Preclinical Evaluation of AHT-102, a CLDN18.2 × CD3 Bispecific Antibody: Pharmacokinetics, Anti-Tumor Efficacy, Tissue Distribution, and Safety Profile

Abstract

Background and Objectives

Claudin18.2 (CLDN18.2) is a promising therapeutic target overexpressed in various tumor tissues. While CD3-engaging bispecific antibodies show great potential, their clinical application is often limited by poor efficacy in solid tumors and significant safety risks, such as cytokine release syndrome (CRS). The objective of this study was to comprehensively evaluate the preclinical anti-tumor efficacy, pharmacokinetics, tissue distribution, and safety profile of AHT-102, a novel Fc-free CLDN18.2 × CD3 bispecific antibody with a low-affinity CD3 arm, to determine its potential for clinical development.

Methods

This study evaluated the anti-tumor efficacy of AHT-102 in CLDN18.2-positive gastric cancer mouse models (NUGC4-CLDN18.2). We further assessed its pharmacokinetic characteristics, quantitative tissue distribution using ¹²⁵I-labeling, and long-term toxicity in human CD3EDG transgenic mice.

Results

AHT-102 (0.1, 0.3, and 1 mg/kg) demonstrated significant dose-dependent anti-tumor effects, with tumor weight inhibition reaching 51% at the 1-mg/kg dose. Pharmacokinetic analysis in CD3EDG mice revealed linear characteristics with refined terminal half-lives of 0.762 h, 3.05 h, and 4.25 h for doses of 0.1, 0.5, and 2.5 mg/kg, respectively. Tissue distribution studies confirmed superior targeting specificity; the tumor-to-muscle ratio exceeded 15, and the molecule successfully bypassed the ‘T-cell sink’ by showing lower accumulation in the lymph nodes compared with the tumor. In vitro, AHT-102 did not induce target-independent cytokine release from peripheral blood mononuclear cells (PBMCs). The maximum tolerated dose (MTD) in human CD3EDG mice reached 13.65 mg/kg, a 136.5-fold margin over the projected clinical starting dose. Observed gastric tissue damage was dose-dependent and reversible upon drug discontinuation.

Conclusions

AHT-102, a novel Fab-like bispecific antibody, achieves an optimized therapeutic balance through its low-affinity CD3 arm and Fc-free format. It demonstrates significant anti-tumor efficacy and exceptional targeting specificity while avoiding systemic immunotoxicity typically associated with CD3-targeting agents. These findings provide a robust scientific rationale for the clinical translation of AHT-102 in patients with CLDN18.2-positive cancers.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40268-026-00535-y.

Key Points

AHT-102 is a novel bispecific antibody that effectively targets CLDN18.2-expressing gastric cancer cells while maintaining a favorable safety profile by minimizing excessive immune activation.

Preclinical evaluations demonstrated that AHT-102 provides significant dose-dependent tumor suppression and possesses predictable pharmacokinetic properties in animal models.

The specific tissue targeting and low risk of systemic toxicity observed in this study support the further investigation of AHT-102 as a promising therapeutic candidate for patients with CLDN18.2-positive cancers.

Introduction

In recent decades, significant progress has been made in cancer immunotherapy. However, due to the limited availability of specific antigen targets, high tumor heterogeneity, and the challenging tumor microenvironment (TME), immunotherapy for solid tumors remains a formidable challenge. There is an urgent need for novel therapies with tumor-targeted immune activation properties to address these characteristics.

Claudins (CLDNs) constitute a protein family essential for maintaining tight junctions that regulate molecular exchange between cells. Within this family, CLDN18 exists in two isoforms, CLDN18.1 and CLDN18.2. CLDN18.2, a highly selective biomarker, exhibits limited expression in normal tissues but frequently undergoes abnormal expression in the development of various primary malignancies, such as gastric cancer (GC), pancreatic cancer, esophageal cancer, breast cancer, colorectal cancer, liver cancer, head and neck tumors, bronchial carcinoma, and non-small cell lung cancer [1]. In 2023, the CLDN18.2-targeted drug zolbetuximab released phase III data, demonstrating that combination therapy with chemotherapy significantly improves median progression-free survival (mPFS) and median overall survival (mOS) in previously untreated CLDN18.2⁺/HER2^- locally advanced unresectable or metastatic gastric or gastroesophageal junction (mG/GEJ) adenocarcinoma patients [2].

The CD3 antigen, serving as a crucial marker on the surface of T cells, is expressed on the cell membranes of CD8⁺ T cells and CD4⁺ T cells. It is a protein complex structure composed of four protein chains (CD3γ, CD3δ, CD3ε, CD3ζ) [3]. CD3 is a transmembrane protein on the surface of T cells, which, through non-covalent binding with the T-cell receptor (TCR), forms a receptor complex known as the TCR-CD3 complex. This complex facilitates the transmission of signals triggered by antigen binding into the cell, subsequently activating T cells and exerting cytotoxic effects against tumors [3]. Currently, CD3 serves as a major target in the development of bispecific drugs and has shown certain efficacy in hematologic malignancies. However, its application in solid tumors faces numerous challenges, such as limited T-cell infiltration and the presence of an immunosuppressive tumor microenvironment. These factors collectively restrict its anti-tumor efficacy. Moreover, severe cytokine release syndrome (CRS), is also a limited factor for CD3-bispecific antibody treatments [4, 5]. These challenges significantly diminish the success rate of clinical trials for CD3-targeted antibody drugs, necessitating the urgent development of new technologies. The exploration of tumor-specific targets holds the potential to reduce ‘on-target off-tumor’ toxicity by enabling antibodies to selectively bind to tumors with high target expression while avoiding binding to normal cells with low expression. This approach can significantly enhance the anti-tumor efficacy of CD3 antibodies, minimizing the occurrence of adverse reactions [5].

Utilizing the specific XFab^® platform (non-IgG-like molecules, Chinese patent number: CN113416258A), we designed and constructed the dual-function antibody AHT-102 (Fig. 1). AHT-102 targets both CLDN18.2 antigen and CD3 antigen, allowing the recruitment and activation of T cells in the tumor microenvironment to achieve tumor cell destruction. One advantage of the XFab^® platform is the positioning of the anti-CD3 module at the C-terminus, reducing the affinity for CD3 on T cells and thereby lowering the risk of T-cell non-specific activation and overactivation. Previous studies [6] have indicated that the binding affinity of AHT-102 (ZWB67) to CLDN18.2⁺ cells is 4.3–9.2 times higher than that to CD3⁺ T cells. Co-culturing AHT-102 with CLDN18.2⁺ cells and CD3⁺ T cells demonstrated dose-dependent activation, T-cell cytotoxicity, and cytokine release. AHT-102 also exhibited significant anti-tumor efficacy in the colorectal cancer model [6].

Fig. 1.

Schematic of the CD3/CLDN18.2 XFab^® bispecific antibody (AHT-102). AHT-102 is designed with the anti-hCD3e variable heavy chains and variable light chains fused to anti-hCLDN18.2 Fab frameworks (anti-hCLDN18.2 Fab binds human and mouse CLDN18.2, respectively, and is from Beijing Immunoah Pharma Tech). This fusion is facilitated by an IgG1 hinge region, which includes a deletion mutation at the C266 position, and is situated at the C-terminus of the Fab frameworks. This schematic representation details the structural composition and genetic modifications that enable the dual targeting mechanism of AHT-102, optimizing its efficacy in recognizing and binding to both CD3 on T cells and CLDN18.2 on tumor cells

Here, we described anti-tumor efficacy of AHT-102 in hCLD18.2⁺ gastric cancer models, and evaluated the pharmacokinetics, tissue distribution, and safety profile of AHT-102 in humanized CD3EDG mice.

Materials and Methods

Experimental Animals

Mice possessing humanized CD3 EDG genes (including CD3ε, CD3δ, and CD3γ) on a C57BL/6J genetic background (hCD3EDG mice) (established by Beijing Immunoah Pharma Tech) were procured from Shanghai Sinostone Biotechnology Co., Ltd. Sixty male NSG mice were procured from Shanghai Southern Model Biotechnology Co., Ltd.

All animals were maintained in individually ventilated cages under specific pathogen-free conditions. The controlled environment was set at a consistent temperature range of 20–26 °C, with a relative humidity of 40–70%. A standard 12-h light/12-h dark cycle was established. Mice had ad libitum access to a standard maintenance diet and were provided with sterilized drinking water. The number of animals, experimental design, and animal handling were approved by the Institutional Animal Care and Use Committee (IACUC) of Hubei Tianqin Xinshi Biotechnology Co., Ltd. (Approval No.: IACUC-A2022011-T011-01).

Pharmacokinetic Analysis in hCD3EDG Mice

Following intravenous administration of various doses of AHT-102 in hCD3EDG mice, the pharmacokinetic properties of AHT-102 were assessed by measuring serum drug concentrations. A total of 108 mice were divided into three dosing groups: 0.1, 0.5, and 2.5 mg/kg, with equal distribution of males and females. After a single dose, blood samples were collected at 11 specific time points: pre-dose (0 min), 2 min, 15 min, 30 min, 45 min, 1 h, 2 h, 4 h, 8 h, 12 h, and 24 h post-dose. Approximately 0.05 mL of whole blood was drawn from the posterior tibial vein of the mice at each time point. Upon collection, the samples were transferred to coagulation tubes and temporarily stored in ice boxes. After clotting, the samples were centrifuged (2–8 °C, 2000g, 10 min) to obtain serum, which was stored at temperatures between − 60 and − 90 °C.

AHT-102 concentrations in mouse serum were quantitatively detected using an electrochemiluminescence immunoassay (ECLIA) method based on the MSD platform (details of this method are available in the electronic supplementary material [ESM]). The raw data generated by the plate reader were imported into the Watson LIMS™ v.7.5 SP1 (Thermo Scientific Inc.) system via a software interface. The standard curve was fitted using a four-parameter logistic (4PL) model (Marquardt), with a weighting factor of 1/Y². Pharmacokinetic (PK) parameters were calculated using the non-compartmental model in Phoenix WinNonlin v.8.2. The fitting method was set to ‘best fit,’ with a weighting factor of ‘Uniform,’ and the calculation method employed was linear trapezoidal linear interpolation. For the determination of the linear relationship between pharmacokinetic parameters and dose, a power function model was applied using Phoenix WinNonlin v.8.2 to calculate the 90% confidence interval of the slope after linear regression. The conclusions were drawn as follows: (a) If the confidence interval is entirely within the judgment range, the PK parameters are linearly related to the dose. (b) If the confidence interval falls entirely outside the judgment range, there is no linear relationship between the PK parameters and the dose. (c) If the confidence interval overlaps with the judgment range, no definitive conclusion can be drawn.

The formula for calculating the judgment range (for AUC, θ_L = 0.8, θ_H = 1.25; for C_max, θ_L = 0.7, θ_H = 1.43) is as follows:

$$1+\frac{\text{ln}{(\theta }_L)}{\text{ln}(r)}<\beta <1+\frac{\text{ln}{(\theta }_H)}{\text{ln}(r)},$$

where H represents the highest administered dose, L represents the lowest administered dose, r = H/L, and θ_H and θ_L represent the upper and lower limits of the confidence interval, respectively.

In Vivo Efficacy Study in Mouse Models of Gastric Cancer

We evaluated the antitumor activity of AHT-102 in a subcutaneous xenograft mouse model using human gastric cancer NUGC4-CLDN18.2 cells complemented with a humanized peripheral blood mononuclear cell (PBMC) immune system. The NUGC4-CLDN18.2 cells, overexpressing human Claudin18.2, were cultured in RPMI-1640 medium (Meilunbio, MA0215-Nov-04H) supplemented with 10% heat-inactivated fetal bovine serum (Excell Bio, 12A222), penicillin 100 U/mL, and streptomycin 100 µg/mL (Solarbio, 20221231), provided by Yi Kang (Beijing) Pharmaceutical Technology Co., Ltd. The cell-line NUGC-4 was purchased from Biovector NTCC Inc. Male NSG mice were subcutaneously inoculated on the right anterior flank with 1 × 10⁷ NUGC4-CLDN18.2 cells in 100 μL. One day following implantation, humanized PBMCs (2 × 10⁶ cells per mouse; Donor #: P121101205C, Beijing Rudebaiao Biotechnology Co., Ltd) were administered intraperitoneally.

Once tumors reached an average volume of approximately 53 mm³, mice (n = 60) were randomized into four cohorts (n = 10): vehicle control and three AHT-102 dose groups (0.1, 0.3, and 1.0 mg/kg). Treatment was given intravenously daily for 3 days, followed by a 3-day break, for a total of four cycles. Tumor volume (V) and body weight were recorded twice weekly. When the tumor volume exceeded 2000 mm³, the mice were euthanized and excluded from the experiment. Tumor volume was calculated as follows: V = length (mm) × width (mm)² × 0.5.

To evaluate the efficacy, the relative tumor volume (RTV) was calculated as V_t/V₀, where V_t is the volume at a specific time point and V₀ is the volume at the start of treatment. The T/C (%) value was determined as the ratio of the mean RTV of the treatment group to the mean RTV of the vehicle group × 100%. The tumor growth inhibition rate (TGI_TV, %) was calculated using the formula: TGI_TV (%) = (1 − T/C) × 100%.

At the end of the study (PG-D21), animals were euthanized and tumors were harvested, weighed, and photographed. The tumor weight inhibition rate (TGI_TW, %) was calculated as: TGI_TW (%) = [1 − (mean tumor weight of the treatment group / mean tumor weight of the vehicle group)] × 100%.

Statistical analysis of tumor volume was performed using IBM SPSS Statistics 25.0 software, applying one-way ANOVA to assess intergroup differences. A p value of <0.05 was considered indicative of a statistically significant difference.

Tissue Distribution Study of ¹²⁵I-labeled AHT-102 in Colorectal Cancer Models

To evaluate tissue distribution of AHT-102, colorectal cancer models were established in 24 hCD3EDG mice by inoculating 5 × 10⁶ MC38-hCLDN18.2 cells (maintained in RPMI-1640 medium supplemented with 10% fetal bovine serum [FBS] and puromycin 5 µg/mL) (Kyinno Biotechnology Co., Ltd, KC-1449), with an equal gender distribution. The investigational agent AHT-102 was labeled with ¹²⁵I (McMaster) using the Iodogen method [7]. When the tumor volume of each mouse exceeded 150 mm³, they were randomly divided into four groups based on body weight, with one group at each time point. A single intravenous dose of ¹²⁵I-AHT-102 2.5 mg/kg was administered, and blood, urine, and tissue samples were collected at 2 min, 4 h, 8 h, and 12 h. Urine and blood samples were centrifuged (room temperature, 2500 relative centrifugal force [RCF], 10 min). After weighing, 400 µL of 20% TCA (Adamas, P1864099) was added to the tissue, urine supernatant, and serum sample tubes, and mixed thoroughly. Following centrifugation, the total radioactivity and precipitate radioactivity were measured using a γ-counter (automatic gamma counter, 3470 WIZARD2, PerkinElmer). Radioactivity equivalent concentrations (Equ ng/g) in serum, urine, and tissues were calculated based on γ counting results.

Equ ng/g = Sample count value/Decay constant/Specific activity of the test substance/Mass of serum (urine, tissue)

In Vitro Cytokine Release Assay

AHT-102 at concentrations of 0.0001, 0.001, 0.01, 0.1, 1, and 10 µg/mL was co-cultured with PBMCs (Heyiou Biotechnology Co., Ltd., 013K009) (Shanghai Aoneng Biotechnology Co., Ltd, LP230216003, LP200706) at a density of 2×10⁵ cells in RPMI-1640 medium supplemented with 10% FBS, 1% Glutamax (Gibco, 35050-061), and 1% penicillin/streptomycin. The cells were incubated for 2, 4, 24, and 48 h. These time points (2, 4, 24, and 48 h) were selected based on preliminary activation marker assays and the known secretion kinetics of early, peak, and late-phase cytokines. After incubation, cell culture supernatants were collected. Serum-free culture medium with T cells served as a blank control, and PHA 2 µg/mL (Sigma, L2769) was used as a positive control. The concentrations of cytokines (interleukin [IL]-2, IL-4, IL-6, IL-8, IL-10, interferon [IFN]-γ, and tumor necrosis factor [TNF]-α) in the cell culture supernatant were measured using the Luminex® Assay (Luminex 200 analyzer, Luminex) and assay kit (Cat.:, Company). The data were analyzed using the Analysis of Variance (ANOVA) method within IBM SPSS Statistics 25.0 software. Statistical differences between groups were considered significant at p < 0.05, highly significant at p < 0.01, and extremely significant at p < 0.001.

Toxicity Study in Healthy hCD3EDG Mice

A total of 654 hCD3EDG mice, evenly divided between males and females, were randomly assigned into vehicle control, AHT-102 low-dose, medium-dose, and high-dose groups based on body weight. Each AHT-102 dose group consisted of 174 mice/group (120 for the main study, 48 for toxicokinetic satellite groups, and 6 additional backup animals for each toxicokinetic satellite group). The control group comprised 132 mice (120 for the main study, 8 for toxicokinetic satellite groups, and 4 additional backup animals). The AHT-102 low, medium, and high doses were 1.50, 4.50, and 13.65 mg/kg, respectively, with a dosing volume of 15 mL/kg. The control group received an equivalent volume of solvent (0.9% sodium chloride injection). Intravenous administration was performed once a week for 5 consecutive weeks, with a 6-week recovery period after the last dose.

The animals were monitored for mortality, clinical signs, local irritation, body weights, food consumption, and ophthalmic examinations. Laboratory investigations included clinical pathology (hematology, coagulation, serum chemistry, and urinalysis), histopathology, and toxicokinetics. Samples were collected at the end of the dose period (R1) and the end of the recovery phase (R43). Tissues were fixed with formaldehyde and stained with H&E. Toxicokinetic blood samples were collected at various time points: before the first and last dose, 15 min, 2 h, 6 h, 24 h, 48 h, 96 h, and 168 h after the first and last dose for each AHT-102 dose group. For the control group, samples were collected before and 24 h after the first and last dose of administration.

All quantitative indicators were statistically analyzed using SPSS 25.0 software. Initially, Levene's test for homogeneity of variance was applied. When the variance was homogeneous (p > 0.05), the results of the ANOVA could be directly referenced to determine if the overall difference between groups was statistically significant. If the overall difference was statistically significant (p < 0.05), the Dunnett t test was employed to compare differences between groups. Conversely, if the overall difference was not statistically significant (p ≥ 0.05), the statistical analysis was concluded. In cases where Levene's test indicated that the variance was not homogeneous (p < 0.05), a non-parametric test (Kruskal–Wallis H test) was used. If the Kruskal–Wallis H test revealed a statistically significant overall difference (p < 0.05), the Mann–Whitney U test was applied to compare differences between groups. However, if the Kruskal–Wallis H test indicated no statistically significant overall difference (p ≥ 0.05), the statistical analysis was concluded.

Results

Pharmacokinetics of AHT-102

We evaluated the pharmacokinetics of AHT-102 in hCD3EDG mice (Table 1). Single intravenous injections of 0.1, 0.5, and 2.5 mg/kg AHT-102 in hCD3EDG mice resulted in dose ratios of 1:5:25, maximum concentration (C_max) ratios of 1:4.76:25.1, and area under the plasma concentration–time curve from zero to the time of the last quantifiable concentration (AUC_(0-t)) ratios of 1:5.18:26.4 (Fig. 2). The C_max and AUC_(0–t) ratios were close to the dose ratios. The half-life of the 0.1, 0.5, and 2.5-mg/kg dose groups was 0.762 h, 3.05 h, and 4.25 h, respectively, with similar clearances rates among the dose groups (78.2–81.8 mL/h/kg). Upon dose-proportional analysis, the pharmacokinetic parameters demonstrated linear characteristics within the dose range of 0.1–2.5 mg/kg. After a single intravenous injection of AHT-102 0.1, 0.5, and 2.5 mg/kg in hCD3EDG mice, the apparent volume of distribution (Vd, range across dose groups: 89.9–705 mL/kg) was greater than the plasma volume of mice (approximately 50 mL/kg), suggesting that AHT-102 also distributes outside the blood vessels.

Table 1.

Pharmacokinetic parameters of different doses of AHT-102 after a single intravenous injection in hCD3EDG mice

Parameters	Unit	AHT-102 (0.1 mg/kg)	AHT-102 (0.5 mg/kg)	AHT-102 (2.5 mg/kg)
Kel	1/h	0.910	0.228	0.111
t½	h	0.762	3.05	4.25
Tmax	h	0.0333	0.0333	0.0333
Cmax	ng/mL	2330	11,100	58,400
AUC(0–t)	h*ng/mL	1210	6270	31,900
AUCinf	h*ng/mL	1220	6330	32,000
AUCextr	%	1.45	0.958	0.391
Vd	mL/kg	89.9	347	705
CL	mL/h/kg	81.8	79.0	78.2
MRTinf	h	0.676	1.01	1.04

AUC_(0–t) area under the plasma concentration–time curve from zero to the time of the last quantifiable concentration, AUC_extr percentage of AUC extrapolated from the last measurable concentration to infinity, AUC_inf AUC from time zero to infinity, CL total body clearance, C_max maximum plasma concentration, Kel terminal elimination rate constant, MRT_inf mean residence time from time zero to infinity, t_½ terminal half-life, T_max time to C_max, V_d volume of distribution.

Fig. 2.

Absorption of AHT-102 in hCD3EDG mice. Mean plasma concentration–time curves (mean ± SD, n = 12) following a single intravenous injection of 0.1-mg/kg, 0.5-mg/kg, and 2.5-mg/kg doses of AHT-102 in human CD3EDG transgenic mice. Blood samples were collected at 11 specific time points: pre-dose (0 min), 2 min, 15 min, 30 min, 45 min, 1 h, 2 h, 4 h, 8 h, 12 h, and 24 h post-dose

AHT-102 Demonstrated Significant Anti-tumor Efficacy in Humanized PBMC GC Models

To evaluate the inhibitory effect of AHT-102, we utilized a humanized PBMC immune system gastric cancer (NUGC4-CLDN18.2) CDX mouse model. AHT-102 was administered intravenously at three doses (0.1, 0.3, and 1.0 mg/kg) once daily in four cycles (3 days on/3 days off).

The results demonstrated that AHT-102 significantly inhibited tumor growth in a dose-dependent manner. At the end of the study (PG-D21), the tumor growth inhibition rates based on volume (TGI_TV) for the 0.1, 0.3, and 1.0-mg/kg groups were 16%, 34%, and 47%, respectively. The tumor volumes in all AHT-102 treated groups were significantly lower than those in the vehicle group (p < 0.01; Fig. 3a). Statistical analysis further revealed that the inhibitory effect in the high-dose group was significantly superior to that in the medium-dose group, which in turn was significantly superior to the low-dose group (p < 0.01).

Fig. 3.

AHT-102 demonstrates significant antitumor efficacy in mouse models of gastric cancer (GC). a Tumor growth inhibition in humanized PBMC immune system mice subcutaneously implanted with human gastric cancer NUGC4-CLDN18.2 cells. AHT-102 was administered intravenously once daily (3 days on, followed by 3 days off for each cycle) for 4 cycles (mean ± SD, n = 10). **p < 0.01 compared with the vehicle control group. b Statistical analysis of tumor weights harvested at the end of the study (PG-D21). AHT-102 exhibited dose-dependent inhibition of tumor weight, reaching a maximum inhibition of 51% in the 1.0-mg/kg group (mean ± SD, n = 10). **p < 0.01 compared with the vehicle group. c Representative photograph of the harvested tumors from the vehicle and AHT-102 treatment groups (0.1, 0.3, and 1.0 mg/kg) at the end of the experiment (PG-D21). d Body weight changes across all groups during the study period. A downward trend in body weight was observed in all cohorts as time progressed, attributed to PBMC-induced graft-versus-host disease (GvHD) (mean ± SD, n = 10). No statistically significant differences were observed between the AHT-102 treated groups and the vehicle group (p > 0.05). PBMC peripheral blood mononuclear cell

Consistent with the tumor volume data, the tumor weight inhibition rates (TGI_TW) at study termination were 20%, 39%, and 51% for the low, medium, and high-dose groups, respectively. The tumor weights in all treatment groups were significantly reduced compared with the vehicle group (p < 0.01; Fig. 3b/c).

During the treatment period, the body weight of mice in each group showed a decreasing trend over time (Fig. 3d). This weight loss was attributed to graft-versus-host disease (GvHD) typically induced by humanized PBMCs rather than drug-specific toxicity, as the trend was observed across all cohorts including the vehicle. Notably, no animal mortality was observed in any group throughout the study period, representing a 100% overall survival rate (Fig. S1 in the ESM).

Distribution of AHT-102 in hCD3EDG Mice

Following a single intravenous injection of ¹²⁵I-AHT-102 2.5 mg/kg in the hCD3EDG mouse colorectal cancer model, the exposure levels (AUC_last) in various tissues were ranked as follows: serum > kidney > lung > stomach > liver > tumor > spleen > adrenal gland > heart > esophagus > duodenum > femur > ovaries > bladder > intestinal contents (duodenum) > thymus > pancreas > testes > skeletal muscle > mesenteric lymph node > gallbladder > body fat > eyeball > brain. These results indicate that AHT-102 is primarily distributed in the systemic circulation, followed by major excretory organs (kidney), blood-perfused tissues (lung, liver), and antigen-expressing target tissues (stomach, tumor).

The pharmacokinetic profiles in major tissues revealed a distinct divergence between target and non-target organs (Fig. 4). In blood-rich tissues, such as the kidney, lung, and liver, concentration–time curves followed a trend similar to the serum, peaking at 2 minutes and declining rapidly thereafter. In contrast, target tissues—including the CLDN18.2-positive stomach and tumor—exhibited an active sequestration process. Concentrations in these tissues reached C_max at 4 h post-dosing, significantly later than in the circulation, suggesting specific binding and accumulation.

Fig. 4.

Distribution of AHT-102 in hCD3EDG mice. Time–concentration curves of precipitated radioactivity (mean ± SD, n = 6) in serum and major tissues following a single administration of I¹²⁵-AHT-102 2.5 mg/kg in human CD3EDG transgenic mouse models of colorectal cancer. Samples were collected at 2 min, 4 h, 8 h, and 12 h post-dose

To further evaluate the targeting specificity, the tumor-to-blood (T/B) ratio was calculated (Table 2). The T/B ratio increased progressively from 0.01 at 2 minutes to 3.09 at 12 h, representing a 300-fold enrichment of the drug within the tumor relative to the serum over time. At the 12-h terminal point, the drug concentration in the tumor (495 Equ ng/g) was substantially higher than in vital organs like the heart (32.8 Equ ng/g) and muscle (32.6 Equ ng/g), with a tumor-to-heart ratio of 15.1. These quantitative metrics, combined with the minimal distribution observed in the brain (indicating no penetration of the blood–brain barrier), demonstrate that AHT-102 possesses favorable tumor-targeting specificity and a safe biodistribution profile without significant long-term accumulation in non-target tissues.

Table 2.

Quantitative targeting ratios of ¹²⁵I-AHT-102 in tumor-bearing mice

Ratio category	Parameter	2 min	4 h	12 h
Targeting efficiency	Tumor/serum	0.01	1.83	3.09
Selectivity vs organs	Tumor/heart	0.15	5.21	15.1
	Tumor/liver	0.10	1.83	4.96
	Tumor/lung	0.07	2.03	3.09
	Tumor/muscle	2.19	5.19	15.18
	Tumor/kidney	0.06	0.54	2.31
	Tumor/brain	1.19	29.4	69.0
	Tumor/ovary	0.23	5.47	11.8
	Tumor/spleen	0.22	1.51	2.86
Endogenous target	Stomach/serum	0.02	2.73	2.08

AHT-102 Demonstrated No Non-Specific Cytokine Release In Vitro

Next, we investigated whether AHT-102 could non-specifically stimulate PBMCs to release cytokines in the absence of CLDN18.2⁺ cells. The results demonstrated that when co-cultured with PBMCs from three donors, the levels of IL-2, IL-4, IL-6, IL-8, IL-10, IFN-γ, and TNF-α in the cell culture supernatant did not increase with either incubation time or AHT-102 concentration (Fig. 5). This indicates that in the absence of target cells, AHT-102 does not induce cytokine release from PBMCs.

Fig. 5.

AHT-102 does not induce non-specific activation of PBMCs. PBMCs from three healthy volunteers were co-cultured with AHT-102 at concentrations of 0.0001, 0.001, 0.01, 0.1, 1, and 10 µg/mL and incubated for 2 h, 4 h, 24 h, and 48 h. Serum-free culture medium with PBMCs served as a blank control, and 2 µg/mL PHA was used as a positive control. The cytokine levels (IL-2, IL-4, IL-6, IL-8, IL-10, IFN-γ, and TNF-α) in the cell culture supernatant were measured. IFN-γ interferon gamma, IL interleukin, PBMCs peripheral blood mononuclear cells, TNF-α tumor necrosis factor alpha

AHT-102 Demonstrated Good Tolerance in hCD3EDG Mice

Finally, we conducted a comprehensive evaluation of the safety profile of AHT-102 in healthy hCD3EDG mice. The experimental mice were administered AHT-102 intravenously at doses of 1.50, 4.50, and 13.65 mg/kg, respectively. The treatment regimen consisted of weekly administrations for a total of 5 weeks, followed by a 6-week recovery period. Notably, the MTD of AHT-102 was established at 13.65 mg/kg, at which point the average AUC_0–168h post-final administration reached 236 h*μg/mL, with a mean C_max of 158 μg/mL. There were no discernible gender differences in the systemic exposure (AUC_0–t and C_max) of AHT-102 across dose groups following both the initial and final administrations, and no evidence of accumulation was observed. Throughout the duration of the study, no instances of drug-related mortality or severe morbidity were recorded.

Comprehensive monitoring encompassed general observations, food consumption patterns, body weight measurements, gross pathological assessments, ophthalmic examinations, hematological parameters, coagulation status, and urinalysis, all of which failed to reveal any drug-related abnormalities. Furthermore, no irritation was observed at the site of administration.

Blood biochemistry analyses (Table 3 ) following the final administration revealed dose-dependent decreases in normal albumin (ALB) and total protein (TP) levels in female mice from the mid- and high-dose groups, as well as reductions in ALB and albumin/globulin (A/G) ratios in male mice. However, these changes remained within the normal range and did not significantly differ from the vehicle control group at the end of the recovery period (Table 4).

Table 3.

Blood biochemistry parameters (mean ± SD) following 5-week repeated intravenous administration of AHT-102 in hCD3EDG mice on day 1 of the recovery period

Parameters	Unit	Sex	Test date	Vehicle	AHT-102 (1.5 mg/kg)	AHT-102 (4.5 mg/kg)	AHT-102 (13.65 mg/kg)
ALB	g/L	Female	R1	36.3 ± 1.42	35.4 ± 1.15	34.1 ± 0.94**	33.9 ± 1.23**
TP	g/L	Female	R1	42.8 ± 1.65	41.9 ± 1.36	40.2 ± 0.97**	40.0 ± 1.95**
ALB	g/L	Male	R1	40.2 ± 1.45	38.1 ± 1.70	37.5 ± 3.07*	37.9 ± 1.04*
A/G		Male	R1	3.85 ± 0.313	3.32 ± 0.241**	3.64 ± 1.767*	3.23 ± 0.226**

R recovery phase, n = 10

A/G albumin/globulin ALB normal albumin, TP total protein

*Indicating statistical significance compared with vehicle control group of p < 0.05

**Indicating statistical significance compared with vehicle control group of p < 0.01

Table 4.

Blood biochemistry parameters (mean ± SD) following 5-week repeated intravenous administration of AHT-102 in hCD3EDG mice on day 43 of the recovery period

Parameters	Unit	Sex	Test date	Vehicle	AHT-102 (1.5 mg/kg)	AHT-102 (4.5 mg/kg)	AHT-102 (13.65 mg/kg)
ALB	g/L	Female	R43	37.2 ± 0.66	36.9 ± 1.60	36.7 ± 0.54	36.9 ± 0.63
TP	g/L	Female	R43	44.0 ± 1.31	44.2 ± 1.72	43.8 ± 1.22	44.5 ± 0.86
ALB	g/L	Male	R43	34.9 ± 1.07	35.3 ± 0.94	35.2 ± 0.86	35.4 ± 0.78
A/G		Male	R43	3.78 ± 0.224	3.75 ± 0.375	4.19 ± 0.373	3.95 ± 0.474

R recovery phase, n = 5

A/G albumin/globulin ALB normal albumin, TP total protein

Histopathological examination highlighted the induction of adverse reactions within the stomach tissue by AHT-102 (Fig. 6). Microscopic findings encompassed mild to severe enlargement of mucous cells, mild to moderate diffuse regenerative hyperplasia of the mucosa, minimal single-cell necrosis within the mucosa, mild to moderate neutrophil infiltration in the submucosa/basal layer, and mild to moderate reactive squamous epithelial hyperplasia. Notably, female mice from the middle-/high-dose groups and male mice from the high-dose group exhibited mild focal erosion/ulceration. Encouragingly, all aforementioned pathological alterations had fully resolved by the end of the recovery period.

Fig. 6.

Treatment with AHT-102 is well tolerated in human CD3EDG transgenic mouse. The stomach sections in the 5-week hCD3EDG mouse toxicity study were stained with H&E and examined microscopically (4 ×)

Discussion

AHT-102 (ZWB67) represents a novel CLDN18.2 × CD3 bispecific T-cell engager (TCE) engineered via the XFab^® platform. Its design philosophy prioritizes a favorable safety-to-efficacy ratio through two distinct structural features: an Fc-free format and a reduced-affinity CD3 arm.

The Rationale for an Fc-Free Design

The necessity of an Fc region in TCEs remains a subject of therapeutic strategy. While the Fc fragment enhances molecular stability and extends circulation time via FcRn-mediated recycling, it poses significant risks in the context of CD3-targeted therapies. Fc-mediated effector functions can trigger off-target immune activation and cytokine release syndrome (CRS) through binding to Fc receptors on non-target cells. Historically, catumaxomab’s clinical utility was hampered by high immunogenicity and systemic toxicity attributed to its fully functional Fc portion [8, 9]. Furthermore, research by Wang et al. demonstrated that intact Fc domains may actually sequester T cells in the periphery, hindering their infiltration into subcutaneous solid tumors [10].

By adopting the Fc-free XFab^® platform, AHT-102 minimizes these risks. Despite a shorter systemic half-life typical of Fab-type antibodies (~ 80 kDa), the reduced molecular size facilitates superior tumor tissue penetration. This is evidenced by our finding that several TCE molecules based on this platform, including a DC-T cell engager, have successfully progressed into clinical trials [11].

Optimized Safety and Controlled Activation

The systemic toxicity of CD3-bispecific antibodies, often manifesting as dose-limiting toxicities, has historically constrained their clinical development [12–14]. AHT-102 addresses this through structural constraints; its CD3-binding domain is positioned at the C-terminus, where the upstream hinge region restricts the flexibility of the variable region. This design effectively reduces its affinity for CD3 on T cells, preventing excessive systemic activation.

Our in vivo results corroborate this safety profile. In hCD3EDG mice, AHT-102 was well tolerated up to 13.65 mg/kg—a dose 136.5 times the proposed clinical starting dose—with no significant immunotoxicity or systemic abnormalities. Although histopathological changes were observed in the stomach, this was an expected on-target effect due to endogenous CLDN18.2 expression, a phenomenon also observed with zolbetuximab [16]. However, the absence of an Fc domain in AHT-102 further mitigates the risk of cytokine storms, making the associated immunotoxicity highly manageable.

Targeted Distribution and the ‘T-Cell Sink’ Effect

A critical challenge for TCEs is avoiding the ‘T-cell sink,’ where high-affinity CD3 binding leads to drug sequestration in T-cell-rich secondary lymphoid organs. AHT-102’s CD3ε affinity (K_D ≈ 10.4 nM) is strategically lower than its CLDN18.2 affinity. As proposed by Mandikian et al., this affinity disparity allows the molecule to bypass peripheral T cells and preferentially accumulate in target-positive tumors [15].

Our tissue distribution studies provided robust quantitative validation of this targeting strategy. Unlike non-target tissues where concentrations followed the rapid decline of the serum, CLDN18.2-positive stomach and tumor tissues exhibited an active enrichment phase, reaching C_max at 4 h post-administration. Notably, the T/B ratio increased 300-fold over 12 h (from 0.01 to 3.09), and the tumor concentration was 15-fold higher than in the heart or skeletal muscle. Furthermore, the lower exposure levels in the spleen and mesenteric lymph nodes compared with the tumor confirm that AHT-102 preferentially targets CLDN18.2+ tissues over T-cell-rich regions.

Functional Validation of Anti-tumor Efficacy

Although direct immunohistochemical (IHC) evidence of T-cell infiltration was not collected, the observed dose-dependent tumor suppression (TGI_TW of 51%) serves as strong functional validation of T-cell recruitment. According to our prior characterization [6], AHT-102 facilitates the formation of a localized ‘cytolytic synapse,’ triggering CD69 expression and T_h1 cytokine release. The observed in-vivo efficacy suggests that the optimized CD3 affinity effectively promotes the redistribution of activated T cells from the circulation into the solid tumor bed [15].

Synergistic Potential and Clinical Outlook

The significant anti-tumor activity and safety profile of AHT-102 provide a strong rationale for combination regimens. The phase III SPOTLIGHT and GLOW trials established CLDN18.2-targeted therapy (zolbetuximab) plus chemotherapy as a new standard of care [2, 17]. Beyond chemotherapy, combining TCEs with immune checkpoint inhibitors (ICIs) is particularly compelling. CLDN18.2-targeted agents can modulate the tumor microenvironment by promoting CD8⁺ T-cell infiltration [18]. Adding PD-1/PD-L1 inhibitors may prevent T-cell exhaustion, which often limits TCE monotherapy. Such strategies could be transformative for patients with MSS/pMMR subtypes, who traditionally show poor responses to ICIs.

Conclusion

The Fc-free design and optimized CD3 affinity of AHT-102 successfully balance potent tumor suppression with a superior safety profile. Its distinct targeting specificity and manageable immunotoxicity position AHT-102 as a promising candidate for the treatment of CLDN18.2-positive malignancies, both as a monotherapy and in synergistic combination strategies.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

The study was funded by Guangxi Ardeon Therapeutics Co., Ltd.

Declarations

Ethics approval

All animal experiments were conducted in strict compliance with the American Veterinary Medical Association (AVMA) Guidelines for the Euthanasia of Animals (2020 Edition) and the Laboratory Animal Guidelines for Euthanasia from China (standard number: GB/T 397602021). Throughout the study, we adhered to the 3R principles, emphasizing the reduction in the number of animals used, the replacement of animals where possible, and the refinement of experimental conditions and procedures to minimize distress or pain. The number of animals, experimental design, and animal handling were approved by the Institutional Animal Care and Use Committee (IACUC) of Hubei Tianqin Xinshi Biotechnology Co., Ltd. (Approval No.: IACUC-A2022011-T011-01).

Consent to participate

Not applicable. This study did not involve human participants; human peripheral blood mononuclear cells used in this study were obtained from commercial sources.

Consent for publication

Not applicable.

Availability of data and material

Data are available upon reasonable request.

Conflict of interest

The authors declare that they have no competing interests.

Author contributions

Conception and design: Huilun Chu, Junping Lv, Haifeng Song. Antibody development: Guili Xu, Niliang Qian, Xin gao. In vivo efficacy study: Huilun Chu, Junping Lv. Pharmacokinetic study: Yunlong Liu, Lun Ou, Huilun Chu. In vitro study: Yujie Liu, Xiujie Pan, Guili Xu, Huilun Chu. Toxicity Study: Yanchuan Li, Huilun Chu. Analysis and interpretation of data: all authors. Writing of the manuscript: Huilun Chu, Guili Xu, Xin gao. All authors have read and approved the final version of the manuscript and consent to its publication.

Code availability

Not applicable.

Data availability and materials

The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request. For inquiries regarding data and materials, please contact Huilun Chu at Guangxi Ardeon Therapeutics Co., Ltd. (email: chuhl@ardeontx.com).