, the overall survival of patients with advanced or metastatic gastric cancer rarely exceeds 1 year (3,4). Clinical trials including CheckMate 649 and KEYNOTE-062 have demonstrated that only a subset of patients—particularly those with programmed death-ligand 1 (PD-L1) positivity or microsatellite instability—derive durable benefits from immunotherapy (5,6). Consequently, identifying new molecular targets and therapeutic strategies to extend survival in the broader gastric cancer population remains an urgent priority.

Claudins are integral tight junction proteins that regulate epithelial barrier function and cell polarity (7). Among them, claudin-18.2 (CLDN18.2) has recently emerged as a promising therapeutic target due to its limited expression in normal gastric mucosa and aberrant exposure on the surface of malignant cells during carcinogenesis (8,9). Preclinical and clinical studies have identified CLDN18.2 as a potential pancancer target, and monoclonal antibodies (mAbs) such as zolbetuximab (IMAB362) have advanced to phase III clinical evaluation (10,11). However, reports on the expression pattern of CLDN18.2 in gastric cancer have been inconsistent, with some studies suggesting high positivity in primary and metastatic sites but others indicating a loss of expression during malignant progression (12,13). Moreover, the functional role of CLDN18.2 in tumorigenesis and its impact on the tumor microenvironment remain poorly understood.

Bispecific antibodies represent an innovative therapeutic modality that can simultaneously engage tumor-associated antigens and immune effector molecules (14). CLDN18.2-CD3 bispecific antibodies are designed to mediate the interaction between gastric cancer cells and T lymphocytes, thereby redirecting cytotoxic T cells into the tumor microenvironment and overcoming immune resistance. In contrast with conventional mAbs or immune checkpoint inhibitors, this approach holds the potential to convert immunologically “cold” tumors into “hot” tumors, thereby broadening the population that may benefit from immunotherapy. Despite its promise, there is a paucity of mechanistic studies examining how CLDN18.2-CD3 bispecific antibodies modulate immune cell composition, spatial interactions, and systemic immune responses in vivo.

Therefore, in this study, we investigated the pharmacological efficacy and immunological effects of a CLDN18.2-CD3 bispecific antibody (IBI389) using both humanized CD3 mouse models and patient-derived tumor samples. Through a combination of flow cytometry, histopathology, cytokine profiling, and multiplex immunofluorescence, we sought to delineate how IBI389 shapes the tumor immune microenvironment (TIME), alters immune cell spatial distribution, and affects systemic immune responses. By providing insights into the mechanisms underlying the interplay between CLDN18.2 targeting and T-cell activation, we hope to establish a rationale for the clinical application of CLDN18.2-CD3 bispecific antibodies and to inform potential combination strategies with existing immunotherapies. We present this article in accordance with the ARRIVE and MDAR reporting checklists (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0474/rc).

Methods

Cell culture and establishment of stable cell lines

The human gastric carcinoma cell line AGS and murine pancreatic cancer cell line KPC, both exhibiting low endogenous CLDN18.2 expression, were acquired from Sunncell Biotechnology Co., Ltd. (cat. No. SNL-103; Wuhan, China) and Immocell Biotechnology Co., Ltd. (cat. No. IM-M167; Xiamen, China), respectively. These cells were cultured in RPMI-1640 for AGS or Dulbecco’s modified Eagle medium (DMEM) for KPC, with the media being supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin, under standard conditions of 37 ℃ in a 5% CO₂ humidified atmosphere. To generate stable CLDN18.2-expressing cell models, both lines were transduced with a lentiviral construct, pLenti-CMV-CLDN18-GFP-Puro (cat. No. PPL01971-4a; Public Protein/Plasmid Library), containing a 786-bp human CLDN18 insert (reference sequence: NM_001002026.3) fused in-frame to a C-terminal copepod green fluorescent protein (CopGFP) reporter. The correct insertion and reading frame were verified via Sanger sequencing. Following transduction, stable pools were selected via puromycin at a concentration of 2 μg/mL, yielding the following cell lines: AGS-CLDN18.2, AGS-Empty, KPC-CLDN18.2, and KPC-Empty. Successful overexpression of CLDN18.2 was confirmed through Western blot analysis and immunofluorescence staining, which assessed both protein expression and subcellular localization.

Western blotting

For protein analysis, cells were lysed using radioimmunoprecipitation buffer containing both protease and phosphatase inhibitors. The total protein concentration was determined with a bicinchoninic acid assay kit (Thermo Fisher Scientific, Waltham, MA, USA). Equal quantities of protein (ranging from 20 to 40 µg per lane) were separated electrophoretically on 10% sodium dodecyl sulfate-polyacrylamide gels and subsequently transferred to polyvinylidene fluoride membranes (MilliporeSigma, Burlington, MA, USA). Membranes were blocked for 1 hour at ambient temperature with 5% nonfat milk prepared in tris-buffered saline with Tween 20 (TBST). They were then probed with primary antibodies targeting CLDN18.2 (dilution 1:1,000; cat. No. ER1902-86; Hua’an Biotechnology Co., Ltd., Hangzhou, China) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (dilution 1:1,000; cat. No. YA874; MedChemExpress, Monmouth Junction, NJ, USA), with the latter serving as a loading control, during an overnight incubation at 4 ℃. After being washed three times with TBST, the membranes were exposed to horseradish peroxidase (HRP)-linked secondary antibodies (dilution 1:3,000; Cell Signaling Technology, Danvers, MA, USA) for 1 hour at room temperature. Signal detection was carried out with an enhanced chemiluminescence (ECL) substrate (Bio-Rad Laboratories, Inc., Hercules, CA, USA), and band intensities were quantified through densitometry with ImageJ software (United States National Institutes of Health, Bethesda, MD, USA).

Immunofluorescence analysis

For immunofluorescence assays, cells were plated onto sterile glass coverslips within 24-well plates and cultured overnight to facilitate attachment. Following two washes with phosphate-buffered saline (PBS), the samples were fixed with 4% paraformaldehyde at room temperature for 15 minutes and subsequently permeabilized with 0.1% Triton X-100 in PBS for 10 minutes. To minimize nonspecific antibody interactions, blocking was performed with 5% bovine serum albumin (BSA) dissolved in PBS for 1 hour at ambient temperature. The cells were then treated overnight at 4 ℃ with a primary antibody targeting the Flag epitope (dilution 1:1,000; cat. No. 20543-1-AP; Proteintech, Rosemont, IL, USA). After thorough rinsing with PBS, Alexa Fluor 488-labeled secondary antibodies (dilution 1:1,000; Invitrogen, Thermo Fisher Scientific) were applied and incubated for 1 hour at room temperature under light-protected conditions. Nuclear staining was conducted with 4’,6-diamidino-2-phenylindole (DAPI) (1 µg/mL; Sigma-Aldrich, St. Louis, MO, USA) for 5 minutes, and coverslips were finally affixed to slides with an antifade mounting reagent. All fluorescence images were acquired with a fluorescence microscope (DMi8, Leica, Wetzlar, Germany; LSM880, Carl Zeiss AG, Oberkochen, Germany) under consistent imaging parameters. Quantitative assessment of the fluorescence signal was performed with ImageJ software.

Colony formation assay

Cells in the logarithmic growth phase were trypsinized, counted, and seeded into six-well plates at a density of 500–1,000 cells per well in complete medium. After a gentle shaking to ensure even distribution, plates were incubated at 37 ℃ with 5% CO₂ for 10–14 days, during which colonies were allowed to form. The medium was replaced every 3 days. At the endpoint, cells were washed twice with PBS, fixed in 4% paraformaldehyde for 15 minutes, and stained with 0.1% crystal violet solution for 20 minutes at room temperature. Excess dye was rinsed off with tap water, and plates were air-dried. Colonies containing more than 50 cells were counted under a microscope. Images were captured, and colony numbers were quantified with ImageJ software.

Transwell migration and invasion assays

Cell migratory and invasive capabilities were evaluated with 24-well Transwell plates equipped with porous membranes (8-μm pore size; Corning Inc., Corning, NY, USA). For completion of the migration assay, cells in the logarithmic growth phase were first serum-deprived for 6 to 12 hours. Subsequently, they were collected and suspended in medium without serum, and aliquots of 5×10⁴ to 1×10⁵ cells in 200 μL of this medium were introduced into the upper compartments. The lower chambers received 600 μL of full medium supplemented with 10% FBS to serve as a chemoattractant. Following a 12- to 24-hour incubation period at 37 ℃, nonmigratory cells remaining on the top side of the membrane were gently wiped away with a cotton swab. Cells that had migrated to the underside were fixed with 4% paraformaldehyde for 15 minutes and then stained with a 0.1% crystal violet solution for 20 minutes. For the invasion assay, the membrane’s upper side was precoated with Matrigel (BD Biosciences, Milpitas, CA, USA), diluted at ratios between 1:8 and 1:12 in serum-free medium, and allowed to solidify for 30 to 60 minutes at 37 ℃. Subsequently, 1×10⁵ to 2×10⁵ cells were added into each Matrigel-coated insert and cultured for 24 to 48 hours. After non-invading cells and residual matrix were carefully eliminated, the membranes were rinsed, air-dried, and imaged under a microscope. The number of cells that had migrated or invaded was quantified by counting at least five randomly selected visual fields per membrane. All assays were carried out in triplicate and independently repeated three times. Data are presented as the average cell count per field or normalized to the control group.

Tumor xenograft models

For the gastric cancer model, NOG mice were reconstituted with human peripheral blood mononuclear cells (PBMCs) via tail vein injection (6 weeks/female, huPBMC-NOG-DKO; Beijing Vital River Laboratory Animal Technology Co., Ltd., Beijing, China) and, after engraftment, were subcutaneously inoculated with 2×10⁶ AGS or AGS-CLDN18.2 cells suspended in 200 µL of PBS into the right axilla. For the pancreatic cancer model, B-hCD3EDG mice (4 weeks/male, B-hCD3EDG; Biocytogen Pharmaceuticals, Beijing, China) were subcutaneously implanted with 2×10⁶ KPC or KPC-CLDN18.2 cells prepared in the same manner. Tumor formation was monitored twice weekly, and mice were enrolled when palpable tumors reached approximately 5 mm in diameter. Animals were then randomized into four groups: immunoglobulin G2a (IgG2a) control, anti-programmed death-1 (anti-PD-1), IBI389, and IBI389 combined with anti-PD-1. Five tumor-bearing mice were allocated to each experimental group. Treatments were administered intravenously every 2 weeks at doses of 1–10 mg/kg according to group allocation. Body weight and tumor volumes were measured with calipers, with the latter being calculated with the following formula: tumor volume = (length × width²)/2. At the experimental endpoint or when tumor burdens reached ethical limits, mice were killed, and tumors were excised, weighed, and photographed. Excised tumors were also processed for downstream flow cytometry, multiplex immunofluorescence, cytokine analysis, and histopathology as described above. All animal experimental procedures in this study were conducted at the Specific Pathogen-Free (SPF) Animal Center of Xinchuan Campus, Sichuan University. Animal experiments were approved by the Laboratory Animal Ethics Committee of West China Hospital (approval No. 20240111004), in compliance with national guidelines for the care and use of animals were followed.

Flow cytometry of the TIME

Tumor tissues were freshly excised, mechanically minced, and enzymatically digested to obtain single-cell suspensions, which were then filtered through a 40-µm strainer and washed with PBS. After red blood cell (RBC) lysis, viable cells were resuspended in staining buffer and incubated with Fc receptor blocking solution and then fluorescently conjugated antibodies. All flow antibodies were purchased from BioLegend Inc. (San Diego, CA, USA; the specific catalog numbers are provided in Table S1) according to the experimental design. After being washed, cells were analyzed in a flow cytometer, and data were processed via FlowJo software (BD Biosciences), with a standard gating strategy being used to identify immune cell subsets within the tumor microenvironment.

Cytokine profiling via bead-based flow cytometry

Peripheral blood was collected from experimental mice at designated time points, and plasma was separated via centrifugation at 3,000 rpm for 10 minutes and stored at −80 ℃ until analysis. Cytokine levels were quantified with a multiplex bead-based flow cytometry kit (CBA/Luminex platform, BEADSTAR-Plex Mouse Multi-Factor Detection Kit/BS-MC01, and BEADSTAR-Plex Human Multi-Factor Detection Kit/BS-HC01) according to the manufacturer’s instructions. Briefly, plasma samples were incubated with fluorescence-labeled capture beads specific for interleukin (IL)-2, IL-4, IL-6, IL-10, IL-17A, tumor necrosis factor-α (TNF-α), and interferon-γ (IFN-γ), which was followed by incubation with biotinylated detection antibodies and streptavidin-phycoerythrin (PE) labeling. After unbound reagents were removed through washing, samples were acquired on a flow cytometer equipped with the appropriate lasers, and at least 5,000 beads per analyte were collected. Standard curves were generated from serial dilutions of recombinant cytokine standards, and sample concentrations were calculated with a five-parameter logistic regression model. Data were analyzed according to the manufacturer’s analysis software, and cytokine levels are expressed as absolute concentrations (pg/mL) or normalized to the percentage of total cytokine signal as required.

Multiplex immunofluorescence and spatial analysis

Paraffin-embedded tumor sections from mouse xenografts were subjected to multiplex immunofluorescence staining for CLDN18.2 (cat. No. ER1902-86; Hua’an Biotechnology Co., Ltd.) and CD3 (CD3 Recombinant Rabbit mAb KU; cat. No. BS49586; Hua’an Biotechnology Co., Ltd.), followed by scanning with a multispectral imaging system. Spatial distances between CLDN18.2⁺ tumor cells and CD3⁺ T cells were quantified via image analysis software in order to determine the immune cell infiltration and proximity across different treatment groups. In addition, a human gastric cancer tissue microarray was analyzed in parallel, and the distribution of CLDN18.2⁺ tumor cells and CD3⁺ T cells was evaluated through the use of the same staining and quantification approach to validate the findings from the animal models.

Hematoxylin and eosin (HE) staining

Liver, kidney, spleen, and tumor tissues were harvested, fixed in 4% paraformaldehyde, embedded in paraffin, sectioned, and stained with HE. Pathological changes and potential toxicities were evaluated under light microscopy.

Hematology and biochemistry

Peripheral blood samples were collected from mice at designated time points during treatment via retro-orbital or tail vein puncture under light anesthesia, and plasma was separated via centrifugation at 3,000 rpm for 10 minutes. Routine hematological parameters, including white blood cell (WBC), RBC, hemoglobin (HGB), hematocrit (HCT), and platelet (PLT) count, were measured with an automated hematology analyzer to monitor potential drug-induced myelosuppression or hematologic toxicity. In parallel, a clinical chemistry analyzer was used to evaluate serum biochemical indices, including alanine aminotransferase (ALT) and aspartate aminotransferase (AST) for liver function, blood urea nitrogen (BUN) and creatinine (Cr) for renal function, and additional markers when indicated. These analyses were performed in duplicate, and values were compared across treatment groups to assess the systemic safety of IBI389 alone or in combination with PD-1 blockade.

Statistical methods

All experimental data were derived from a minimum of three independent biological replicates. Results are expressed as the mean ± standard error of the mean (SEM). For comparisons involving two groups, either the Student’s t-test or Mann-Whitney U test was applied. Comparisons among multiple groups were conducted via one-way analysis of variance (ANOVA) followed by the Tukey post-hoc test or the Kruskal-Wallis test with Dunn correction. Analysis of tumor growth kinetics was carried out via repeated-measures ANOVA. All statistical evaluations were executed with GraphPad Prism 10 (Dotmatics, Boston, MA, USA) and R v.4.3 (The R Foundation for Statistical Computing, Vienna, Austria), with a P value <0.05 being considered statistically significant.

Results

Establishment and validation of CLDN18.2-overexpressing gastric and pancreatic cancer cell lines

To determine the baseline expression of CLDN18.2 in gastric cancer cells, Western blotting was performed across a panel of gastric cancer cell lines. CLDN18.2 was highly expressed in MKN28 and MKN45, whereas AGS cells showed relatively low expression, similar to other lines such as SGC-7901, MGC-823, BGC-803, and the normal gastric epithelial line GES-1 (Figure 1A,1B). Based on this, AGS cells were selected for subsequent stable transfection with a CLDN18.2 expression vector. Construction of the CLDN18.2-overexpressing plasmid was confirmed (Figure 1C), and Western blot analysis demonstrated markedly elevated CLDN18.2 protein levels in AGS-overexpression (AGS-OE) cells compared with AGS-Empty controls (Figure 1D,1E). Similarly, in murine pancreatic cancer KPC cells, stable CLDN18.2 overexpression was achieved, as shown by significantly increased protein expression relative to KPC-Empty cells (Figure 1F).

Figure 1 Establishment and validation of CLDN18.2-overexpressing gastric and pancreatic cancer cell lines. (A) Protein expression profile of CLDN18.2 in gastric cancer cell lines. CLDN18.2 expression levels were detected by Western blotting in multiple gastric cancer cell lines (MKN28, MKN45, AGS, SGCC-7901, MGC-823, and BGC-803) and the normal gastric epithelial cell line GES-1. (B) Quantitative analysis of CLDN18.2 protein expression by Western blotting. CLDN18.2 protein expression levels were significantly higher in MKN28 and MKN45 cell lines than in other lines, while AGS, SGC-7901, MGC-823, BGC-803, and GES-1 maintained low expression levels. (C) Plasmid map of the CLDN18.2-overexpressing lentiviral vector. The schematic illustrates the structure of the recombinant plasmid plenti-CMV-CLDN18.2/puro for stable CLDN18.2 overexpression, constructed based on the plenti-CMV-GFP/puro backbone, containing a CMV promoter, MCSs, and puro resistance marker. (D) Protein validation of stably CLDN18.2-overexpressing cell lines. CLDN18.2 protein band signals were markedly stronger in AGS-OE (human gastric cancer) and KPC-OE (murine pancreatic cancer) cells stably transfected with the CLDN18.2-overexpression plasmid compared to empty vector controls. (E) Quantitative validation of CLDN18.2 overexpression in AGS gastric cancer cells. Grayscale value analysis of Western blot results showed that CLDN18.2 protein expression levels were significantly higher in AGS-OE cells than in the AGS-Empty control cells. (F) Quantitative analysis of CLDN18.2 protein levels in KPC cells. The relative CLDN18.2 protein expression was significantly elevated in KPC-OE cells as compared to the KPC-Empty control cells. (G) Expression and localization of CLDN18.2 in AGS cells as detected by immunofluorescence (FITC staining, ×200). AGS-OE cells exhibited strong membrane-localized CLDN18.2 signals (green), while signals in the AGS-Empty control cells were weak. Nuclei were stained with DAPI (blue). (H) Quantitative analysis of CLDN18.2 immunofluorescence in AGS cells. The histogram shows that the MFI of CLDN18.2 was significantly higher in AGS-OE cells than in the AGS-Empty control cells. (I) Expression and localization of CLDN18.2 in KPC cells detected by immunofluorescence (FITC staining, ×200). Representative images show strong membrane-localized CLDN18.2 signals (green) in KPC-OE cells and weak signals in the KPC-Empty control cells. Nuclei were stained with DAPI (blue). (J) Quantitative validation of CLDN18.2 expression levels in human tumor tissue and adjacent tissue by immunofluorescence. Quantitative analysis of immunofluorescence images revealed that the MFI of CLDN18.2 was significantly higher in the tumor tissue group than in the adjacent tissue control group. ***, P<0.001; ****, P<0.0001. CLDN18.2, claudin-18.2; CMV, cytomegalovirus; DAPI, 4',6-diamidino-2-phenylindole; FITC, fluorescein isothiocyanate; GFP, green fluorescent protein; IF, immunofluorescence; MCS, multiple cloning site; MFI, mean fluorescence intensity; OE, overexpression; puro, puromycin.

Immunofluorescence staining further validated these findings. AGS-OE cells exhibited robust CLDN18.2 signals with higher mean fluorescence intensity as compared to AGS-Empty cells (Figure 1G,1H). Similarly, KPC-overexpression (KPC-OE) cells displayed markedly enhanced CLDN18.2 fluorescence intensity as compared to KPC-Empty controls (Figure 1I,1J). Together, these data confirm the successful generation of stable CLDN18.2-overexpressing gastric (AGS) and pancreatic (KPC) cancer cell lines, providing reliable models for subsequent functional and therapeutic studies.

CLDN18.2 overexpression enhanced the proliferation and migration of gastric and pancreatic cancer cells

To investigate the functional role of CLDN18.2 in tumor cell behavior, colony formation, and Transwell assays were performed in human gastric (AGS) and murine pancreatic (KPC)cancer cell lines with stable CLDN18.2 overexpression. Colony formation assays revealed that CLDN18.2 overexpression significantly promoted the proliferative capacity of all tested cell lines. Compared with empty controls, AGS-CLDN18.2 cells generated markedly more colonies (Figure 2A), and led to a significant increase in colony number (Figure 2B). Similarly, overexpression of CLDN18.2 in KPC cells led to a significant increase in colony number (Figure 2C,2D). Transwell migration assays further confirmed the protumorigenic role of CLDN18.2. AGS-CLDN18.2 and KPC-CLDN18.2 cells exhibited a significantly higher number of migrated cells than did Empty controls (Figure 2E-2H). Collectively, these results indicate that CLDN18.2 overexpression promotes both the proliferation and migratory potential of gastric and pancreatic cancer cells, supporting a functional role of CLDN18.2 in tumor progression.

Figure 2 CLDN18.2 overexpression enhanced the proliferation and migration of gastric and pancreatic cancer cells. (A-D) Representative images showing the effect of CLDN18.2 overexpression on proliferation in AGS and KPC cells (0.1% crystal violet staining, ×1). Colony formation assay demonstrated that AGS-OE and KPC-OE cells generated more colonies than did the empty control cells. (E-H) Transwell assay showed that the AGS-OE and KPC-OE group had stronger invasiveness than did the empty vector control group (0.1% crystal violet staining, ×200). **, P<0.01; ***, P<0.001; ****, P<0.0001. CLDN18.2, claudin-18.2; OE, overexpression.

IBI389 inhibited tumor growth in vivo and synergized with PD-1 blockade

Representative tumor photographs showed marked size reductions with IBI389 monotherapy and the greatest shrinkage with IBI389 + PD-1 in both the AGS (Figure 3A) and KPC (Figure 3B) models. Quantification confirmed this pattern: in AGS, the IgG2a control had the largest tumor volume and weight, PD-1 monotherapy produced a modest decrease, IBI389 monotherapy significantly lowered both metrics as compared to IgG2a and PD-1, and the combination achieved the lowest values (Figure 3C,3D). In the AGS xenograft model, the tumor weight inhibition rate of the combination group was reduced about 74.8% (vs. IgG2a control), which was 27.3% lower than that of IBI389 monotherapy. Safety-related readouts included in the AGS model show no significant differences in liver or spleen weights between the groups (Figure 3E,3F). The KPC model mirrored these results, with robust suppression in the IBI389-alone group and the strongest effect in the combination group (Figure 3G,3H). In the KPC model, the tumor volume inhibition rate of the combination group reduced 38.3% (vs. IgG2a control), 8.6% lower than that of the monotherapy group, and the tumor weight was comparable with the monotherapy group. Safety-related readouts included in the KPC model show no significant differences in liver or spleen weights between the groups (Figure 3I,3J). Together, the data in Figure 3 indicate that IBI389 effectively suppresses tumor growth in vivo and, when combined with PD-1 blockade, yields the greatest antitumor benefit, consistent with a synergistic effect.

Figure 3 IBI389 inhibited tumor growth in vivo and synergized with PD-1 blockade. (A,B) Tumor photographs from each treatment group in the AGS and KPC xenograft model. Images show markedly reduced tumor volumes in the IBI389 monotherapy and IBI389 + PD-1 combination groups compared to the IgG2a control group. (C) Tumor volume changes in each treatment group of the AGS model. Quantitative data demonstrated that IBI389 monotherapy and combination therapy significantly suppressed tumor growth as compared to control groups. (D) Tumor weights in each treatment group of the AGS model. Bar graphs show that tumor weights in the IBI389 monotherapy and combination groups were significantly lower than those in control groups. (E,F) Liver and spleen weights in each treatment group of the AGS model. No significant differences were observed between the groups. (G) Tumor volume changes in each treatment group of the KPC model. Quantitative data demonstrated that IBI389 monotherapy and combination therapy significantly suppressed tumor growth. (H) Tumor weights in each treatment group of the KPC model. Bar graphs show that tumor weights in the IBI389 monotherapy and combination groups were significantly lower than those in control groups. (I,J) Liver and spleen weights in each treatment group of the KPC model. No significant differences were observed between the groups. **, P<0.01; ***, P<0.001; ns, not significant. h-IgG, human immunoglobulin G; IgG2a, immunoglobulin G2a; PD-1, programmed death-1.

IBI389 combined with PD-1 blockade reshaped the TIME

To clarify the immunological mechanisms underlying the antitumor activity of the CLDN18.2-CD3 bispecific antibody, flow cytometry was performed to profile tumor-infiltrating immune cells in both AGS (human gastric cancer xenograft with PBMC reconstitution) and KPC (murine pancreatic cancer) models. In the AGS model, IBI389 monotherapy significantly increased the proportion of CD45⁺ immune cells compared with the IgG2a or PD-1 treatment, and the combination of IBI389 with PD-1 blockade resulted in the highest infiltration (Figure 4A). Further analysis revealed that both total CD3⁺ T cells and CD8⁺ cytotoxic T cells were markedly elevated in the IBI389 + PD-1 group (Figure 4B,4C). In parallel, macrophage infiltration was enhanced by combination therapy (Figure 4D). Quantitative analysis of the proportion of tumor-infiltrating in the different treatment groups of the AGS model suggests a broad remodeling of the immune landscape (Figure 4E-4H). Similarly, in the KPC model, treatment with IBI389 significantly increased the frequency of intratumoral immune cells, and combination with PD-1 blockade led to the most pronounced infiltration (Figure 5A). The abundance of both CD3⁺ and CD8⁺ T cells was markedly increased in the combination group (Figure 5B,5C). Interestingly, the proportion of regulatory T cells (Tregs; CD4⁺CD25⁺FoxP3⁺) was also significantly elevated following IBI389 + PD-1 treatment as compared to treatment with the monotherapies (Figure 5D), indicating that bispecific antibody-driven T-cell activation may be accompanied by compensatory immune regulation. Statistical analysis procedures are elaborated in (Figure 5E-5H). Together, these results demonstrate that IBI389 therapy markedly enhances immune cell infiltration in the tumor microenvironment, particularly effector CD8⁺ T cells. Importantly, its combination with PD-1 blockade can achieve the most profound remodeling, converting tumors toward a more immune-inflamed phenotype, although this may potentially be accompanied by increased Treg recruitment.

Figure 4 IBI389 combined with PD-1 blockade reshaped the TIME in human gastric cancer xenograft with PBMC reconstitution models. (A-D) Representative flow cytometry plots of tumor-infiltrating CD45⁺ immune cells, CD3⁺ T cells, CD8⁺ T cells, and macrophages in the different treatment groups of the AGS model. (E-H) Quantitative analysis of the proportion of tumor-infiltrating CD45⁺ immune cells, CD3⁺ T cells, CD8⁺ T cells, and macrophages in the different treatment groups of the AGS model. Bar graphs show that IBI389 monotherapy and combination therapy significantly increased the proportion of CD3⁺ T cells, CD8⁺ T cells, and macrophages, with the most pronounced effect observed in the combination group, while no significant difference was observed between the PD-1 monotherapy group and the control group. **, P<0.01; ***, P<0.001; ns, not significant. PBMC, peripheral blood mononuclear cell; PD-1, programmed death-1; TIME, tumor immune microenvironment.

Figure 5 IBI389 combined with PD-1 blockade reshaped the TIME in murine pancreatic cancer models. (A-D) Representative flow cytometry plots of tumor-infiltrating CD45⁺ immune cells, CD3⁺ T cells, CD8⁺ T cells, and Tregs in the different treatment groups of the KPC model. (E-H) Quantitative analysis of the proportion of tumor-infiltrating CD45⁺ immune cells, CD3⁺ T cells, CD8⁺ T cells, and Tregs in the different treatment groups of the KPC model. Bar graphs show that IBI389 monotherapy and combination therapy significantly increased the proportion of CD3⁺ T cells, CD8⁺ T cells, and Tregs, with the most pronounced effect observed in the combination group, while no significant difference was observed between the PD-1 monotherapy group and the control group. **, P<0.01; ***, P<0.001. PD-1, programmed death-1; TIME, tumor immune microenvironment; Tregs, regulatory T cells.

IBI389 combined with PD-1 blockade skewed cytokine production toward an antitumor profile

Multiplex bead-based flow cytometry was performed to evaluate systemic cytokine responses in the AGS xenograft model (Figure 6A). The results revealed that IBI389 treatment led to a pronounced increase in IL-2 levels compared with treatment with IgG2a or PD-1 monotherapy, and the combination of IBI389 with PD-1 blockade further boosted IL-2 secretion to the highest level, indicating enhanced activation and expansion of effector T cells (Figure 6B). Concomitantly, IFN-γ increased with PD-1 and IBI389 treatment, whereas TNF-α decreased across treatment groups; the combination did not further raise IFN-γ levels, indicating no generalized proinflammatory surge (Figure 6C,6D). In contrast, the level of cytokines associated with immune suppression or tumor promotion was reduced following treatment. The level of IL-4, typically linked with T helper (Th)2—skewing and tumor-permissive immunity, was reduced by IBI389 monotherapy and combination treatment, while the level of IL-10 was also reduced in all treatment arms. The level of IL-4, typically linked with a Th2 skew and tumor-permissive immunity, was significantly reduced in both the IBI389 monotherapy and combination therapy groups (Figure 6E); meanwhile, the level of IL-10, another key immunoregulatory cytokine, showed a modest reduction in both the IBI389 monotherapy and combination therapy groups (Figure 6F). Notably, the level of IL-17A, which contributes to chronic inflammation and tumor progression, was strongly suppressed by IBI389 alone and further reduced by treatment with combination therapy. Notably, the level of IL-17A, which contributes to chronic inflammation and tumor progression, was significantly reduced in both the IBI389 monotherapy and combination therapy groups (Figure 6G). Furthermore, the level of IL-6 decreased in the IBI389 and combination groups, suggesting the attenuation of nonspecific inflammatory pathways in these arms (Figure 6H). Collectively, these findings indicate that IBI389, particularly when combined with PD-1 blockade, remodels the systemic cytokine network by amplifying IL-2, TNF-α, and IFN-γ while simultaneously suppressing IL-4, IL-10, and IL-17A, thereby establishing a systemic immune profile that is strongly supportive of antitumor immunity.

Figure 6 IBI389 combined with PD-1 blockade skewed cytokine production toward an antitumor profile. (A) Schematic overview of systemic cytokine profiling in the AGS xenograft model. Multiplex bead-based flow cytometry analysis was performed to evaluate the cytokine network in the different treatment groups (IgG2a control, PD-1 monotherapy, IBI389 monotherapy, and IBI389 + PD-1 combination). (B) IBI389 combination therapy significantly enhanced serum IL-2 levels. Compared with the IgG2a control and PD-1 monotherapy regimens, the IBI389-containing regimens significantly increased IL-2 concentrations, with the highest secretion level achieved by IBI389 combined with PD-1 blockade. (C,D) TNF-α levels in the different treatment groups. IFN-γ level increased with PD-1 and IBI389 treatment, whereas TNF-α level decreased across the treatment groups; the combination did not further raise IFN-γ levels, indicating no generalized proinflammatory surge. (E) IL-4 levels in the different treatment groups. IL-4 was reduced in the IBI389 monotherapy and the combination groups. (F) IL-10 levels in the different treatment groups. IL-10 showed a modest reduction in both IBI389 monotherapy and combination therapy groups. (G) IL-17A levels in the different treatment groups. IL-17A was strongly suppressed by IBI389 alone and further reduced in the combination therapy group. (H) IL-6 levels in the different treatment groups. IL-6 decreased in the IBI389 and combination groups, suggesting attenuation of nonspecific inflammatory pathways in these arms. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001; ns, not significant. h-IgG, human immunoglobulin G; IFN-γ, interferon-γ; IgG2a, immunoglobulin G2a; IL, interleukin; PD-1, programmed death-1; TNF-α, tumor necrosis factor-α.

Histopathological evaluation of major organs demonstrated the favorable safety profile of IBI389

To assess the potential systemic toxicity of IBI389, HE staining was performed on liver, kidney, and spleen tissues from the treated animals (Figure 7A,7B). Both the AGS xenograft model in NOG mice reconstituted with human PBMCs and the KPC murine pancreatic cancer model were examined following repeated administration of IBI389, PD-1 antibody, or their combination (AGS: Figure 7A; KPC: Figure 7B). In all the treatment groups, histological analysis revealed the preserved architecture of hepatic lobules and no hepatocyte degeneration, necrosis, or inflammatory infiltration (Figure 7A,7B). Similarly, renal tissues exhibited intact glomerular and tubular structures without detectable signs of edema, tubular dilatation, or interstitial fibrosis (Figure 7A,7B). In the spleen, normal white pulp and red pulp organization was maintained, with no abnormal lymphoid depletion or hyperplasia observed (Figure 7A,7B). Importantly, there were no histological features indicative of drug-induced organ injury in either model (Figure 7A,7B). These findings collectively demonstrate that IBI389, alone or in combination with PD-1 blockade, does not cause overt histopathological damage to major organs, supporting its favorable preclinical safety profile.

Figure 7 Histopathological evaluation of major organs demonstrated the favorable safety profile of IBI389. (A) Representative histopathological images (HE staining, ×100) of major organs from the different treatment groups in the AGS model. Liver, kidney, and spleen tissues maintained normal architecture across all treatment groups, with no evident pathological damage being observed. (B) Representative histopathological images (HE staining, ×100) of major organs from the different treatment groups in the KPC model. Liver, kidney, and spleen tissues maintained normal architecture across all the treatment groups, with no evident pathological damage being observed. h-IgG, human immunoglobulin G; HE, hematoxylin and eosin; IgG2a, immunoglobulin G2a; PD-1, programmed death-1.

IBI389 increased intratumoral CD3+ T-cell infiltration

Multiplex immunofluorescence was used to quantify intratumoral CD3⁺ T-cell infiltration (cells/mm²) within CLDN18.2⁺ tumor regions of interest (Figure 8A-8F; red: CLDN18.2; green: CD3; scale bar: 20 µm). In the AGS xenograft cohort, representative images showed visibly greater CD3⁺ accumulation after IBI389 and combination therapy (Figure 8A). Quantification confirmed that IBI389 significantly increased intratumoral CD3⁺ T-cell density as compared to IgG2a and PD-1 monotherapy (Figure 8A), with the IBI389 + PD-1 group attaining the highest count, significantly exceeding that of IBI389 alone (Figure 8D). For the KPC model, the images similarly suggested treatment-driven increases (Figure 8B), and quantification showed a significant rise with IBI389 relative to IgG2a and PD-1 (Figure 8B). The combination displayed a further numerical increase that did not differ significantly from that of IBI389 (Figure 8E). To contextualize these results, human gastric cancer tissues exhibited baseline CD3⁺ T-cell densities comparable to those of untreated mouse controls (Figure 8C,8F), indicating that the marked elevations observed in AGS and KPC were treatment-dependent and not baseline features of CLDN18.2⁺ tumors. Collectively, these data demonstrate that IBI389 robustly increases the number of CD3⁺ T cells within CLDN18.2⁺ tumor nests, with additional benefit from PD-1 co-therapy in AGS, and a strong single-agent effect being observed in KPC.

Figure 8 IBI389 increased intratumoral CD3+ T-cell infiltration in xenograft tumor models. (A) Representative mIF images of intratumoral CD3⁺ T-cell infiltration in the different treatment groups of the AGS model (red: CLDN18.2; green: CD3; scale bar: 20 µm; mIF staining, ×100). (B) Representative mIF images of intratumoral CD3⁺ T-cell infiltration in the different treatment groups of the KPC model (red: CLDN18.2; green: CD3; scale bar: 20 µm, mIF staining, ×100). (C) Baseline T-cell infiltration in human gastric cancer tissues as detected by mIF (mIF staining, ×100). Representative images show the distribution of CD3⁺ T cells (green) within CLDN18.2-positive (red) tumor regions, with infiltration levels comparable to those of untreated mouse model controls. (D) Quantitative analysis of intratumoral CD3⁺ T-cell density in the different treatment groups of the AGS model. Bar graphs show that IBI389 monotherapy and combination therapy significantly increased CD3⁺ T-cell density, with the most pronounced effect observed in the combination group. (E) Quantitative analysis of intratumoral CD3⁺ T-cell density in the different treatment groups of the KPC model. Bar graphs show that IBI389 monotherapy and combination therapy significantly increased CD3⁺ T-cell density, with no significant difference between the combination and monotherapy groups. (F) Comparative analysis of intratumoral CD3⁺ T-cell density between human gastric cancer tissues and mouse models. Bar graphs show no significant difference in CD3+ T-cell density between human gastric cancer tissues and untreated mouse control tissues. *, P<0.05; ***, P<0.001; ns, not significant. CLDN18.2, claudin-18.2; h-IgG, human immunoglobulin G; IgG2a, immunoglobulin G2a; mIF, multiplex immunofluorescence; PD-1, programmed death-1.

Hematologic redistribution in AGS and preserved hepatic biochemistry in KPC after IBI389 ± PD-1

Peripheral blood analyses revealed treatment-dependent hematologic remodeling in the AGS cohort (Figure 9A-9G). Total leukocyte count (WBC) was increased in the IBI389-containing regimens, with the highest values in the IBI389 + PD-1 group relative to IgG2a and PD-1 (Figure 9A). This rise was accompanied by a reduction in lymphocytes in terms of both absolute counts and percentages (Figure 9B,9C) and a concomitant increase in neutrophils in counts and fractions (Figure 9D,9E). Monocytes showed a modest but consistent elevation in counts and percentages with IBI389 ± PD-1 (Figure 9F,9G). In contrast, serum chemistry in the KPC cohort remained stable across groups, with ALT, AST, albumin (ALB), and γ-glutamyl transferase (γ-GT) showing no significant differences (Figure 9H-9M). Collectively, these data indicate that IBI389 ± PD-1 reshapes circulating leukocyte composition in AGS—shifting toward a myeloid-biased profile—and does not produce hepatotoxicity in KPC (Figure 9).

Figure 9 Hematologic redistribution in AGS and preserved hepatic biochemistry in KPC after IBI389 ± PD-1. (A) Changes in total peripheral WBC count in the different treatment groups of the AGS model. Bar graphs show that IBI389-containing regimens increased WBC counts, with the most pronounced effect observed in the combination group. (B) Changes in absolute lymphocyte counts in the peripheral blood of the different treatment groups in the AGS model. Statistical results indicated that IBI389 monotherapy and combination therapy reduced absolute lymphocyte counts as compared to IgG2a and PD-1 monotherapy. (C) Changes in lymphocyte percentages in the peripheral blood of the different treatment groups in the AGS model. Statistical results indicated that IBI389 monotherapy and combination therapy significantly reduced lymphocyte percentages as compared to IgG2a and PD-1 monotherapy. (D) Changes in absolute neutrophil counts in the peripheral blood of the different treatment groups in the AGS model. Statistical results indicated that IBI389 monotherapy and combination therapy significantly increased absolute neutrophil counts as compared to IgG2a and PD-1 monotherapy. (E) Changes in neutrophil percentages in the peripheral blood of the different treatment groups in the AGS model. Statistical results indicated that IBI389 monotherapy and combination therapy significantly increased neutrophil percentages compared to the IgG2a and PD-1 monotherapy. (F) Changes in absolute monocyte counts in the peripheral blood of different treatment groups in the AGS model. Statistical results indicated that IBI389 monotherapy and combination therapy induced a moderate increase in absolute monocyte counts as compared to IgG2a and PD-1 monotherapy. (G) Changes in monocyte percentages in the peripheral blood of the different treatment groups in the AGS model. Statistical results indicated that IBI389 monotherapy and combination therapy significantly increased monocyte percentages as compared to IgG2a and PD-1 monotherapy. (H,I) Changes in HGB and PLT counts in the peripheral blood of the different treatment groups in the AGS model. Statistical results indicated that IBI389 monotherapy and combination therapy exerted a relatively small impact on HGB and PLT. (J) Quantitative analysis of serum ALT levels in the different treatment groups of the KPC model. Bar graphs show no significant differences in ALT levels between the treatment groups. (K) Quantitative analysis of serum AST levels in the different treatment groups of the KPC model. Bar graphs show no significant differences in AST levels between the treatment groups. (L) Quantitative analysis of the serum ALB levels in the different treatment groups of the KPC model. Bar graphs show no significant differences in ALB levels between the treatment groups. (M) Quantitative analysis of serum γ-GT levels in the different treatment groups of the KPC model. Bar graphs show no significant differences in γ-GT levels between the treatment groups. Data are the mean ± SEM from one-way ANOVA with the Tukey post-hoc test. *, P<0.05; **, P<0.01; ***, P<0.001; ns, not significant. ALB, albumin; ALT, alanine aminotransferase; ANOVA, analysis of variance; AST, aspartate aminotransferase; h-IgG, human immunoglobulin G; HGB, hemoglobin; IgG2a, immunoglobulin G2a; Lym, lymphocyte; Mon, monocyte; Neu, neutrophil; PD-1, programmed death-1; PLT, platelet; γ-GT, γ-glutamyl transferase; SEM, standard error of the mean; WBC, white blood cell.

Discussion

In this study, we examined the antitumor activity and immunologic mechanism of a CLDN18.2-CD3 bispecific antibody (IBI389) across two complementary systems—a human PBMC-reconstituted AGS xenograft and a B-hCD3EDG KPC model—and integrated these findings with histopathology, cytokine profiling, and spatial multiplex immunofluorescence. Four observations were particularly notable. First, CLDN18.2 overexpression in gastric and pancreatic cancer cells augmented proliferative and migratory phenotypes in vitro, suggesting that CLDN18.2 is not merely a passive surface tag but may be associated with tumor-promoting programs. Second, redirecting T cells to CLDN18.2 with IBI389 suppressed tumor growth in vivo, and coadministration with PD-1 blockade further enhanced efficacy, yielding the smallest volumes and endpoint weights in both models. Third, IBI389 remodeled the TIME toward an effector-dominant state—with marked increases in CD45⁺ infiltrates and CD8⁺ T cells—while also causing a compensatory rise in Treg abundance, especially under combination therapy. Fourth, at the systemic level, cytokine profiling demonstrated a coordinated skewing toward an antitumor milieu (upregulation of IL-2, TNF-α, and IFN-γ; downregulation of IL-4, IL-10, and IL-17A), and multiplex immunofluorescence quantification showed that IBI389—especially in combination with PD-1—elevated CD3⁺ T-cell density (cells/mm²) within CLDN18.2⁺ tumor regions. Across both models, HE staining of the liver, kidney, and spleen indicated preserved architecture without overt drug-related injury, supporting a favorable preclinical safety profile.

These data together suggest a mechanistic model in which antigen density on tumor cells provides docking sites for IBI389 to bridge and activate T cells locally (15-17), seeding the following positive feedback loop: redirected cytotoxicity and IFN-γ/TNF-α release promote further recruitment and retention of effector cells, while systemic IL-2 reinforces proliferation and survival of activated clones. The spatial convergence of CD3⁺ T cells and CLDN18.2⁺ nests—recapitulated in human gastric cancer tissue microarrays—offers orthogonal evidence that the drug does more than raise absolute immune counts and can additionally improve where those cells are positioned relative to their targets, a parameter being increasingly recognized as critical for effective killing (18-20). That this spatial signature matches the direction of change in our cytokine readouts [more Th1/cytotoxic T lymphocyte (CTL)-supportive and less Th2/Th17/IL-10-supportive] strengthens the internal coherence of the mechanism.

The rise in Tregs with the combination regimen is also biologically plausible and clinically meaningful. Amplification of IL-2 levels can preferentially support high-affinity IL-2Rα (CD25)-expressing Tregs (21,22), and inflamed tumors often upregulate chemokines (e.g., CCL22/CCL28) that attract Tregs once effector responses intensify (22,23). Functionally, this feedback may cap the upper bound of cytotoxicity and help explain why some tumors plateau rather than regress completely despite robust infiltration. The translational implication is that clinicians should not avoid combination therapy but rather layer rational modulators that neutralize this brake [e.g., anti-CD25 variants sparing effector IL-2 signaling, low-dose cyclophosphamide, or metabolic/transforming growth factor-β (TGF-β)/adenosine pathway inhibitors] and use scheduling that pulses IL-2 exposure without chronically favoring Tregs (24).

Our safety readouts are reassuring: there was no histologic evidence of hepatotoxicity, nephrotoxicity, or splenic pathology under repeated dosing. Given CLDN18.2’s restricted normal distribution, on-target off-tumor risk is likely manageable; nevertheless, clinical translation should incorporate prospective monitoring for cytokine-release features and gastric mucosal effects, particularly at higher antigen loads or with PD-1 co-stimulation (25,26). The cytokine panel in our models did not show an increase in IL-6 levels, which is consistent with the absence of pathological inflammation, but larger species-translation studies are necessary to define the margins between pharmacodynamic activation and toxicity.

This work also helps reconcile the field’s heterogeneous literature on CLDN18.2. Our baseline screens confirm that expression varies across lines; engineered overexpression augments malignant traits yet makes tumors more addressable via CLDN18.2-directed T-cell engagement (27,28). Thus, CLDN18.2 can be both a facilitator of tumor biology and a susceptibility that therapy exploits. The decisive factor for response may be the spatially realized triad of (I) sufficient antigen density; (II) the ability to corral T cells into immediate proximity; and (III) a checkpoint context that allows synapse formation to mature into killing rather than exhaustion. Our observation that spatial proximity is improved by IBI389 and tracks with effects observed in human tissues suggests that companion diagnostics should move beyond bulk immunohistochemistry percent positives toward quantitative spatial metrics [e.g., intratumoral CD3⁺ T-cell density (cells/mm²)] that better predict bispecific performance (29).

Certain limitations to this study should be acknowledged. First, the PBMC-humanized NOG mice constitute an incomplete model for myeloid education and ectopic homing, while the B-hCD3EDG model may not fully recapitulate human checkpoint dynamics. Second, our in vitro overexpression systems simplify the complexity of endogenous CLDN18.2 regulation, including splice variants and membrane trafficking, that may affect epitope exposure. Third, while our cytokine and flow panels were informative, deeper single-cell multiomics (e.g., single-cell RNA sequencing or cellular indexing of transcriptomes and epitopes by sequencing with T-cell receptor clonotyping) would refine maps of effector vs. regulatory circuits and identify chemokine/adhesion programs that underlie the observed spatial convergence.

Future work should therefore (I) profile antigen density and T-cell spatial metrics prospectively in clinical samples as potential biomarkers of benefit; (II) test combinatorial regimens that counteract the Treg rebound [anti-CD25 and CCR4 blockade (30,31) or adenosine/TGF-β pathway inhibition (32-35)]; (III) identify myeloid-reprogramming partners (e.g., CSF1R and CD47/SIRPα) that can achieve sustain antigen presentation and antigen-specific killing (36,37); (IV) evaluate dose and sequence optimization with PD-1 blockade to maximize effector expansion while minimizing regulatory capture; and (V) expand safety characterization with sensitive gastric mucosa assays and longitudinal biochemistry.

In summary, IBI389 redirects and activates T cells against CLDN18.2-expressing tumors, exerting potent growth inhibition that is amplified by PD-1 blockade. Mechanistically, therapy with IBI389 increases effector infiltration, skews systemic cytokines toward a tumoricidal profile, and—critically—reduces the spatial gap between T cells and their targets. The simultaneous emergence of regulatory counter responses suggests rational avenues for next-generation combinations. Together, these findings provide a mechanistic rationale and translational blueprint for deploying CLDN18.2-CD3 bispecific antibodies for the treatment of gastric cancer and related CLDN18.2-positive malignancies.

Conclusions

In this study, we demonstrated that the CLDN18.2-CD3 bispecific antibody IBI389 exerts potent antitumor activity in gastric and pancreatic cancer models, with the strongest effects observed when the combination with PD-1 blockade was applied. IBI389 effectively inhibited tumor growth, enhanced CD8⁺ T-cell infiltration, remodeled systemic cytokine responses, and reduced the spatial distance between tumor cells and T cells, all without inducing overt organ toxicity. These findings support IBI389 as a promising therapeutic candidate for the treatment of CLDN18.2-positive tumors and provide a strong rationale for its further clinical development.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the ARRIVE and MDAR reporting checklists. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0474/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0474/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0474/prf

Funding: This work was supported by the Project for Disciplines of Excellence, West China Hospital, Sichuan University (No. ZYGD24004), the Sichuan Medical Science and Technology Innovation Research Association (No. YCH-KY-YCZD2024-031), and the Wu Jieping Medical Foundation (No. 320.6750.2024-13-13).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0474/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Animal experiments were approved by the Laboratory Animal Ethics Committee of West China Hospital (approval No. 20240111004), in compliance with national guidelines for the care and use of animals were followed.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Smyth EC, Nilsson M, Grabsch HI, et al. Gastric cancer. Lancet 2020;396:635-48. [Crossref] [PubMed]
2. Zeng Y, Jin RU. Molecular pathogenesis, targeted therapies, and future perspectives for gastric cancer. Semin Cancer Biol 2022;86:566-82. [Crossref] [PubMed]
3. Bang YJ, Van Cutsem E, Feyereislova A, et al. Trastuzumab in combination with chemotherapy versus chemotherapy alone for treatment of HER2-positive advanced gastric or gastro-oesophageal junction cancer (ToGA): a phase 3, open-label, randomised controlled trial. Lancet 2010;376:687-97. [Crossref] [PubMed]
4. Li Y, Zheng Y, Huang J, et al. CAF-macrophage crosstalk in tumour microenvironments governs the response to immune checkpoint blockade in gastric cancer peritoneal metastases. Gut 2025;74:350-63. [Crossref] [PubMed]
5. Janjigian YY, Ajani JA, Moehler M, et al. First-Line Nivolumab Plus Chemotherapy for Advanced Gastric, Gastroesophageal Junction, and Esophageal Adenocarcinoma: 3-Year Follow-Up of the Phase III CheckMate 649 Trial. J Clin Oncol 2024;42:2012-20. [Crossref] [PubMed]
6. Lee KW, Van Cutsem E, Bang YJ, et al. Association of Tumor Mutational Burden with Efficacy of Pembrolizumab±Chemotherapy as First-Line Therapy for Gastric Cancer in the Phase III KEYNOTE-062 Study. Clin Cancer Res 2022;28:3489-98. [Crossref] [PubMed]
7. Tsukita S, Tanaka H, Tamura A. The Claudins: From Tight Junctions to Biological Systems. Trends Biochem Sci 2019;44:141-52. [Crossref] [PubMed]
8. Kubota Y, Shitara K. Zolbetuximab for Claudin18.2-positive gastric or gastroesophageal junction cancer. Ther Adv Med Oncol 2024;16:17588359231217967. [Crossref] [PubMed]
9. Qi C, Gong J, Li J, et al. Claudin18.2-specific CAR T cells in gastrointestinal cancers: phase 1 trial interim results. Nat Med 2022;28:1189-98. [Crossref] [PubMed]
10. Sahin U, Koslowski M, Dhaene K, et al. Claudin-18 splice variant 2 is a pan-cancer target suitable for therapeutic antibody development. Clin Cancer Res 2008;14:7624-34. [Crossref] [PubMed]
11. Jin WM, Zhu Y, Cai ZQ, et al. Progress of Clinical Studies Targeting Claudin18.2 for the Treatment of Gastric Cancer. Dig Dis Sci 2024;69:2631-47. [Crossref] [PubMed]
12. Jun KH, Kim JH, Jung JH, et al. Expression of claudin-7 and loss of claudin-18 correlate with poor prognosis in gastric cancer. Int J Surg 2014;12:156-62. [Crossref] [PubMed]
13. Rohde C, Yamaguchi R, Mukhina S, et al. Comparison of Claudin 18.2 expression in primary tumors and lymph node metastases in Japanese patients with gastric adenocarcinoma. Jpn J Clin Oncol 2019;49:870-6. [Crossref] [PubMed]
14. Goebeler ME, Stuhler G, Bargou R. Bispecific and multispecific antibodies in oncology: opportunities and challenges. Nat Rev Clin Oncol 2024;21:539-60. [Crossref] [PubMed]
15. Ellerman D. Bispecific T-cell engagers: Towards understanding variables influencing the in vitro potency and tumor selectivity and their modulation to enhance their efficacy and safety. Methods 2019;154:102-17. [Crossref] [PubMed]
16. Betts A, Haddish-Berhane N, Shah DK, et al. A Translational Quantitative Systems Pharmacology Model for CD3 Bispecific Molecules: Application to Quantify T Cell-Mediated Tumor Cell Killing by P-Cadherin LP DART AAPS J 2019;21:66. [Crossref] [PubMed]
17. Albayrak G, Wan PK, Fisher K, et al. T cell engagers: expanding horizons in oncology and beyond. Br J Cancer 2025;133:1241-9. [Crossref] [PubMed]
18. Lee DW, Santomasso BD, Locke FL, et al. ASTCT Consensus Grading for Cytokine Release Syndrome and Neurologic Toxicity Associated with Immune Effector Cells. Biol Blood Marrow Transplant 2019;25:625-38. [Crossref] [PubMed]
19. Schürch CM, Bhate SS, Barlow GL, et al. Coordinated Cellular Neighborhoods Orchestrate Antitumoral Immunity at the Colorectal Cancer Invasive Front. Cell 2020;182:1341-1359.e19. [Crossref] [PubMed]
20. Xia Y, Ma J, Yang X, et al. Identifying the Spatial Architecture That Restricts the Proximity of CD8(+) T Cells to Tumor Cells in Pancreatic Ductal Adenocarcinoma. Cancers (Basel) 2024;16:1434. [Crossref] [PubMed]
21. Shouse AN, LaPorte KM, Malek TR. Interleukin-2 signaling in the regulation of T cell biology in autoimmunity and cancer. Immunity 2024;57:414-28. [Crossref] [PubMed]
22. Koreth J, Matsuoka K, Kim HT, et al. Interleukin-2 and regulatory T cells in graft-versus-host disease. N Engl J Med 2011;365:2055-66. [Crossref] [PubMed]
23. Curiel TJ, Coukos G, Zou L, et al. Specific recruitment of regulatory T cells in ovarian carcinoma fosters immune privilege and predicts reduced survival. Nat Med 2004;10:942-9. [Crossref] [PubMed]
24. Solomon I, Amann M, Goubier A, et al. CD25-T(reg)-depleting antibodies preserving IL-2 signaling on effector T cells enhance effector activation and antitumor immunity. Nat Cancer 2020;1:1153-66. [Crossref] [PubMed]
25. Shitara K, Lordick F, Bang YJ, et al. Zolbetuximab plus mFOLFOX6 in patients with CLDN18.2-positive, HER2-negative, untreated, locally advanced unresectable or metastatic gastric or gastro-oesophageal junction adenocarcinoma (SPOTLIGHT): a multicentre, randomised, double-blind, phase 3 trial. Lancet 2023;401:1655-68.
26. Wang N, Patel H, Schneider IC, et al. An optimal antitumor response by a novel CEA/CD3 bispecific antibody for colorectal cancers. Antib Ther 2021;4:90-100. [Crossref] [PubMed]
27. Cao W, Xing H, Li Y, et al. Claudin18.2 is a novel molecular biomarker for tumor-targeted immunotherapy. Biomark Res 2022;10:38. [Crossref] [PubMed]
28. Liu S, Zhang Z, Jiang L, et al. Claudin-18.2 mediated interaction of gastric Cancer cells and Cancer-associated fibroblasts drives tumor progression. Cell Commun Signal 2024;22:27. [Crossref] [PubMed]
29. Williams HL, Frei AL, Koessler T, et al. The current landscape of spatial biomarkers for prediction of response to immune checkpoint inhibition. NPJ Precis Oncol 2024;8:178. [Crossref] [PubMed]
30. Maeda S, Motegi T, Iio A, et al. Anti-CCR4 treatment depletes regulatory T cells and leads to clinical activity in a canine model of advanced prostate cancer. J Immunother Cancer 2022;10:e003731. [Crossref] [PubMed]
31. Zamarin D, Hamid O, Nayak-Kapoor A, et al. Mogamulizumab in Combination with Durvalumab or Tremelimumab in Patients with Advanced Solid Tumors: A Phase I Study. Clin Cancer Res 2020;26:4531-41. [Crossref] [PubMed]
32. Willingham SB, Ho PY, Hotson A, et al. A2AR Antagonism with CPI-444 Induces Antitumor Responses and Augments Efficacy to Anti-PD-(L)1 and Anti-CTLA-4 in Preclinical Models. Cancer Immunol Res 2018;6:1136-49. [Crossref] [PubMed]
33. Mariathasan S, Turley SJ, Nickles D, et al. TGFβ attenuates tumour response to PD-L1 blockade by contributing to exclusion of T cells. Nature 2018;554:544-8. [Crossref] [PubMed]
34. Gulley JL, Schlom J, Barcellos-Hoff MH, et al. Dual inhibition of TGF-β and PD-L1: a novel approach to cancer treatment. Mol Oncol 2022;16:2117-34. [Crossref] [PubMed]
35. Lind H, Gameiro SR, Jochems C, et al. Dual targeting of TGF-β and PD-L1 via a bifunctional anti-PD-L1/TGF-βRII agent: status of preclinical and clinical advances. J Immunother Cancer 2020;8:e000433. [Crossref] [PubMed]
36. Maute R, Xu J, Weissman IL. CD47-SIRPα-targeted therapeutics: status and prospects. Immunooncol Technol 2022;13:100070. [Crossref] [PubMed]
37. Sato T, Sugiyama D, Koseki J, et al. Sustained inhibition of CSF1R signaling augments antitumor immunity through inhibiting tumor-associated macrophages. JCI Insight 2025;10:e178146. [Crossref] [PubMed]

(English Language Editor: J. Gray)