Comprehensive modeling of the Correa cascade in precancerous lesions of gastric cancer: Leveraging animal, cellular, and organoid systems for translational insights

Abstract

BACKGROUND

Precancerous lesions of gastric cancer (PLGC), including chronic atrophic gastritis, intestinal metaplasia, and dysplasia, constitute key transitional stages in the Correa cascade toward gastric cancer. Understanding these stages is essential for early detection, prevention, and therapeutic intervention.

AIM

To provide a comprehensive and critical overview of current animal, cellular, and organoid models used in PLGC research, emphasizing their applications, translational relevance, and emerging technologies.

METHODS

A scoping systematic review was performed according to the PRISMA-ScR framework. Literature was identified from PubMed and Web of Science up to June 2025 using predefined keywords and Boolean operators. Eligible studies included experimental reports describing animal, cellular, or organoid models of PLGC. Two authors independently screened titles, abstracts, and full texts, and relevant data were charted and synthesized narratively.

RESULTS

This study delineates the modeling strategies for PLGC across various platforms. Animal models, including chemical induction with N-methyl-N’-nitro-N-nitrosoguanidine, N-nitroso-N-methylurea, Helicobacter pylori infection, transgenic approaches, and composite modeling, recapitulate the histopathological spectrum of PLGC and facilitate mechanistic discovery. Cellular models simulate chronic inflammation and carcinogen exposure, enabling high-throughput mechanistic screening. Gastric organoids derived from patients or animals offer a physiologically relevant three-dimensional culture system, reproducing epithelial transformation and Helicobacter pylori-induced pathology, with recent advances in co-culture systems and immune-organoid interactions. Integration of gene editing, single-cell transcriptomics, and organoid-on-chip platforms is expected to further refine PLGC modeling and accelerate clinical translation.

CONCLUSION

Integrated PLGC models provide a robust framework for elucidating the cellular and molecular basis of early gastric tumorigenesis. Their continued evolution will enhance biomarker discovery, immunologic investigation, and therapeutic development.

Key Words: Precancerous lesions of gastric cancer; Correa cascade; Animal model; Cellular model; Organoid; Helicobacter pylori; Tumor microenvironment; Translational research

Core Tip: Precancerous lesions of gastric cancer (GC), including atrophic gastritis, intestinal metaplasia, and dysplasia, represent the pivotal transition in the Correa cascade. Accurate modeling of these lesions is crucial for dissecting early gastric tumorigenesis and identifying chemopreventive strategies. This scoping study systematically summarizes current precancerous lesions of GC models, including chemical- and infection-induced animal systems, transformed gastric epithelial cell lines, and organoid platforms derived from stem cells. We highlight their advantages, limitations, and recent advances such as genome editing, single-cell technologies, and immune co-culture. These insights provide a roadmap for selecting appropriate experimental systems and accelerating translational research in early GC prevention and therapy.

INTRODUCTION

Gastric cancer (GC) remains one of the most common digestive malignancies worldwide, ranking fifth in incidence and mortality[1]. Intestinal-type gastric adenocarcinoma, strongly associated with Helicobacter pylori (H. pylori) infection, follows the well-recognized Correa cascade described in 1992: Superficial gastritis, chronic atrophic gastritis (CAG), intestinal metaplasia (IM), dysplasia (Dys), carcinoma[2]. Clinically, CAG, IM, and Dys are grouped as precancerous lesions of GC (PLGC); their severity correlates with subsequent GC risk[3]. Because PLGC represents a critical window between chronic gastritis and malignancy, the “golden turning point”, intervention at this stage may reduce GC incidence[4].

The etiology and pathogenesis of GC have not been fully elucidated. The establishment of a simple and stable gastric epithelial “inflammation-cancer” transformation animal model is the premise and basis for studying the mechanism of GC and conducting intervention research on PLGC. The most commonly used mouse models of PLGC involve chemical carcinogens such as N-methyl-N’-nitro-N-nitrosoguanidine (MNNG) and N-nitroso-N-methylurea (MNU), as well as H. pylori infection to induce lesions; however, these approaches have shortcomings, such as poor stability and long modeling time. In recent years, the rapid development of gene editing technology has provided a new perspective for the development of PLGC mouse models. PLGC cell model is used to simulate the process of GC by culturing PLGC cells in vitro. It is also commonly used in acute and chronic models induced by chemical induction and H. pylori infection of gastric epithelial cell-1 (GES-1) cells to study the mechanisms of cell proliferation, gene mutation and immune response. Although this model is suitable for drug screening and mechanistic studies, it cannot fully reproduce the complex microenvironment of PLGC.

The organoid model is based on the three-dimensional (3D) culture of gastric epithelial structure, which can simulate the function and microenvironment of gastric tissue, closer to human tissue, and has high clinical relevance. However, it is still difficult to fully reproduce the immune and vascular microenvironment of GC. There are great differences among PLGC models, but there are no relevant reviews. In this paper, the common animal, cell and organoid models of PLGC were classified and summarized, the modeling methods, modeling time, pathological changes, observation methods of models were compared and analyzed, and the relevant points of model evaluation were commented, in order to provide reference for PLGC research.

MATERIALS AND METHODS

This study was conducted following the PRISMA-ScR checklist. Although the protocol was not prospectively registered, the methodological approach was predefined. We systematically searched PubMed, EMBASE, and Web of Science from January 2000 to June 2025 using a combination of Medical Subject Headings and free-text keywords related to PLGC, the Correa cascade, atrophic gastritis, IM, Dys, animal models, organoids, and cell culture. Boolean operators were applied, for example: (“precancerous lesions” OR “atrophic gastritis” OR “intestinal metaplasia” OR “dysplasia”) AND (“gastric” OR “stomach”) AND (“animal model” OR “rat” OR “mouse” OR “Mongolian gerbils” OR “organoid” OR “cell line”). Only peer-reviewed original studies and reviews were included, and the reference lists were screened for additional relevant articles.

Studies were included if they reported experimental models (animal, cellular, or organoid) of PLGC, described induction methods, histological outcomes, or mechanistic insights related to the Correa cascade. Case reports, clinical-only studies without modeling experiments, non-gastric models, studies focused exclusively on advanced gastric carcinoma, abstract-only publications, conference proceedings, and non-peer-reviewed materials were excluded.

Two reviewers independently screened titles and abstracts, followed by full-text assessment for eligibility, with disagreements resolved by consensus. Data extracted from each included study comprised the type of model, induction methods (chemical, bacterial, dietary, or genetic), duration of observation, histological endpoints (atrophy, metaplasia, Dys, carcinoma), and reported strengths and limitations. The results were narratively synthesized and structured according to model category, with summary tables created to facilitate comparison. The literature selection process is summarized in a PRISMA-ScR flow diagram, showing the number of records identified, screened, excluded, and finally included in the systematic review. The literature screening and selection process and representative experimental models for each stage of the Correa cascade are summarized in Figures 1 and 2.

Figure 1 Literature screening and selection process for the scoping review. A total of 813 records were identified (715 from databases and 98 from other sources). After removing duplicates, 527 records remained, of which 279 were screened. Following title and abstract screening, 192 full-text articles were assessed for eligibility. Of these, 37 were excluded with reasons, leaving 155 studies that were finally included in the review.

Figure 2 Representative experimental models for each stage of the Correa cascade, from chronic and atrophic gastritis to intestinal metaplasia, dysplasia, and early gastric cancer. It includes: (1) Animal models: N-methyl-N’-nitro-N-nitrosoguanidine-induced lesions in rats, Helicobacter pylori-infected mice, and spasmolytic polypeptide-expressing metaplasia models induced by tamoxifen or genetic modification; (2) Cellular models: Human gastric epithelial cell lines (e.g., gastric epithelial cell-1) exposed to chemical, inflammatory, or infectious stimuli to simulate precancerous changes; and (3) Organoid models: Patient-derived organoids from intestinal metaplasia or dysplastic tissue, pluripotent stem cell-derived gastric organoids via directed differentiation, and organoids from genetically engineered mice. Together, these models offer complementary platforms for investigating the stage-specific pathogenesis, immune responses, and therapeutic targets of precancerous gastric lesions. H. pylori: Helicobacter pylori; S. anginosus: Streptococcus anginosus; MNNG: N-methyl-N’-nitro-N-nitrosoguanidine; MNU: N-Nitroso-N-methylurea; PSC: Pluripotent stem cell.

RESULTS

PLGC animal model

Chemical carcinogens or dietary induction models: MNNG induction model, long-term low-dose exposure to N-nitrosamines is a key factor in GC[5]. Among chemically induced PLGC models, MNNG is widely used due to its strong mutagenic and carcinogenic properties. It directly alkylates DNA bases, promoting tumorigenesis[6]. Given that over 95% of human GCs are adenocarcinomas, MNNG’s methylation specificity toward gastric glandular tissue makes it ideal for mimicking the progression from CAG to GC[7].

Wistar and Sprague-Dawley rats are commonly used due to their docility, resistance to infections, low spontaneous tumor incidence, and suitability for oral administration[8]. IM and Dys can be induced by administering 200 μg/mL MNNG in drinking water for 16 weeks, with progression to gastric adenocarcinoma by 40 weeks[9,10], Alternatively, gavage with 800 mg/L MNNG for 10 weeks can induce PLGC, with progression to carcinoma observed by 34 weeks[11]. However, high-dose gavage increases mortality and reduces model success rate. In mice, free drinking of 150 μg/mL MNNG for 6 weeks has been sufficient to induce PLGC[12,13]. Because PLGC is multifactorial, compound models combining MNNG with agents such as ranitidine, ammonia, sodium salicylate, ethanol, or irregular feeding better simulate real-world pathogenesis[14], Wistar rats are preferred for these models due to more stable weight gain and lower tumor background compared to Sprague-Dawley rats[15].

MNNG concentration and administration route are critical. Free drinking reflects natural exposure, while gavage enhances mutagenicity and reduces off-target toxicity[16]. Cai et al[17] demonstrated that 200 μg/mL MNNG for 16 weeks induced IM and Dys, and Wang et al[18] observed mucosal atrophy with focal atypia after 40 weeks of free drinking 100 μg/mL MNNG. Higher doses via gavage (250-50 mg/kg) tend to induce forestomach lesions, whereas free intake targets gastric glandular tissue. Thus, model success depends on dose, duration, and delivery. Free drinking of 200 μg/mL MNNG for 16 weeks induces PLGC efficiently but only partially mimics human pathogenesis due to lack of sustained immune infiltration and stromal remodeling.

MNU induction model: The utility of another N-nitroso compound, MNU as a gastric carcinogen was tested in mice and found to be more effective. Researchers have verified that the MNU-induced PLGC model in mice can be successfully established by short-term gavage of MNU solution (120 ppm). The experimental approach consists of enabling the mice to freely drink the MNU solution for 8 weeks, after which the mice exhibit IM and Dys in the gastric mucosa[19]. When the mice are subjected to the treatment for 30 weeks, GC can be induced[20]. However, the MNU model of gastric carcinogenesis does not proceed through a classical atrophy-metaplasia-Dys sequence, and the time, location, quantity and size of cancer are uncertain, with large genetic heterogeneity, which is the drawback of the MNU mouse model. Therefore, MNU is commonly used in combination with H. pylori infection and a high-salt diet to establish animal models of PLGC and GC[21].

Deoxycholic acid (DCA) and chenodeoxycholic acid (CDCA) induction model: During bile reflux, phospholipase A in pancreatic fluid converts lecithin into lysolecithin, which disrupts the phospholipid bilayer of GESs, increasing permeability and enhancing mucosal injury by gastric acid[22]. Bile acids are classified as free (e.g., cholic acid, DCA, CDCA, and lithocholic acid) or conjugated. Among these, DCA and CDCA are commonly used to model gastric IM both in vitro and in vivo[23,24]. In a seminal study, FVB/N insulin-gastrin (INS-GAS) gene transgenic mice were given 0.2% DCA in drinking water for five months. Compared to C57BL/6 mice, these mice exhibited marked glandular atrophy, accelerated IM, and low-grade Dys in the gastric mucosa after six months[23]. CDCA-induced models typically involve administering 20 mmol/L CDCA to rats daily, often combined with 2 mL of 2% sodium salicylate or 0.1% ammonia plus irregular feeding. After 12-16 weeks, consistent pathological features of CAG and IM are observed in the antrum[24-26].

Acute drug- and H. pylori-induced spasmolytic polypeptide-expressing metaplasia (SPEM) model: SPEM, which is considered a potential neoplastic precursor of IM[27], is characterized by spasmolytic polypeptide (trefoil factor 2) or mucin6 (MUC6)/GSII expression in the basal region of the gland, comprising the chief cells after parietal cell loss[28,29]. DMP-777 is a protonophore that blocks acid secretion and induces rapid parietal cell loss, leading to SPEM after 10-14 days. These changes are reversible within 7-14 days after withdrawal. While these acute chemical injury models efficiently generate SPEM, they lack sustained immune-mediated remodeling, limiting their relevance to chronic human PLGC. However, long-term DMP-777 treatment does not cause Dys, likely due to a lack of inflammation[30]. In contrast, L635, a structural analog of DMP-777 that lacks neutrophil elastase inhibition, induces parietal cell loss accompanied by significant inflammation. This inflammatory context accelerates SPEM development via transdifferentiation of chief cells[28].

High-dose tamoxifen also induces parietal cell loss by acting as a proton carrier. Daily administration of 5 mg emulsified tamoxifen for 3 days reduces parietal cell numbers by 90%, with SPEM emerging within 3 days in mice treated with 250 mg/kg. These lesions are similarly reversible within 2-3 weeks after drug withdrawal[31,32]. Tamoxifen also ablates tuft cells, which play key roles in early carcinogenesis. In gp130^Y757F/Y757F mice, two tamoxifen injections at 16 weeks induced tuft cell loss, leading to spontaneous development of SPEM-associated intestinal-type adenomas as early as 4 weeks of age[33]. Recent studies have identified a novel mechanism by which H. pylori induces SPEM through the CagA-telomerase reverse transcriptase (TERT)-Wnt signaling axis. Importantly, this mechanism was elucidated using the CagA-positive Sydney strain 1 (PMSS1). In mice infected with PMSS1, H. pylori stabilizes TERT by suppressing synovial apoptosis inhibitor 1-mediated ubiquitination and degradation, thereby activating the Wnt/β-catenin pathway and driving SPEM development. Genetic deletion of TERT abolished these pathological changes, confirming that the CagA-TERT-Wnt axis is essential for H. pylori-induced metaplastic transformation[34].

High-salt and salting diet induced models: A high-salt and pickled diet is a high-risk factor for GC. High salt diet combined with H. pylori Sydney strain (HpSS1) infection resulted in higher rates of atrophy and epithelial proliferation in Mongolian gerbils[35], and high salt diet also increased the incidence of MNU-induced gastric tumors in C57BL/6 mice[36]. High salt intake alone does not induce GC, and it is more likely to cause cancer synergistically with other factors and increase the risk of GC.

Helicobacter infection model: H. pylori infection is a major etiological factor in gastric carcinogenesis, and various animal models have been developed to study its pathogenic mechanisms. Among them, Mongolian gerbils and C57BL/6 mice are the most widely used due to their reproducible pathological responses. Notably, disease severity and temporal progression in these models are strongly influenced by both bacterial strain virulence and host genetic background, underscoring the importance of strain-host interactions in PLGC modeling.

Watanabe et al[37] first established a Mongolian gerbil model of H. pylori-induced GC, with 37% of animals developing tumors resembling human intestinal-type cancer after 62 weeks of infection. Because C57BL/6 mice are resistant to most H. pylori strains, Helicobacter felis, a closely related species, has been used instead and can cause gastric metaplasia, Dys, and invasive cancer with prolonged infection[38,39]. Lee et al[40] then isolated the HpSS1 carrying CagA and VacA, which successfully colonized C57BL/6 mice, inducing chronic gastritis and atrophy but no carcinoma even after 2 years. Later, Arnold et al[41] introduced PMSS1, a stable mouse-adapted strain that induces metaplasia/SPEM, though with a lower colonization rate than HpSS1.

Effect of H. pylori infection in the mouse stomach: The pathogenic outcome in mice depends on the strain of H. pylori and the duration of infection. In C57BL/6 mice, HpSS1 infection causes lesions ranging from mild inflammation to atrophy and hyperplasia over 5 days to 9 months; hyperplasia and IM were observed at 3 and 18 months, respectively[42]. The more virulent PMSS1 strain accelerates this process, with metaplasia detected at 8 weeks[43] and inflammation by 3-6 months[44]. In BALB/c mice, HpSS1 induces chronic gastritis within 40 days to 8 weeks[45]. These results highlight the influence of host and bacterial factors. Notably, PMSS1 exhibits stronger pathogenicity with accelerated induction of metaplasia and inflammation, whereas HpSS1 favors long-term colonization with relatively milder pathological outcomes[46]. Recent studies also identified SlyD as a novel virulence factor that promotes IM via stabilization of tumor protein translationally-controlled 1 and activation of caudal-related homeobox transcription factor 2 (CDX2)[47].

Effect of H. pylori infection in the Mongolian gerbil: Mongolian gerbils are highly susceptible to H. pylori-induced gastric lesions, closely mimicking human disease progression. Long-term infection leads to gastritis, atrophic gastritis, IM, Dys, and adenocarcinoma[37,48]. Different H. pylori strains exhibit varying carcinogenic potentials. For instance, TN2GF4 strain induces IM and carcinoma after 18 months[49,50], ATCC 43504 strain leads to IM within 3-12 months[51], and strain 7.13 causes rapid progression from IM (7 weeks) to adenocarcinoma (2 months)[52]. These comparisons underscore the importance of considering strain-specific virulence when designing murine PLGC studies.

Streptococcus anginosus (S. anginosus) mouse model: Although H. pylori initiates gastric pathology, its influence wanes during later disease stages. Using metagenomic analysis, Fu et al[53] identified S. anginosus as a non-H. pylori species associated with gastric inflammation and tumorigenesis. In murine models, S. anginosus induced acute gastritis within two weeks, and 9-12 months infection led to chronic gastritis, mucinous metaplasia, and Dys in both conventional and germ-free C57BL/6 mice. Notably, co-infection with S. anginosus and MNU significantly accelerated gastric tumorigenesis compared to H. pylori alone.

Multi-factorial induction model: The single-factor modeling method has a single intervention and a long modeling time, which is not fully compatible with the view that atrophic gastritis and its precancerous lesions are complex diseases with multifactorial participation. Therefore, scholars have tried to superimpose other carcinogenic factors on top of MNU and MNNG for composite induction, in order to more closely resemble and mimic the human pathogenic process.

Controversy and exploration of MNU combined Helicobacter infection modeling: Several studies have explored the combined use of MNU and H. pylori infection in inducing gastric carcinogenesis in mice, but results remain inconsistent. Toyoda et al[36] administered 120 ppm MNU via alternate-week free drinking with a 10% NaCl high-salt diet, and a subset received additional H. pylori inoculation biweekly. After 40 weeks, all mice developed antral tumors, suggesting that H. pylori and/or a high-salt diet significantly enhance MNU-induced tumorigenesis, However, whether H. pylori alone synergizes with MNU remains debated.

For this controversy, Lee et al[54] demonstrated that mice co-treated with H. pylori and MNU had the highest GC incidence, significantly exceeding that of either treatment alone. Supporting this, another study reported that 240 ppm MNU combined with oral HpSS1 inoculation induced GC after 50 weeks[55]. Additionally, infection with more virulent strains such as PMSS1[56] and ATCC 43504[57] resulted in gastric mucosal atrophy, IM, and heterotopic hyperplasia. These findings confirm that MNU co-administered with H. pylori accelerates gastric disease progression and tumorigenesis.

MNU combined diet induction model: In terms of modeling of MNU solution combined with dietary factors, Liao et al[58] established a model using 300 ppm MNU in drinking water combined with an irregular feeding schedule (starvation-to-satiation cycles), leading to IM and Dys within 20 weeks. Similarly, Sethi et al[59] developed a long-term model in leucine-rich repeat-containing G protein-coupled receptor 5 (Lgr5)-p53^-/^- mice exposed to MNU and DCA for 12 months. These mice developed dysplastic and metaplastic lesions, highlighting the critical role of gene–environment interactions in promoting premalignant gastric transformation.

MNNG integrated induction model: The administration mode of MNNG solution and its combined solution is often alternate, and the concentration and dose are also adjusted according to the different administration mode. The concentration is lower and the dose is higher when drinking freely. The concentration was higher and the dose was lower during intragastric administration. In the selection of joint factors, dietary habits, chemical factors and physical factors are usually combined to form an auxiliary effect with carcinogens, and combined factors such as ammonia or ethanol or sodium salicylate solution are used on fasting days. Compound models combining MNNG with ranitidine, ammonia, sodium salicylate, ethanol, or irregular feeding partially improve physiological relevance, yet still fall short of fully recapitulating human PLGC immune and stromal features.

(1) Ammonia and H. pylori simulation. The pH of normal gastric mucosa is 2.0, and ammonia can simulate the high ammonia concentration caused by H. pylori infection, destroying the acidic environment of gastric mucosa. Wei et al[16] by compared different mouse models found that the combination of H. pylori, MNNG, and ammonia water is an effective method of developing a mouse model of human CAG. Lv et al[60] MNNG, ammonia and irregular combined with fasting, PLGC rat model was successfully constructed; (2) Diet and lifestyle and stress-emotion factors: High-salt diet can promote gastric mucosal damage and inflammation, and it plays a role in combination with other factors. Yin et al[61] found that saturated salt plus MNNG for 25 weeks induced mild atrophy, while CAG with IM appeared after 35 weeks. Some studies mimicked poor eating habits by alternating fasting and satiation (1-day cycles) alongside MNNG administration for 10 weeks, successfully inducing CAG[62,63]. Zhu et al[64] established a PLGC model combining 24 weeks of MNNG gavage with an irregular diet (2 days satiation, 1 day fasting) and weekly tail-pinching to induce emotional stress. Such stress-emotion PLGC models provide a framework to explore the interactions between physiological, environmental, and psychosomatic factors in gastric precancerous lesion development; (3) Chemical reagents: Sodium salicylate is a non-steroidal anti-inflammatory drug, which causes inflammation by directly stimulating the local gastric mucosa, resulting in the shedding of gastric mucosa cells[18]. Zhu et al[65] combined MNNG with sodium salicylate (2%) to establish a PLGC rat model. At 21 weeks, gastric mucosa atrophy, IM and atypical hyperplasia occurred. In addition, long-term administration of MNNG with 5% aqueous alcohol solution successfully induced GC in rats after 14 months[66]; and (4) Role of ranitidine in accelerating MNNG mutation: Ranitidine can inhibit gastric acid secretion, cause weak acid environment in stomach, promote nitrate to nitrite and accelerate MNNG mutagenesis. The modified MNNG-induced PLGC model was established by oral MNNG, sodium salicylate, 40% alcohol, irregular fasting, and 0.03%-0.05% ranitidine, and such model rats exhibit the phenotypes of gastric mucosa pathology including atrophy, metaplasia, Dys, and adenocarcinoma[67-70]. This composite model is widely used in the evaluation of drug therapy effect.

MNNG combined with Helicobacter infection induction model: In Mongolian gerbils, 50 μg/mL MNNG via free drinking combined with H. pylori infection induced atrophic gastritis by 32 weeks and GC by 52 weeks, although IM was not observed[71]. In contrast, C57BL/6 mice receiving 40 μg/mL MNNG and HpSS1 infection developed moderate gastric hyperplasia at 48 weeks without tumor formation[7]. These findings suggest that H. pylori infection synergizes with chemical carcinogens like MNNG to promote gastric carcinogenesis, though outcomes vary by model and experimental conditions. Commonly used strategies for constructing animal models of precancerous gastric lesions are illustrated in Figure 3, and detailed modeling methods, agents, and animal types used in precancerous lesions of gastric cancer research are summarized in Table 1[72-78].

Figure 3 Commonly used strategies for constructing animal models of precancerous lesions of gastric cancer. Including: (1) Chemical carcinogen-induced models using agents such as N-methyl-N’-nitro-N-nitrosoguanidine or N-Nitroso-N-methylurea in combination with high-salt diet, ethanol, bile acids, or sodium deoxycholate; (2) Infection-based models using Helicobacter pylori in mice or Mongolian gerbils to mimic inflammation-driven transformation; (3) Combination models integrating chemical exposure with Helicobacter pylori infection, high-salt feeding, ethanol administration, fasting, or ammonia water to accelerate disease progression; and (4) Spasmolytic polypeptide-expressing metaplasia models induced by tamoxifen, L635, or DMP-777, which recapitulate key features of gastric glandular cell transdifferentiation. DCA: Deoxycholic acid; CDCA: Chenodeoxycholic acid; MNNG: N-methyl-N’-nitro-N-nitrosoguanidine; MNU: N-Nitroso-N-methylurea; H. pylori: Helicobacter pylori; S. anginosus: Streptococcus anginosus; H. felis: Helicobacter felis; Trx1: Thioredoxin 1; SD: Sprague-Dawley.

Table 1 Summary of modeling methods, agents, and animal types used in precancerous lesions of gastric cancer research.

Type	Modeling method	Model animals	Modeling time	Observation time	Atrophy	Metaplasia	Dysplasia	Carcinoma	Ref.
MNU	MNU (120 ppm)	C57BL/6 mice	8 weeks	8 weeks	Yes	IM	Yes	No	[19]
	MNU (100-120 ppm)	BALB/c and C57BL/6 mice, Wistar rats	16-30 weeks	31-54 weeks	Yes	Yes	Yes	Yes	[20,72]
	MNU (120 ppm)	db/db mice	20 weeks	30 weeks	No	No	Yes	Yes	[73]
	MNU (300 ppm)	BALB/c mice	16 weeks	16 weeks	No	IM (16 weeks)	No	No	[74]
MNNG	MNNG (200 μg/mL)	SD rats	16 weeks	26 weeks	No	IM (16 weeks)	Yes (16 weeks)	No	[10,75]
	MNNG (200 μg/mL)	SD rats	40 weeks	40 weeks	No	No	Yes (40 weeks)	Yes (40 weeks)	[9]
	MNNG (800 mg/L)	SD rats	10 weeks	34 weeks	Yes (10 weeks)	No	Yes (22 weeks)	Yes (34 weeks)	[11]
	MNNG (150 μg/mL)	BALB/c mice, C57BL/6 mice	6 weeks	14-36 weeks	Yes	IM	Yes	No	[12,13]
CDCA	DCA (0.2%)	INS-GAS mice	6 months	6 months	Yes (5 months)	IM	Yes	No	[23]
	CDCA (20 mmol/L) + 2% sodium salicylate	SD rats	12 weeks	12 weeks	Yes	IM	No	No	[24,25]
	CDCA (20 mmol/L) + 0.1% ammonia + irregular diet	SD rats	16 weeks	16 weeks	Yes				[26]
Acute drug and H. pylori induction	DMP-777	CD-1 rats, C57BL/6 mice	7-14 days	7-14 days	SPEM	No	No	No	[30,76]
	L635	C57BL/6 mice	3 days	3 days	SPEM	No	No	No	[77]
	Tamoxifen (250 mg/kg)	C57BL/6 mice	3 days	3 days	SPEM	Yes	No	No	[31,32]
	Tamoxifen (50 mg/kg)	gp130Y757F/Y757F mice	2 days	16 weeks	SPEM	Yes	Yes	Yes	[33]
	CagA-positive PMSS1	C57BL/6 mice, Tert-/- mice	12 weeks	12 weeks	SPEM	No	No	No	[34]
H. pylori strain	H. felis	C57BL/6 mice	15 months	15 months	Yes	Yes	Yes	Yes	[39]
	H. pylori SS1	C57BL/6 mice	18 months	18 months	Yes	Yes	No	No	[42]
	H. pylori PMSS1	C57BL/6 mice	8 weeks	8 weeks	Yes	Yes	No	No	[43]
	H. pylori HpslyD + strain	C57BL/6 mice	73 weeks	73 weeks	Yes	IM	No	No	[47]
	H. pylori TN2GF4 strain	Mongolian gerbils	18 weeks	18 weeks	Yes (2 weeks)	IM	No	No	[50]
	H. pylori ATCC 43504 strain	Mongolian gerbils	16-18 months	16-18 months	Yes	IM	Yes	Yes	[49]
	H. pylori 7.13 strain	Mongolian gerbils	7-9 weeks	7-9 weeks	Yes	IM (7 weeks)	Yes	Yes (9 weeks)	[52]
Non-H. pylori gastric microbiome	S. anginosus	C57BL/6 mice, germ-free BALB/c mice	48 weeks	48 weeks	Yes (9 months)	Yes (12 months)	Yes (12 months)	No	[53]
	S. anginosus + MNU (240 ppm)	C57BL/6 mice	36 weeks	36 weeks	Yes	Yes	Yes	Yes (9 months)	[53]
MNU + H. pylori	MNU (240 ppm) + HpSS1	C57BL/6 mice	12-16 weeks	50-80 weeks	Yes	IM	Yes	Yes	[55]
	MNU (200 ppm) + H. felis	FVB and B6 mice	36 weeks	36-50 weeks	Yes	Yes	Yes	Yes	[54,78]
	MNU + PMSS1	C57BL/6 mice	12 weeks	40 weeks	Yes	IM	Yes	No	[56]
	MNU (10 ppm) + ATCC 43504	Mongolian gerbil	24 weeks	53 weeks	Yes	IM	Yes	No	[57]
MNU + dietary	MNU (120 ppm) + 10% NaCl	C57BL/6J mice	40 weeks	40 weeks				Yes	[36]
	MNU (120 ppm) + HpSS1 + 10% NaCl	C57BL/6J mice	40 weeks	40 weeks				Yes	[36]
	MNU + DCA	C57BL/6 mice, Lgr5-p53-/- mice	48 weeks	48 weeks		Yes	Yes	No	[59]
	MNU (300 ppm) + irregular diet	BALB/c mice	20 weeks	20 weeks	Yes	IM	Yes	No	[58]
MNNG + H. pylori	MNNG (50 μg/mL) + KMUH-G926	Mongolian gerbil	20 weeks	52 weeks	Yes (32 weeks)	No	Yes	Yes (52 weeks)	[71]
	MNNG (40 μg/mL) + HpSS1	C57BL/6J mice	48 weeks	48 weeks	No	No	Yes (48 weeks)	No	[7]
MNNG composite model	MNNG (200 μg/mL) + saturated salt	Wistar rats	25 weeks	35 weeks	Yes (25 weeks)	IM	Yes (35 weeks)	No	[61]
	MNNG (200 μg/mL) + 0.1% ammonia + irregular fasting	SD rats	20 weeks	20 weeks	Yes	IM	Yes	No	[60]
	MNNG (120 μg/mL) + H. pylori + ammonia	Kunming mice	7 days	120 days	Yes	No	No	No	[16]
	MNNG (170-200 μg/mL) + irregular diet	SD rats, Wistar rats	10-12 weeks	14 weeks	Yes	No	No	No	[62,63]
	MNNG (200 μg/mL) + irregular diet + emotionally stimulated	Wistar rats	24 weeks	24 weeks	Yes	IM	Yes	No	[64]
	MNNG (100 μg/mL) + 2% sodium salicylate	Wistar rats	17 weeks	21 weeks	Yes	IM	Yes	No	[65]
	MNNG (120 μg/mL) + 5% aqueous alcohol solution	Wistar rats	56 weeks	56 weeks	No	No	No	Yes (14 months)	[66]
	MNNG (120-150 μg/mL) + 2% sodium salicylate + irregular fasting + 0.05% ranitidine	Wistar rats	32-40 weeks	32-40 weeks	Yes	Yes	Yes	Yes	[67,68,70]
	MNNG (200 μg/mL) + 40% alcohol + irregular fasting + 0.03% ranitidine	SD rats	40 weeks	40 weeks	Yes	IM	Yes	No	[69]

MNU: N-Nitroso-N-methylurea; MNNG: N-methyl-N’-nitro-N-nitrosoguanidine; CDCA: Chenodeoxycholic acid; H. pylori: Helicobacter pylori; DCA: Deoxycholic acid; H. felis: Helicobacter felis; S. anginosus: Streptococcus anginosus; HpSS1: Helicobacter pylori Sydney strain; INS-GAS: Insulin-gastrin; SD: Sprague-Dawley; Lgr5: Leucine-rich repeat-containing G protein-coupled receptor 5; IM: Intestinal metaplasia.

Gene editing mouse mode: Gene-edited mouse models are now widely used to study gastric carcinogenesis both domestically and internationally. By inducing aberrant expression of specific inflammatory mediators or mutations in oncogenes and tumor suppressor genes, these models more accurately recapitulate disease pathogenesis and enable functional analysis of key genes and signaling pathways involved in PLGC progression. Compared with Helicobacter infection or chemically induced models, gene-edited models offer advantages such as controlled induction timing and synchronized lesion development.

GAS correlation model: GAS knockout (GAS^-/^-) mice. GAS, secreted by antral G cells, regulates acid secretion via parietal and enterochromaffin-like cells. GAS^-/^- mice exhibit antral inflammation, parietal/enterochromaffin-like cell loss, and develop bacterial overgrowth, IM, and spontaneous tumors by 4-5 months under conventional housing[79]. By 12 months, they progress to gastritis, atrophy, IM, and invasive cancer. DMP-777 induces rapid SPEM (1-3 days) in GAS^-/^- mice, faster than in wild-type (> 10 days)[76]. These mice are also more susceptible to MNU-induced tumorigenesis, whereas GAS-overexpressing mice show resistance via epigenetic regulation of trefoil factor family 1 (TFF1)[78].

INS-GAS transgenic mice. INS-GAS transgenic mice, which express human GAS under the insulin promoter, develop sustained hypergastrinemia that initially (1-4 months) increases parietal cell mass and acid secretion, followed by progressive atrophy, metaplasia, Dys, and carcinoma by 20 months[80]. These mice are commonly combined with H. pylori infection to model gastric carcinogenesis, with pathological outcomes depending on bacterial strain and infection duration. For example, HpSS1 infection leads to Dys after 4-9 months[80], while PMSS1 causes invasive carcinoma as early as 4 months[81,82]. B129 strain infection induces Dys and tumors within 3-6 months[83]. INS-GAS and GAS^-/^- mice are often used together to investigate GAS’s role in tumorigenesis. Notably, tumors in INS-GAS mice localize mainly to the corpus, whereas GAS^-/^- lesions occur predominantly in the antrum, necessitating precise histological distinction between gastric regions in these models[84]. The accelerated neoplastic progression observed in H. pylori-infected INS-GAS mice, particularly with virulent strains such as PMSS1 or B129, parallels the enhanced cancer risk seen in humans. A cluster-randomized controlled trial has demonstrated that timely eradication of H. pylori reduces the incidence of GC[85] and decreases metachronous tumor development after endoscopic resection[86]. These findings highlight INS-GAS mice as a genetically sensitized host background in which H. pylori strain-dependent effects on PLGC progression are markedly amplified, underscoring the importance of host-pathogen interactions in murine GC modeling.

Models related to the induction of inflammatory transmitters. Interleukin-1β-Tg mice (H⁺/K⁺-ATPase promoter-driven) develop spontaneous oxyntic atrophy, SPEM, IM, and high-grade Dys or carcinoma[87]. Interferon-γ (IFN-γ) overexpression in parietal cells induces corpus/antral inflammation, parietal/chief cell loss, SPEM, and atypical hyperplasia or adenomatous change[88]. The TxA23 TCR-Tg model, featuring CD4+ T cells against H⁺/K⁺-ATPase, mimics autoimmune gastritis with progressive oxyntic atrophy and metaplasia by 2-4 months; nearly all mice develop atypical hyperplasia by 12 months[89]. These models underscore the critical role of chronic cytokine signaling and immune injury in promoting gastric pre-neoplasia.

Oncogene mutation oncogene mutation modeling: Activation of oncogenic Kirsten rat sarcoma viral oncogene homolog (K-ras) in gastric chief cells induces rapid metaplastic transformation, with SPEM appearing within one month and IM by four months[90]. Several transgenic mouse lines expressing mutant K-ras under constitutive or inducible promoters, such as Mist1CreERT[91] and eR1CreERT[92], have been established. In K19-K-ras-V12 mice, parietal cell loss, SPEM, and Dys resemble Helicobacter felis-induced preneoplasia[93], underscoring Ras pathway activation as central to PLGC progression.

Loss of Runt-related transcription factor 3 leads to sequential SPEM and IM, with MNU accelerating adenocarcinoma[94]. Usf1^-/^- mice infected with H. pylori develop atrophic gastritis, IM, and Dys[95], while p27^-/^- mice infected with HpSS1 show inflammation, pseudopyloric metaplasia, atrophy, and Dys by 70 weeks[96]. More recently, gastric-specific deletion of SET domain containing 2, a histone methyltransferase linked to poor prognosis, combined with oncogenic Myc mutation (MYCR26StopFL/+; Setd2^-/^-), produced high-grade IM and Dys as early as 10 weeks, modeling early-onset GC[97].

Glandular cell injury: Suzuki et al[98] generated claudin-18 (stomach-specific claudin-18)-deficient mice, which exhibited increased gastric acid secretion and spontaneous SPEM as early as postnatal day 3. Notably, 20%-30% developed gastric tumors between 60 and 100 weeks, indicating claudin-18’s essential role in parietal cell differentiation and glandular homeostasis. Liu et al[99] demonstrated that both global and parietal cell-specific deletion of Slc26a9 led to spontaneous premalignant and malignant gastric lesions, underscoring its importance in epithelial integrity and acid-base balance. Katsha et al[100] showed that TFF1-knockout mice developed antral inflammation by 8 weeks, with over half progressing to high-grade Dys and some to invasive adenocarcinoma by 12 months, implicating TFF1 as a tumor suppressor in the antrum. More recently, Zeng et al[101] identified mitochondrial gene associated with retinoid-IFN-induced mortality-19 as a pathogenic driver of SPEM. Parietal cell-specific Grim-19 knockout disrupted glandular differentiation and promoted spontaneous gastritis and SPEM via activation of the NLR family pyrin domain-containing 3-interleukin-33 axis, mediated by the reactive oxygen species-nuclear factor erythroid-2-related factor 2-heme oxygenase-1-nuclear factor-κB signaling cascade.

Pathway misalignment: Several transgenic and knockout mouse models have delineated key pathways in gastric tumorigenesis. Peroxisome proliferator-activated receptor delta-overexpressing mice (villin promoter) sequentially developed SPEM, IM, Dys, and adenocarcinoma by 55 weeks, implicating peroxisome proliferator-activated receptor delta in epithelial transformation[102]. Smad3^-/^- mice exhibited early SPEM by 6 months and invasive neoplasia by 10 months, particularly at the forestomach/glandular junction[103]. MyD88-deficient mice infected with Helicobacter rapidly developed atrophy, metaplasia, and Dys (25-47 weeks), underscoring the role of innate immunity[104]. Wnt activation via Rnf43 mutation further aggravated Helicobacter-induced lesions[105]. Collectively, these models demonstrate how dysregulated metabolic, inflammatory, and developmental signaling drives PLGC progression and highlight the utility of genetic models in GC research.

Models of polygenic mutations or deletions in specific glandular cells. Seidlitz et al[106] developed a gastric-specific Anxa10-CreERT2 mouse model harboring Kras^G12D/+, p53^R172H/+, and Smad4^fl/fl mutations, which showed Dys by 3 weeks, T1-T2 tumors by 2-8 weeks, T3-T4 tumors with liver/Lung metastases by 10 weeks. Li et al[107] inactivated Smad4 and Pten in Lgr5⁺ stem cells, leading to rapid antral adenoma and intramucosal carcinoma formation. Hayakawa et al[91] introduced Kras activation and/or Apc loss into Mist1⁺ isthmic stem cells; Kras alone induced Dys with IM, while combined mutations led to intestinal-type GC within 4 months. Fang et al[108] used H⁺/K⁺ ATPase-Cre to inactivate Lkb1 and Pten, resulting in hyperplastic polyps (approximately 20 weeks) progressing to metastatic adenocarcinoma (approximately 40 weeks). Genetically engineered mouse models of precancerous gastric lesions and their histological characteristics are summarized in Table 2[109-112].

Table 2 Genetically engineered mouse models of precancerous gastric lesions and their histological characteristics.

Type	Mouse model	Observation time	Intervention	SPEM	Atrophy	Metaplasia	Dysplasia	Carcinoma	Ref.
Gastrin-related	Gastrin-/- mice	12 months	No		Yes	IM	Yes	Yes	[79]
	Gastrin-/- mice	28 days	DMP-777	Yes					[76]
	Gastrin-/- mice	36 weeks	MNU + H. felis		Yes	Yes	Yes	Yes	[78]
	INS-GAS mice	36 weeks	MNU		Yes	Yes	Yes	Yes	[78]
	INS-GAS mice	6-18 months	H. pylori SS1		Yes	Yes	Yes	No	[80]
	INS-GAS mice	3-4 months	H. pylori PMSS1, B128, NCTC11637		Yes	IM	Yes	Yes	[81-83]
Inflammatory transmitter induction	H/K-IL-β transgenic mice	12 months	No	Yes	Yes	Yes	Yes	Yes	[87,109]
	H/K-IFN-γ transgenic mice	15 months	No	Yes	Yes	Yes	Yes	Yes	[88]
	TxA23 TCR transgenic mice	12 months	No	Yes	Yes	Yes	Yes	Yes	[89]
	K19-Wnt1/C2mE transgenic mice	30 weeks	No	Yes	No	Yes	Yes	Yes	[110]
Mutations in oncogenes and tumor suppressors	Mist1-Kras transgenic mice	1-4 months	No	Yes	Yes	Yes	Yes	Yes	[90]
	K19-K-ras-V12 transgenic mice	20 months	No		Yes	Yes	Yes	Yes	[93]
	Runx3-/- mice	52 weeks	MNU	Yes	No	IM	Yes	Yes	[94]
	Usf1-/- mice	12 months	H. pylori		Yes	IM	Yes	No	[95]
	p27-/- mice	70 weeks	H. pylori SS1		Yes	Yes	Yes	No	[96]
	MYCR26StopFL/+; Setd2-/-(AMC) mice	10 weeks	No		No	IM	Yes	No	[97]
Glandular cell damage	stCldn18-/- mice	100 weeks	No	Yes	Yes	IM	Yes	Yes	[98]
	Slc26a9-/- mice	18 months	No	Yes	Yes	IM	Yes	Yes	[99]
	Tff1-/- mice	12 months	No		Yes	Yes	Yes	Yes	[100]
	Hip1r-/- mice	5 weeks	No	Yes	No	No	No	No	[110]
	GRIM-19-/- mice	10 weeks	No	Yes	No	No	No	No	[101]
Dysregulation of signalling pathways	PPARD mice	55 weeks	No	Yes	Yes	IM	Yes	Yes	[102]
	Smad3-/- mice	10 months	No	Yes	Yes	Yes	Yes	Yes	[103]
	H/K-noggin mouse	12 weeks	No	Yes	No	No	No	No	[111]
	MyD88-/- mice	47 weeks	H. felis		Yes	IM	Yes	Yes	[104]
	RNF43H292R/H295R mice	6 months	H. pylori; PMSS1		Yes	IM	Yes	Yes	[105]
Multiple gene mutations or deletions in specific glandular cells	Pgc-Cre mice	9 months	No	Yes	Yes	Yes	Yes	Yes	[112]
	Anxa10-Cre mice	10 weeks	No		No	No	Yes	Yes	[106]
	Lgr5-Cre; Smad4fl/fl; PTENfl/fl mice	12 months	No		Yes	IM	Yes	Yes	[107]
	Mist1-Cre; LSL-KrasG12D; Apcfl/fl mice	4 months	No		Yes	IM	Yes	Yes	[91]
	H/K Cre; LKB1L/L; PTENL/L mice	40 weeks	No		No	IM	Yes	Yes	[108]

INS-GAS: Insulin-gastrin; IL-β: Interleukin-β; IFN-γ: Interferon-γ; PPARD: Peroxisome proliferator-activated receptor delta; Lgr5: Leucine-rich repeat-containing G protein-coupled receptor 5; H. pylori: Helicobacter pylori; H. felis: Helicobacter felis; MNU: N-Nitroso-N-methylurea; IM: Intestinal metaplasia.

PLGC cell model

MNNG-induced malignant transformation model of human GES-1: MNNG is commonly used to mimic the in vivo nitrosation process, whereby nitrates are converted into carcinogenic nitrites within the gastric environment. In vitro, MNNG induces malignant transformation in GES-1, promoting enhanced migration, invasion, and survival, as well as activating epithelial-mesenchymal transition (EMT) pathways. This model provides an effective tool for simulating the progression of PLGC[113].

Acute transformation model. Recent studies have established malignantly transformed cells (MCs) by treating GES-1 with MNNG at various concentrations. Doses above 50 μM exhibit significant cytotoxicity, while 30 μM for 24-48 hours is widely accepted for acute induction of transformation[58,74]. In a dose-response assay, MNNG (0-200 μM) exposure led to concentration-dependent reductions in cell viability, with 20-30 μM identified as optimal for inducing DNA damage without compromising survival[114]. At 30 μM for 48 hours, GES-1 cells exhibited approximately 20%-30% viability reduction, sufficient to trigger early transformation events[115,116]. Moreover, exposure to 20 μM MNNG for 24 hours in serum-free medium resulted in extensive cell death, followed by regeneration and stable MC cell lines by the third passage[58,117]. Interestingly, proliferative responses persisted up to 200 μM MNNG, with only 500 μM reversing this trend[118].

Chronic exposure model. Chronic low-dose MNNG effectively induces malignant transformation in GES-1 cells. GES-1 cells treated with 2 μM MNNG for 20 weeks (MC-40) developed malignant phenotypes[119,120]. Another study exposed cells to 1-4 μM MNNG with 0.1% DMSO over 60 passages (approximately 30 weeks), resulting in significantly increased colony formation and EMT marker expression comparable to GC cell line MGC803[121]. Similarly, BGC823 and HGC-27 cells treated chronically with 0.25-0.5 μM MNNG for 60 days showed dose-dependent increases in migration, invasion, vascular mimicry, spheroid formation, and cancer stem cell marker expression[66]. MNU has also been applied in chronic low-dose models to better mimic real-life carcinogen exposure. Song et al[122] cultured GES-1 cells with 0.01 g/L MNU (one-fifth of previously reported doses) for 6 months, successfully inducing malignant transformation while maintaining cell viability. Combined MNNG and H. pylori model. Several studies have combined MNNG treatment with chronic H. pylori infection to enhance transformation efficiency. In a model using the J99 H. pylori strain and daily 5 μM MNNG exposure for 1 year, GES-1 cells exhibited increased proliferation, migration, invasion, EMT, and tumorsphere formation, along with upregulation of cancer stem cell marker CD44[7].

GIM model induced by bile acids in GES-1 cells: Model establishment. DCA and CDCA are used to model bile acid-induced IM in GES-1 cells. GES-1 cells are starved in serum-free medium, then exposed to DCA or CDCA for 24 hours. This method effectively induces IM phenotype[123]. Among various bile acids tested cholic acid, DCA, and CDCA, CDCA was identified as the most potent inducer of IM. Dose- and time-dependent increases in IM markers (e.g., CDX2 and MUC2) were observed following treatment with 0-100 μM of bile acids for 3-24 hours, with CDCA inducing the most significant upregulation of IM-associated markers[123,124]. These findings suggest that CDCA is the optimal inducer for GIM model construction. To optimize the model, GES-1 cells were treated with 50, 100, and 200 μM CDCA. Western blot analysis of CDX2 protein expression revealed a statistically significant increase in CDX2 levels at 100 μM and 200 μM. Based on both efficacy and cell viability, 100 μM CDCA for 24 hours was determined to be the optimal condition for GIM induction[125].

Viability considerations. Cell viability was evaluated using the cell counting kit-8 assay after exposure to various concentrations of DCA and CDCA (50-600 μM) for different durations (6 and 15 minutes; 6, 12, 24, and 48 hours). Interestingly, 200 μM DCA increased GES-1 viability after only 15 minutes of stimulation, simulating the acute onset of bile reflux[23]. However, prolonged exposure (48 hours) to 100 μM or 200 μM of DCA or CDCA significantly reduced cell viability. High concentrations (400-600 μM) were cytotoxic even at short durations. To avoid excessive cytotoxicity while ensuring bile acid stimulation, 24-hour exposure to physiological concentrations (< 200 μM) was selected as the optimal condition for constructing the GIM model[126].

H. pylori-induced transformation model: Acute infection model. Acute H. pylori infection models are commonly established by incubating GES-1 cells (at approximately 80% confluence) in antibiotic-free medium and exposing them to freshly collected H. pylori suspensions at various multiplicities of infection (10:1, 20:1, 50:1, 100:1) for 0, 6, 12, or 24 hours. Bacterial concentrations are quantified using OD600, cell viability, morphology, and proliferation are assessed via high-content screening. Results reveal that H. pylori acutely suppresses GES-1 cell viability and induces morphological changes, suggesting cytotoxic effects of short-term infection[62,127].

Chronic infection model. In the chronic infection model, GES-1 cells are co-cultured with H. pylori at multiplicities of infection of 1:1, 10:1, or 100:1, with daily replacement of fresh bacterial suspension. Co-culture typically continues for up to 6 months. Bacterial density is monitored at OD660 (approximately 2 × 10⁸ CFU/mL per OD). Compared to acute infection, chronic CagA⁺H. pylori exposure induces profound phenotypic alterations in GES-1 cells, including apoptosis resistance, abnormal proliferation, enhanced migration, colony formation, and acquisition of stem cell-like features[128]. This model mimics in vivo transformation during Correa’s cascade and is increasingly applied in organoid-based gastric carcinogenesis research[129].

PLGC gastric organoid model

Patient-derived gastric precancerous organoids: Organoids are 3D clusters of organ-specific cells derived from stem cells with self-organizing capabilities. They retain many functional, molecular, and cellular features of their tissue of origin, including its heterogeneity[130]. Organoids derived from adult stem cells (ASCs) and pluripotent stem cells (PSCs) have become essential preclinical models for cancer research and therapeutic development. Importantly, ASC- and PSC-derived organoids are generated via distinct approaches: ASC organoids rely on niche-specific growth factor-enriched media to maintain physiological stem cell homeostasis, whereas PSC organoids undergo stepwise differentiation through sequential morphogen exposure, mimicking embryonic developmental pathways[131].

Precancerous organoids (PDOs) establishment. PDOs are typically derived from surgical or endoscopic gastric samples, processed via mechanical dissociation and enzymatic digestion, such as Dispase II and Collagenase XI[132], EDTA and TrypLE[133], Liberase TH and TrypLE Express[134], or collagenase and hyaluronidase[135], then embedded in extracellular matrix scaffolds (e.g., Matrigel) for 3D culture. Despite variations in media composition, the core aim is to replicate the gastric stem cell niche to sustain growth and differentiation. One common technical challenge is the contamination or overgrowth of normal epithelial organoids. To enrich for precancerous or neoplastic PDOs, several strategies have been employed, including targeted sampling of lesion-rich areas, selection of regions with high tumor purity, and the use of gene-editing tools[136].

ASC-derived organoids of atrophic gastritis, IM, and Dys. Dysfunction of gastric stem cells underlies atrophic gastritis pathogenesis. Li et al[137] successfully expanded organoids from antral biopsies of non-atrophic gastritis, while most atrophic gastritis samples failed to grow, suggesting impaired gastric stem cell self-renewal. Yoshihiro et al[138] generated atrophic gastritis-derived organoids showing Wnt signaling as essential for maintenance. Bone morphogenetic protein activation induced IM markers CDX2 and MUC2, whereas Smad4 deletion inhibited this. Dual tumor protein 53 (TP53)/Smad4 knockout generated by CRISPR-Cas9 accelerated tumor-like transformation through p21 suppression. These studies highlight the powerful application of gene editing in gastric organoid models, enabling precise interrogation of molecular drivers underlying the transition from atrophy to neoplasia.

IM-derived organoids using optimized stem cell niche factors such as Wnt3A, R-spondin, epidermal growth factor, and Noggin, enabling long-term propagation[139]. Gene-editing of E-cadherin and/or RHOA in normal gastric organoids reproduced IM development[134]. Wei et al[140] group developed IM-mimicking patient-derived organoids revealing OLFM4 and MYH9 cooperation in IM progression. Large-scale RNA-seq revealed heterogeneity and precancerous signatures in IM organoids, identifying STMN1⁺ isthmus stem cells as potential origins. Dysplastic organoids were also derived from resected tissues, showing chromosomal instability and dysplastic stem cell features[141].

For Dys, Yan et al[135] established 46 GC organoid lines, including dysplastic types, from 34 patients. Serosal sampling improved tumor purity, and a modified STAR protocol achieved > 50% success, validated by whole-exome sequencing. Min et al[142] further demonstrated that Meta4 organoids derived from Mist1-Kras (G12D) mice contain dysplastic stem cells with multilineage potential, functioning as key regulators of the transition from PLGC to cancer. Through comparative analyses of dysplastic and Meta4 organoids, cortactin was identified as a mediator of dysplastic progression, highlighting its potential as a clinically relevant biomarker and therapeutic target.

PSC-derived organoids. PSC-derived gastric organoids are generated via stepwise differentiation through definitive endoderm and foregut stages into gastric progenitors within a 3D matrix environment[143]. Lgr5, a Wnt-responsive stem cell marker, has facilitated long-term culture of stable organoids[139]. McCracken et al[144] first derived human PSCs into 3D gastric organoids containing progenitor and differentiated cell types, chief, mucous, and endocrine cells, that responded robustly to H. pylori infection.

Karlsson et al[145] applied CRISPR-Cas9 technology to knock out both alleles of TP53 in human PSCs into 3D gastric organoids and observed their transformation toward IM and tumor-like phenotypes. Single-cell RNA sequencing confirmed loss of mucosal identity. Koide et al[146] used human-induced pluripotent stem cells with CDX2 overexpression to create an incomplete IM model, showing intestinal marker upregulation, MUC5AC loss, and crypt-like morphology with Paneth cell features. The establishment of human gastric organoids from adult and pluripotent stem cells is shown in Figure 4.

Figure 4 Human gastric organoids can be established either from adult stem cells isolated from gastric tissue biopsies or from pluripotent stem cells such as reprogrammed fibroblasts. In the adult stem cell-derived pathway (left), gastric tissue is enzymatically digested to release adult stem cells, which are embedded in Matrigel and cultured to form gastric organoids. In the pluripotent stem cell-derived pathway (right), somatic cells are reprogrammed into induced pluripotent stem cells, differentiated through embryoid body formation and directed endodermal specification to generate gastric organoids. Both organoid systems can be applied to downstream platforms, including patient-derived organoid biobanking, genetic engineering (e.g., CRISPR/Cas9), organ-on-a-chip systems, and advanced assembloid models incorporating stromal or immune components. ASC: Adult stem cell; PDO: Precancerous organoids; PSC: Pluripotent stem cell.

Mouse-derived organoids: Although mouse models cannot fully replicate the physiological and genetic features of the human stomach, organoids derived from murine gastric tissues have become valuable tools for investigating the pathogenesis of PLGC, due to their accessibility, controllability, and compatibility with genetic engineering. These organoids are typically cultured in 3D matrices such as Matrigel and supplemented with niche factors like Wnt3A, epidermal growth factor, and R-spondin1 to maintain stemness and differentiation capacity. They exhibit stable expression of gastric-specific markers and are capable of recapitulating disease progression in vitro[139].

Mouse-derived models of IM and Dys. MNNG-treated mouse gastric organoids exhibit IM-like morphology with CDX2/MUC2 upregulation and ATP4B/MUC6 loss. Lineage tracing confirmed MIST1⁺ chief cells as the source of metaplasia[147]. Dys-Lgr5-p53 knockout organoids derived by Sethi et al[59] showed genome doubling, chromosomal instability, and Wnt/cell cycle pathway activation, mimicking Dys.

Min et al[148] developed two organoid lines Meta3 and Meta4 derived from the gastric corpus of Mist1-Kras(G12D) mice, collected three and four months after tamoxifen induction, respectively. Meta3 organoids recapitulated IM phenotypes, while Meta4 organoids exhibited more pronounced dysplastic characteristics. Using CRISPR-Cas9, they further generated Meta4-cortactin knockout dysplastic organoids and found that cortactin was a critical mediator of membrane dynamics, malignant transformation, and exosome secretion in dysplastic cells, implicating it in the tumor microenvironment remodeling and solid tumor formation. Clinically, its aberrant expression may serve as a biomarker for risk stratification in gastric precancerous lesions, and its inhibition could offer a novel preventive strategy to block progression toward invasive cancer.

Transgenic mouse-derived organoids. Although engineered mouse models do not fully capture the entire spectrum of human GC, particularly distant metastasis[59], they remain indispensable tools for dissecting the molecular mechanisms of PLGC. Lu et al[149] developed TMPC (Trp53^-/^-; Myc; sgPten; sgCdkn2b) gastric organoids that recapitulate Dys-to-metastasis progression post-transplantation. The system allows rapid assembly of driver mutations using CRISPR/Cas9, making it an ideal platform for functional oncogene validation. In a related study, mutations in TP53 and ARID1A were introduced into human gastric organoids, which formed malignant tumors upon subcutaneous transplantation into immunodeficient recipient mice[150]. These findings highlight the power of transplanting genetically modified PDOs into host systems as a versatile strategy for modeling diverse cancer phenotypes.

Organoid models for immune microenvironment and H. pylori infection: Organoid-immune interactions. PLGC development is closely linked to the inflammatory microenvironment. Osaki et al[151] established gastric organoids from transgenic mice and showed that exogenous IFN-γ induced structural disorganization and apoptosis, indicating that chronic inflammatory cytokines directly drive atrophic gastritis and IM in gastric epithelium. Conditioned media from activated immune cells of autoimmune atrophic gastritis mice (TxA23), with or without IFN-γ, were added to organoid cultures. Pathological changes occurred only in organoids expressing IFN-γ receptors, confirming the role of inflammatory signaling. Organoid-immune co-culture systems further enable immune interaction studies. A gastric injury model was created via MNNG exposure, followed by H. pylori microinjection. Co-culture with RAW264.7 macrophages revealed dynamic macrophage-organoid interactions over 24-72 hours, including inflammatory modulation and polarization, modeling immune-driven tumor promotion in the PLGC microenvironment[152].

Organoid models of H. pylori infection. Gastric organoids, with their 3D architecture closely resembling native glandular structure, offer an ideal platform to study H. pylori-induced carcinogenesis. Guenther et al[153] generated human gastric organoids containing pit mucous, gland mucous, chief, and enteroendocrine cells. H. pylori infection strongly activated nuclear factor-κB signaling and pro-inflammatory cytokine secretion. H. pylori-induced tumorigenesis also involves metabolic pathways. In Smox^-/^- organoids, reduced inflammation, DNA damage, and β-catenin signaling confirmed the role of polyamine metabolism[154]. The bacterium also upregulates ACVR1, activating the IRF3/POLD1 axis and impairing DNA repair, promoting genomic instability[82].

Single-cell RNA sequencing has revealed distinct cellular compositions across different organoid configurations. While 2D monolayer cultures are enriched in pit-like epithelial cells, 3D organoids are predominantly composed of glandular progenitor cells[155]. Comparative studies demonstrated that apical-out and 2D organoids, with their luminal surface exposed and greater experimental accessibility, significantly enhance H. pylori adhesion and facilitate the study of early epithelial responses to infection, such as barrier disruption and inflammation[156].

To better mimic physiology, a dual-channel organoid-on-chip platform was developed, supporting H. pylori colonization for up to six days under sustained luminal acidity and promoting pit cell maturation. This system revealed unique infection responses in mature pit cells absent in conventional models, advancing the study of chronic infection and gastric microenvironment[157]. However, current organoid-on-chip platforms still represent partial reconstructions of gastric tissue, as they do not yet fully integrate multicellular stromal networks or systemic immune circulation. Construction and application of organoid models in PLGC research are summarized in Figure 5.

Figure 5 Construction and application of organoid models in precancerous lesions of gastric cancer research. Organoids can be derived from patient tissues, pluripotent stem cells, or gastric tissues of genetically engineered mice: (1) Patient-derived organoids preserve key histological and molecular features of intestinal metaplasia (IM), dysplasia, and early gastric cancer, enabling personalized modeling of disease progression; (2) Pluripotent stem cell-derived organoids recapitulate gastric development and can be genetically modified to simulate metaplastic transitions (e.g., caudal-related homeobox transcription factor 2-driven IM); and (3) Mouse-derived organoids, including those from models like Mist1-Kras and leucine-rich repeat-containing G protein-coupled receptor 5-p53^-/^-, facilitate studies of genetic alterations in spasmolytic polypeptide-expressing metaplasia, IM, and dysplasia. Co-culture systems with immune cells or Helicobacter pylori further allow mechanistic investigations into inflammation, epithelial remodeling, and host-pathogen dynamics. These platforms provide versatile tools for in vitro modeling of the Correa cascade. PLGC: Precancerous lesions of gastric cancer; IM: Intestinal metaplasia; PSC: Pluripotent stem cell; Dys: Dysplasia.

DISCUSSION

GC develops through the Correa cascade, with PLGC representing a pivotal transitional stage. Accurate PLGC modeling is essential for dissecting early oncogenic events and evaluating preventive strategies. Chemically induced models (e.g., MNNG, MNU, bile acids) have recapitulated epithelial injury and transformation, though outcomes vary depending on dose, duration, and route, highlighting the importance of tailored model selection. A decision framework for selecting precancerous lesions of gastric cancer research models is shown in Table 3. However, these models only partially mimic human PLGC, as they lack the sustained immune infiltration and stromal remodeling characteristic of H. pylori-driven disease. Approximate timeframes for the development of key stages in the Correa cascade across different model systems are summarized in Figure 6, providing guidance for aligning study duration with specific lesion progression.

Figure 6 Integrates data from multiple studies to show approximate timeframes for the development of key stages of the Correa cascade. Model systems are grouped by primary induction method. Note that the exact timing can vary based on specific protocol, species, strain, and dose. Temporal scale: Chemical induction (N-methyl-N’-nitro-N-nitrosoguanidine/N-Nitroso-N-methylurea) offers a relatively predictable and accelerated timeline for carcinogenesis, especially in rats. Natural history: Helicobacter pylori (H. pylori) infection models, particularly in Mongolian gerbils, best recapitulate the natural, multi-stage progression of human gastric carcinogenesis but require significantly longer experiments. Acceleration: Composite models (e.g., H. pylori + N-Nitroso-N-methylurea) and genetically engineered models (e.g., insulin-gastrin mice) can dramatically accelerate the development of advanced lesions and cancer. Model selection: This visual aid helps researchers align their experimental timeframe with their chosen model. For instance, a 6-month study might be sufficient to observe metaplasia in a composite model but would only capture early-stage gastritis in a standard H. pylori infection model in mice. MNNG: N-methyl-N’-nitro-N-nitrosoguanidine; IM: Intestinal metaplasia; H. pylori: Helicobacter pylori; SPEM: Spasmolytic polypeptide-expressing metaplasia.

Table 3 Decision framework for selecting precancerous lesions of gastric cancer research models.

Research objective	Model	Induction/approach	Advantages	Limitations
SPEM mechanisms	Mouse (acute drug-induced)	DMP-777 (10-14 days), L635 (10-14 days)	Rapid, reversible, specific parietal cell loss	Lacks chronic inflammatory context
IM	Mouse (bile acid-induced), GES-1 cells, gastric organoids	Mouse: 0.2% DCA in drinking water (6 months); GES-1: 100 μM CDCA for 24 hours. Organoids: BMP activation	In vivo: Recapitulates IM progression. In vitro: Rapid, high-throughput. Organoids: Human-specific, physiologically relevant	In vivo: Long duration. In vitro: Lack tissue complexity. Organoids: Lack full tumor microenvironment
H. pylori pathogenesis	Mongolian gerbil, C57BL/6 mouse	H. pylori inoculation (e.g., TN2GF4, 7.13, PMSS1)	Gerbils: Full Correa cascade to adenocarcinoma. Mice: Genetic tools available, suitable for immune studies	Gerbils: Limited genetic tools. Mice: Resistant to many H. pylori strains
Multifactorial carcinogenesis	Composite rat/mouse models	MNNG/MNU + H. pylori; MNNG/MNU + high-salt diet + ranitidine	Models human synergistic etiology; high clinical relevance	Complex setup, multiple variables to control
Genetic and molecular mechanisms	GEMMs (e.g., INS-GAS, p53KO), CRISPR-edited organoids	Genetic manipulation (e.g., KrasG12D, TP53-/-)	Definitive genotype-phenotype studies; precise temporal control	High cost, technical complexity
Drug screening and toxicity	Cell lines (GES-1, BGC823)	Acute or chronic MNNG/MNU treatment or H. pylori co-culture	High-throughput, low cost, reproducible	Simplified system, lacks in vivo physiology
Immune-microenvironment interactions	Immune-organoid co-culture	Co-culture with macrophages (RAW264.7) or PBMCs	Incorporates human immune components; personalized potential	Technically challenging; not fully vascularized

IM: Intestinal metaplasia; H. pylori: Helicobacter pylori; GES-1: Gastric epithelial cell-1; GEMM: Genetically engineered mouse models; INS-GAS: Insulin-gastrin; CDCA: Chenodeoxycholic acid; DCA: Deoxycholic acid; MNU: N-Nitroso-N-methylurea; MNNG: N-methyl-N’-nitro-N-nitrosoguanidine; TP53: Tumor protein 53; PBMCs: Peripheral blood mononuclear cells.

Compared to chemical agents, H. pylori infection better mimics human pathology, with strain- and host-specific responses in gerbils and mice. Multifactorial models combining H. pylori, carcinogens, diet, and host genetics enhance fidelity by capturing synergistic mechanisms. Acute drug-induced models have elucidated precursor lesions such as SPEM, while genetically engineered mice allow precise manipulation of tumor-related genes and immune dynamics. In addition, stress-related models, such as restraint stress or tail-clamping paradigms, have been used to explore how psychological and neuroendocrine factors modulate gastric mucosal injury and carcinogenic progression, highlighting the influence of host systemic responses[158]. Human GES models offer a practical platform for mechanistic studies and high-throughput screening. While they replicate early molecular alterations induced by carcinogens or infection, their 2D nature and lack of tissue architecture curtail their ability to model disease heterogeneity and cellular crosstalk observed in vivo.

ASC-derived organoids recapitulate renewal, IM, and Dys. CRISPR-Cas9-mediated gene editing has yielded tractable PLGC models for genotype-phenotype correlation studies[134]. PSC-derived organoids mimic developmental signaling and robustly respond to H. pylori infection. Innovations such as organoid-on-chip platforms and apical-out polarity improve physiological relevance[157]. Co-culture with immune cells and organoid assembloids enables exploration of inflammation-driven transformation and tumor-immune interactions[152]. Nevertheless, such co-culture systems remain simplified, typically incorporating limited immune cell subsets and lacking vascular perfusion, thereby constraining their ability to fully recapitulate the spatial and temporal complexity of in vivo immune regulation.

Despite their promise, organoids face limitations, including the absence of vasculature, stromal architecture, and a fully competent immune system, as well as challenges in reproducibility across laboratories. These constraints limit their ability to model chronic inflammation, immune-mediated epithelial remodeling, and stromal-epithelial interactions that are central to human PLGC progression. However, advances in CRISPR gene editing, microfluidics, and single-cell multi-omics are mitigating these barriers. Integration with immune co-culture systems, organoid-on-chip platforms, and PDO biobanks is accelerating biomarker discovery and personalized therapy development[159,160]. Future PLGC modeling will increasingly combine chemical, microbial, dietary, genetic, and stress-related paradigms to recapitulate multifactorial tumorigenesis. Spatial transcriptomics, lineage tracing, and next-generation organoid systems incorporating vasculature, stroma, and immune components are expected to bridge the gap between reductionist models and the human gastric microenvironment.

CONCLUSION

Animal, cellular, and organoid models provide complementary insights into PLGC. Their continued refinement, through genetic engineering, 3D culture, immune modeling, and stress paradigms, will enhance understanding of early gastric tumorigenesis and facilitate development of early detection, prevention, and personalized therapeutic strategies.