Stomach Cancer

• NCI-N87: one of the most classic HER2-positive gastric cancer models, suitable for HER2-targeted agents, HER2 ADCs, HER2 bispecifics, trastuzumab combination chemotherapy, and later-line therapy research.
• NCI-N87-luc: suitable for orthotopic gastric wall models, relapse models, and metastasis models.

• MKN45: a widely used base model for CLDN18.2 direction, suitable for CLDN18.2 antibodies, ADCs, bispecifics, CAR-T/NK, and peritoneal dissemination research.
• MKN45-luc: suitable for orthotopic gastric wall, peritoneal dissemination, and liver metastasis models.
• NUGC-4: commonly used for CLDN18.2-high expression and orthotopic/dissemination-related research, suitable for validating CLDN18.2 membrane expression stability and peritoneal lesion accessibility.

• SNU-16: suitable for FGFR2b antibody, bispecific, ADC, and enriched gastric cancer program early screening.
• KATO III: can serve as an FGFR2b direction supplementary model for target expression stability, combination therapy, and in vivo efficacy validation.

• HGC-27: suitable for high-invasion, poorly differentiated, migration, EMT, and advanced dissemination research.
• MKN45: combines CLDN18.2 direction value and dissemination capacity, suitable for peritoneal dissemination, gastric wall orthotopic, and liver metastasis early platforms.

• Murine gastric cancer syngeneic models: suitable for immune checkpoint, STING, cytokines, perioperative immunotherapy, and TME remodeling research.
• Human gastric cancer cell lines + huPBMC / huHSC: suitable for CLDN18.2 CAR-T, TCE, bispecifics, human cytokines, and human immune-dependent programs.

• HER2: NCI-N87.
• CLDN18.2: MKN45, NUGC-4.
• FGFR2b: SNU-16, KATO III.
• Target expression should be re-verified in vitro, after tumor formation, and at the endpoint stage; CLDN18.2 programs should focus on membrane expression quality rather than total protein expression alone.

Subcutaneous CDX typically involves inoculating human gastric cancer / esophagogastric junction adenocarcinoma cells subcutaneously into immunodeficient mice, followed by randomization and dosing once tumors reach a predetermined volume. Commonly used hosts include BALB/c nude, NOD-SCID, and NSG mice.

• HER2-positive subcutaneous CDX: NCI-N87, NCI-N87-luc
• CLDN18.2-high subcutaneous CDX: MKN45, NUGC-4
• FGFR2b subcutaneous CDX: SNU-16, KATO III
• Diffuse-type and high-invasion subcutaneous CDX: HGC-27, MKN45

Model characteristics
NCI-N87 is suitable for HER2-targeted agents, HER2 ADCs, bispecifics, and combination chemotherapy early efficacy screening. MKN45 or NUGC-4 are suitable for CLDN18.2 antibody, ADC, bispecific, and cell therapy program early validation, with MKN45 more suitable for subsequent escalation to orthotopic, peritoneal dissemination, and liver metastasis models. SNU-16 or KATO III can serve as early screening platforms for FGFR2b antibodies, bispecifics, or ADCs. HGC-27 or MKN45 are suitable for high-invasion, poorly differentiated, migration, EMT, and advanced dissemination direction early screening. This model has high standardization and reproducibility, suitable for early screening and PK/PD research; its limitations are the lack of gastric wall microenvironment, peritoneal dissemination, liver metastasis, local invasion, and postoperative relapse processes.

Modeling and dosing timelines
Subcutaneous inoculation typically reaches enrollment volume within 1–3 weeks, with randomization commonly performed at tumor volumes of approximately 80–150 mm³. The conventional treatment period is 2–4 weeks. For resistance, withdrawal rechallenge, long-term targeted suppression, ADC interval dosing, or immune combination pre-treatment studies, this can be extended to 4–8 weeks. HER2, CLDN18.2, and FGFR2b programs should re-verify target expression before modeling, at enrollment, and at the endpoint stage.

Detection endpoints

• Basic efficacy layer: TGI, RTV, terminal tumor weight, body weight, general condition, and progression-free time
• Molecular stratification layer: HER2, CLDN18.2, FGFR2b, PD-L1, MSI/dMMR-related markers, Ki67, and target retention
• Mechanism layer: cleaved caspase-3, TUNEL, CD31, necrosis ratio, EMT, drug internalization, ADC tissue distribution, and target pathway inhibition
• PK/PD layer: Plasma drug concentration, tumor tissue concentration, target occupancy, pathway inhibition, ADC distribution, ctDNA, and exposure-efficacy relationship

Recommended positive controls
Fluorouracil at 20–50 mg/kg IP, 1–3 times weekly; oxaliplatin at 5–10 mg/kg IP or IV, once weekly. Trastuzumab at 5–20 mg/kg IP or IV, 1–2 times weekly; trastuzumab deruxtecan can be explored at 3–10 mg/kg IV interval dosing. Zolbetuximab at 5–20 mg/kg IV or IP, once weekly or adjusted by half-life. Bemarituzumab-class FGFR2b antibody at 5–20 mg/kg IV or IP.

Human gastric wall orthotopic models typically use MKN45-luc, NUGC-4-luc, SNU-16-luc, or NCI-N87-luc inoculated or implanted into the gastric wall of NSG or NOD-SCID mice. Bioluminescent labeling enables orthotopic tumor burden and subsequent dissemination tracking. Luc-labeled cells or small tumor fragments can be implanted into the gastric wall, with Matrigel or suture fixation as needed to reduce cell leakage and improve modeling consistency.

Model characteristics
MKN45-luc and NUGC-4-luc are more suitable for CLDN18.2, peritoneal dissemination, and relapse directions; SNU-16-luc is more suitable for FGFR2b programs; NCI-N87-luc is more suitable for HER2 targeting, HER2 ADC, and local efficacy evaluation. The gastric wall orthotopic model is closer to the real organ environment than subcutaneous models, suitable for local invasion, relapse, and spontaneous metastasis research. Its limitations are complex surgery, lower throughput, greater variability, and strong dependence on BLI, ultrasound, or CT monitoring.

Modeling and dosing timelines
Gastric wall orthotopic models typically form monitorable burden within 1–3 weeks after modeling, depending on cell line, implantation method, and host background. BLI/IVIS is recommended for enrollment stratification. The conventional treatment period is 2–6 weeks. If studying postoperative relapse, perioperative therapy, or spontaneous metastasis, observation can be extended. Treatment can begin after BLI signal confirmation, or after orthotopic burden reaches a predetermined standard, followed by local resection, local dosing, or perioperative intervention.

Detection endpoints

• Orthotopic burden layer: BLI/IVIS, orthotopic tumor burden, gastric wall pathology, endpoint tumor weight, survival, and body weight
• Local invasion layer: Gastric wall invasion depth, serosal breakthrough, perigastric metastasis, local blood supply, local necrosis, and tissue pathology
• Dissemination and metastasis layer: Peritoneal dissemination, ascites, liver metastasis, lymph node metastasis, ctDNA, and survival
• Target and mechanism layer: HER2, CLDN18.2, FGFR2b, Ki67, cleaved caspase-3, CD31, PD-L1, and drug tissue distribution

Recommended positive controls
Fluorouracil or capecitabine combined with oxaliplatin as general chemotherapy reference; if simulating perioperative therapy, docetaxel can be added to form a FLOT-like split regimen with dose adjusted by tolerance. HER2-positive orthotopic models can set trastuzumab, with trastuzumab deruxtecan for later-line programs. CLDN18.2-positive orthotopic models can set zolbetuximab. Perioperative immunotherapy combination can reference the durvalumab plus FLOT clinical pathway.

Peritoneal dissemination models typically use MKN45-luc, NUGC-4-luc, or SNU-16-luc established via intraperitoneal injection, or formed by spontaneous progression from gastric wall orthotopic models. Commonly used hosts are NSG or NOD-SCID mice.

Model characteristics
Suitable for intraperitoneal local therapy, CLDN18.2 agents, peritoneal dissemination inhibition, nanodelivery, intraperitoneal perfusion, and anti-dissemination strategy research. This model can better reflect gastric cancer peritoneal dissemination and ascites formation, but tumor burden heterogeneity is significant, requiring strict standardization of scoring and sampling procedures.

Detection endpoints
Peritoneal nodule count, PCI-like score, ascites volume, intraperitoneal BLI signal, organ involvement, survival, ascites cytology, and intraperitoneal drug exposure.

Recommended positive controls
Fluorouracil or capecitabine combined with oxaliplatin as basic chemotherapy reference. CLDN18.2 programs can set zolbetuximab; ADC or CAR-T programs require intraperitoneal or systemic dosing regimens based on drug type.

Liver metastasis models can use MKN45 or MKN28 established via intrasplenic or portal vein injection for experimental liver metastasis; if the program emphasizes natural disease course, spontaneous liver metastasis from gastric wall orthotopic models can be used.

Model characteristics
Suitable for liver colonization, liver-targeted delivery, anti-adhesion, anti-microenvironment remodeling, and advanced organ metastasis research. Intrasplenic and portal vein models have relatively high success rates and shorter cycles but are more biased toward the liver colonization stage. Orthotopic spontaneous liver metastasis models are closer to natural disease course but have longer cycles and greater variability.

Detection endpoints
Liver metastatic nodule count, liver weight, BLI/IVIS or other imaging, liver tissue HE, liver tissue drug concentration, and survival.

Recommended positive controls
Fluorouracil, oxaliplatin, capecitabine, and docetaxel can serve as systemic therapy references. For targeted agents, simultaneous detection of HER2, CLDN18.2, or FGFR2b expression retention in liver metastatic lesions is recommended.

After establishing gastric wall orthotopic models with MKN45-luc or NUGC-4-luc, spontaneous peritoneal or liver metastasis processes are continuously observed. Primary lesion resection, perioperative therapy, or long-term maintenance therapy can be added based on program needs.

Model characteristics
Suitable for primary lesion to peritoneal or liver metastasis full-process, postoperative relapse, long-term maintenance, and MRD clearance research. This model has stronger clinical relevance but longer experimental cycles and greater variability, with high requirements for imaging tracking and long-term follow-up.

Detection endpoints
Orthotopic burden, peritoneal dissemination, liver metastasis, BLI, ctDNA, ascites, survival, and metastatic lesion pathology.

Recommended positive controls
Fluorouracil or capecitabine combined with oxaliplatin as basic chemotherapy reference; targeted agents should simultaneously re-verify HER2, CLDN18.2, or FGFR2b expression retention in relapsed or metastatic lesions.

Murine gastric cancer syngeneic models can establish subcutaneous platforms in matched immunocompetent mice, suitable for immune checkpoint, innate immune agonists, cytokines, combination therapy, and preliminary PK/PD research.

Model characteristics
Suitable for immunotherapy early screening, TME mechanisms, combination therapy, and immune drug dose exploration.

Detection endpoints
Tumor volume, TGI, survival, body weight, flow cytometry immune profiling, CD8/Treg, myeloid cell profile, PD-L1, cytokines, and IHC/IF.

Recommended positive controls
anti-PD-1 antibody, anti-PD-L1 antibody, anti-CTLA-4 antibody, STING agonists, or cytokine strategies can serve as immunotherapy references. Antibody doses can start at 100–200 μg/mouse IP, twice weekly or once every 3 days.

Syngeneic gastric wall orthotopic or peritoneal dissemination models can mimic gastric wall microenvironment, peritoneal dissemination, and local immune responses in an immunocompetent background.

Model characteristics
Suitable for immunotherapy, local therapy, tumor microenvironment remodeling, perioperative therapy, and relapse research. Syngeneic immunocompetent models can evaluate efficacy and mechanisms in an intact immune system background, particularly suitable for immunotherapy and TME research. Their limitation is that murine tumors differ from human gastric cancer in targets and molecular background; human target drugs, TCE, CAR-T/NK, and specific human antibodies still require human CDX or humanized model supplementary validation.

Detection endpoints
Orthotopic tumor burden, peritoneal dissemination, CD8/Treg, MDSC, macrophages, PD-L1, cytokines, TCR repertoire, spatial immune analysis, and survival.

Recommended positive controls
anti-PD-1 antibody, anti-PD-L1 antibody, anti-CTLA-4 antibody, STING agonists, or cytokine strategies can serve as immunotherapy references. Perioperative immunotherapy programs can use syngeneic gastric wall orthotopic models with primary lesion resection or local treatment, followed by observation of peritoneal and liver relapse, with FLOT-like chemotherapy, anti-PD-1/PD-L1 pathway, anti-CTLA-4 pathway, or durvalumab-related mapping strategies set.

The huPBMC model establishes a short-term human immune effector environment by infusing human peripheral blood mononuclear cells into immunodeficient mice, combined with human gastric cancer cells. Recommended models include NCI-N87, MKN45, NUGC-4, and SNU-16 target-positive models.

Model characteristics
Suitable for CLDN18.2 CAR-T, CLDN18.2 bispecifics, TCE, CAR-T/NK, human cytokines, and immune-dependent antibody programs. This model has high cost, long cycles, significant donor variability, and carries xGvHD or cytokine release risk.

Detection endpoints
hCD45, hCD3, hCD4, hCD8, hNK, T cell activation, tumor infiltration, IFN-γ, TNF-α, IL-6, Granzyme B, tumor burden, body weight, CRS, and xGvHD risk markers.

Recommended positive controls
anti-human PD-1 antibody, anti-human PD-L1 antibody, or anti-human CTLA-4 antibody can serve as mechanistic references, with doses determined by model window and human immune reconstitution status. CLDN18.2 CAR-T programs can reference satri-cel-class strategies, with target-positive, low/negative, and empty vector controls. TCE and bispecific programs should include target-positive, low/negative, and irrelevant TCE controls.

The huHSC model reconstructs a relatively long-term human immune system via human hematopoietic stem cells, suitable for sustained immunotherapy, resistance evolution, and complex combination strategy evaluation.

Model characteristics
Suitable for long-cycle TCE, CAR-T/NK, bispecifics, immune resistance, sustained immune activation, and complex TME mechanism validation. This model has high cost, long cycles, significant donor variability, and carries xGvHD or cytokine release risk; it is not recommended as a first-round screening platform.

Detection endpoints
Human immune reconstitution ratio, T/B/NK/myeloid cell spectrum, tumor infiltration, cytokines, human immune checkpoint expression, tumor burden, survival, and major organ pathology.

Recommended positive controls
anti-human PD-1 antibody, anti-human PD-L1 antibody, or anti-human CTLA-4 antibody can serve as mechanistic references, with doses determined by model window and human immune reconstitution status. CLDN18.2 CAR-T programs can reference satri-cel-class strategies. TCE and bispecific programs should include target-positive, low/negative, and irrelevant TCE controls.

Relapse and MRD models typically use MKN45-luc, NUGC-4-luc, or SNU-16-luc gastric wall orthotopic models as the base, first establishing trackable orthotopic lesions, followed by primary lesion resection, debulking, or local treatment. After primary lesion resection, peritoneal relapse, liver relapse, local relapse, and survival are continuously tracked. This model is closer to gastric cancer / EGJ adenocarcinoma postoperative peritoneal and liver relapse scenarios. Preoperative therapy, postoperative therapy, perioperative chemotherapy, perioperative immunotherapy combination, targeted maintenance, or local therapy can be set up.

Model characteristics
Relapse and MRD models have strong clinical mapping, but are technically complex, have longer cycles, and require high demands for orthotopic tumor formation, local resection consistency, and long-term follow-up. They are not suitable for first-round large-scale screening, but are suitable when key candidate drugs enter translational validation.

Modeling and dosing timelines
First establish gastric wall orthotopic lesions and confirm burden by BLI/IVIS or imaging. Premature treatment may result in an insufficient relapse window, while delayed treatment may result in an excessively high peritoneal or liver metastasis baseline. Primary lesion resection, debulking, or local treatment can be performed according to the program; postoperative recovery status, body weight, feeding, local healing, and general condition should be recorded. After local treatment, observation typically continues for 4–10 weeks, recording BLI relapse, peritoneal relapse, liver relapse, survival, and ctDNA changes.

Detection endpoints

• MRD and relapse layer: Time to local relapse, BLI relapse, peritoneal relapse, liver relapse, ascites, ctDNA, and survival
• Mechanism layer: HER2, CLDN18.2, FGFR2b, Ki67, cleaved caspase-3, CD31, PD-L1, EMT, and target retention
• Immune layer: In syngeneic or humanized models, CD8/Treg, MDSC, macrophages, hCD45, cytokines, and TCR repertoire can be detected
• Safety layer: Body weight, feeding, postoperative recovery, liver and kidney function, gastric wall local injury, and major organ pathology

Recommended positive controls
FLOT-like regimens can serve as a perioperative chemotherapy clinical mapping framework; in animal studies, they can be split into fluorouracil, leucovorin, oxaliplatin, and docetaxel combinations with dosing intervals set by tolerance. Durvalumab plus FLOT can serve as a perioperative immunotherapy combination clinical mapping reference; if immune effects need to be evaluated in animal studies, syngeneic or humanized models should be selected. HER2-positive programs can set trastuzumab or trastuzumab deruxtecan as references. CLDN18.2-positive relapse or peritoneal relapse programs can set zolbetuximab or CLDN18.2 ADC-class agents as references.