Gastric intestinal metaplasia: Management and surveillance strategies

Abstract

Gastric intestinal metaplasia (GIM) is a common histological finding often associated with Helicobacter pylori infection, placing patients at high risk of gastric cancer. The Correa cascade is a pathway that describes the progression of GIM from precancerous to cancerous following a sequence of inflammation-atrophy-metaplasia-dysplasia and carcinoma. Mortality can be reduced through early detection and follow-up screening. There is a wide variation in clinical management of GIM, which makes it difficult to determine the degree of adherence to established guidelines and recommendations. All guidelines emphasize the importance of high-definition endoscopy and targeted biopsy (taken separately from the antrum and corpus) to improve risk stratification. This highlights the need for standardized protocol for GIM management and risk-based follow-up. The main challenge remains the lack of global standardization leading to inconsistency in follow-up initiation and discontinuation, in addition to low adherence to guidelines due to resource limitations. This is progressively enhanced by the development of a personalized follow-up plan based on the individual patient’s case and a unified risk-based surveillance protocol.

Key Words: Gastric intestinal metaplasia; Gastric cancer prevention; Helicobacter pylori eradication; Endoscopic surveillance; Risk stratification

Core Tip: Challenges in the standardized management and surveillance of gastric intestinal metaplasia persist. Variations between the American Gastroenterological Association and MAPS III guidelines highlight the need for a unified, risk-based global approach tailored to regional cancer incidence and healthcare resources. Early detection through high-definition endoscopy, the eradication of Helicobacter pylori, and precise risk stratification using systems like operative link on gastritis assessment and operative link on gastric intestinal metaplasia have emerged as crucial strategies. Artificial intelligence offers promise in enhancing diagnostic accuracy and predicting malignant transformation, although real-world application requires broader validation. The future of gastric intestinal metaplasia management lies in the integration of multimodal risk assessment, combining endoscopic, histological, molecular, and artificial intelligence driven data, to personalize surveillance and optimize outcomes.

INTRODUCTION

Gastric intestinal metaplasia (GIM) is a pathological process linked to chronic inflammation in which gastric mucosa undergoes intestinal-type differentiation, often following injuries such as gastritis or peptic ulcers[1,2]. It is considered a precancerous condition within the Correa cascade, which describes the progression from chronic inflammation to non-atrophic gastritis (AG), gastric atrophy, intestinal metaplasia (IM), dysplasia, and ultimately gastric adenocarcinoma (GA)[3,4].

GA comprises two biologically and pathogenetically distinct entities as defined by the Lauren classification: Intestinal-type and diffuse-type gastric cancer (GC). Intestinal-type GC develops through a well-characterized, multistep inflammatory cascade, commonly referred to as the Correa pathway, progressing from chronic Helicobacter pylori (H. pylori)-associated gastritis to AG, IM, dysplasia, and ultimately carcinoma. It is more common in older adults, and marked by glandular, well-differentiated cancer cells. In this context, GIM represents an established premalignant lesion and a key target for surveillance strategies[5,6].

In contrast, diffuse-type GC, commonly seen in younger patients arises through a distinct pathogenetic mechanism characterized by poorly cohesive tumor cells, loss of cell-cell adhesion, and frequent alterations in cadherin 1 and other adhesion-related genes. This subtype typically lacks a defined premalignant sequence and may develop in the absence of chronic AG or IM, limiting the utility of endoscopic surveillance strategies based on precursor lesions[7]. Accordingly, risk stratification and surveillance paradigms centered on GIM are intended to identify individuals at increased risk for intestinal-type GA, and should not be extrapolated to diffuse-type disease, which follows a separate molecular and clinical trajectory.

While GIM does not always lead to cancer, the presence of dysplasia significantly raises the risk, warranting careful evaluation and follow-up[2]. Early detection is vital, as most GCs are diagnosed at advanced stages[2,8]. However, standardized protocols for endoscopic surveillance of precancerous gastric lesions remain lacking[9]. Emerging research suggests that improved adherence to national guidelines at the institutional level can enhance early diagnosis and management of GIM, reducing progression to GC[10].

This review aims to synthesize and evaluate recent international guidelines for GIM management, including follow-up strategies and the impact of H. pylori eradication. It will also explore the role of artificial intelligence (AI) and advanced endoscopic tools in risk prediction and monitoring, as well as current challenges, regional disparities, and future directions for personalized surveillance approaches.

RISK STRATIFICATION IN IM

Risk factors associated with GIM

In the United States, certain populations are at higher risk for GC and may benefit from targeted screening[11]. These include first-generation immigrants from high-incidence regions, non-White racial and ethnic groups, individuals with a family history of GC and those with hereditary cancer syndromes such as Lynch syndrome or familial adenomatous polyposis. Universal screening is not practical due to limited endoscopic resources, making personalized risk-based screening essential. Endoscopy remains the only approved screening method.

Risk factors influencing screening decisions include age, H. pylori status, family history, country of origin, immigration generation, socioeconomic status, tobacco use, and residence in persistently impoverished areas—defined as regions where ≥ 20% of residents have lived below the poverty line for over 40 years[12]. Age 45 is proposed as a reasonable starting point for screening high-risk individuals, aligning with colorectal cancer screening and allowing for combined procedures[13]. Earlier screening is recommended for those with hereditary syndromes or a first-degree relative diagnosed with GC typically 10 years before the relative’s age at diagnosis.

Immigrants from high-incidence regions retain elevated risk, which may decline over generations due to modifiable lifestyle and environmental changes. Factors such as poor diet, smoking, and limited healthcare access increase risk, while diets rich in fruits, vegetables, and lean proteins are protective. Given current evidence, personalized risk assessment remains the most appropriate strategy for GC screening in the United States[11,12,14].

The prevalence of GIM in high-risk groups

A 2019 technical review by the American Gastroenterological Association (AGA) analyzed 53 studies, with 44 (83%) reporting H. pylori prevalence. Studies with > 15% H. pylori exposure generally showed higher GIM prevalence, though findings were inconsistent and often biased. Meta-regression found no significant correlation between H. pylori and GIM prevalence (P = 0.85), likely due to diagnostic variability. In six studies (n = 7121), pooled GIM prevalence among H. pylori-exposed patients was 25%, with regional variation[5].

Regarding H. pylori virulence factors, three studies (n = 3068) showed GIM prevalence was highest in cytotoxin-associated gene A (CagA)-positive patients (36.4%), followed by CagA-negative (21.3%) and uninfected individuals (17.8%). However, inconsistent detection of CagA antibodies limited the strength of evidence[5,15-17].

Ethnic and racial disparities in GIM were reported in three United States studies (n = 1434), with Hispanics showing the highest prevalence (P < 0.01), though this finding was driven by a small subgroup (n = 58) reporting 50%, while a larger group (n = 162) showed 15.4%, similar to other groups. One study found GIM in 88.9% of patients with pernicious anemia[13]. Additionally, a noncomparative study by Leung et al[18] reported 30% GIM prevalence in individuals with a first-degree family history of GC.

Endoscopic evaluation for AG and GIM in high risk patients

The risk of GC associated with AG and GIM is influenced by factors such as age, H. pylori infection, the extent and severity of AG/GIM, and GIM subtype (complete vs incomplete)[5,19]. Although histologic subtyping of GIM into complete and incomplete forms has been consistently associated with differential GC risk, its use for risk stratification in real-world clinical practice remains variable. Incomplete-type IM has repeatedly been shown to confer a higher risk of progression to dysplasia and GC compared with complete IM, likely reflecting greater molecular instability and a closer resemblance to intestinal-type GA[9]. Consequently, major international guidelines, including MAPS III, explicitly incorporate incomplete-type IM as a modifier that justifies endoscopic surveillance even in patients with otherwise limited disease[20].

However, in routine practice, systematic reporting of IM subtypes is inconsistent. Several real-world studies and expert reviews highlight substantial interobserver variability among pathologists, limited reproducibility outside specialized centers, and incomplete adoption of IM subtyping in standard pathology reports, particularly in Western, low-incidence settings[5,9]. As a result, clinicians often rely more heavily on the extent of IM (antrum-limited vs extensive) and validated staging systems such as operative link on GIM (OLGIM), which demonstrate superior reproducibility and prognostic value across practice settings[21-23].

Histological interpretation of incomplete GIM remains a challenge, though additional feedback and training improve diagnostic precision, interobserver agreement remains lower than for complete GIM, notably among less experienced pathologists, underscoring the need for standardization and optimized training thus significantly impacting patient care and surveillance strategies[24]. It has also been noted that in the United States, reports of GIM subtype by pathologists are infrequent which raises question about the viability of using IM subtypes for risk stratification[25]. On the other hand, it has been shown that diagnostic performance is higher among pathologists who routinely apply GIM subtyping and use it as risk stratification tool, with low misclassification rate noted among expert pathologists[24].

Hematoxylin and eosin stained slides has been a reliable tool for the assessment of occurrence, extent and subtyping of GIM, with finding measuring up to endoscopic grading, in addition to high interobserver agreement, beyond that seen for non-metaplastic gastric atrophy[25]. Furthermore, digital microscopy demonstrated reduced diagnostic efficiency compared to traditional light microscopy, with issues in the identification of some specific cells. Nonetheless, there has been increasing familiarity with whole slide imaging, especially among less experienced pathologist, which is likely narrowing the disparity in performance between digital and glass slides[24]. Additionally, the probability of detecting GIM on gastric biopsies depends on the number of biopsies taken, which along with the portion of mucosal surface involved by GIM and the pattern, pure or mixed, all affect the diagnostic interobserver agreement[25].

GIM is identifiable with high-definition white light endoscopy (WLE), especially when enhanced with narrow band or linked color imaging, offering over 85%-90% sensitivity and specificity[26]. In East Asia, global endoscopic risk assessment is often sufficient, and biopsies are reserved for suspected neoplasia[27]. In the United States, due to lower detection rates, systematic biopsy following the updated Sydney system is recommended, with at least five site-labeled samples[11,28].

A 2024 study by Fang et al[29] introduced GasMIL, a deep learning algorithm for diagnosing and grading GIM and AG using biopsy specimens. Trained on 2725 whole-slide images from 545 patients, GasMIL outperformed 10 pathologists, achieving area under the curves (AUCs) of 0.884 for IM and 0.877 for atrophy. As a diagnostic aid, GasMIL significantly improved pathologists’ accuracy, demonstrating strong potential to enhance GC risk stratification and clinical decision-making.

UPDATED INTERNATIONAL GUIDELINES ON SCREENING AND SURVEILLANCE OF GIM AND PRECANCEROUS LESIONS

The AGA (2020) and MAPS III (2025) are the two most recent updated guidelines, which can be adapted in the clinical practice of most gastroenterologists regarding the screening and surveillance of patients diagnosed with or at high risk of developing GIM. It is important to emphasize that international guidelines addressing GIM, including those from the AGA, MAPS III (European Society of Gastrointestinal Endoscopy/European Helicobacter and Microbiota Study Group/European Society of Pathology), and Eastern Asian societies, differ not only in content but also in the strength of their recommendations. Many statements particularly those concerning surveillance initiation, surveillance intervals, and management of low-risk GIM are explicitly conditional and intended to guide shared decision-making rather than mandate uniform practice. These recommendations are frequently based on low- to moderate-quality evidence and are highly dependent on regional GC incidence, patient risk profiles, healthcare resources, and patient preferences. Accordingly, guideline adherence should be interpreted within the clinical context rather than as definitive or universally applicable directives, especially in Western, low-incidence settings.

AGA recommendations for GIM screening and surveillance

Several guidelines and recommendations were put forth by the AGA to optimize the level of screening and surveillance for GIM. Only the most important guidelines will be mentioned based on the level of evidence and importance. First, screening should be prioritized for identifiable high-risk groups, including first-generation immigrants from high-incidence GC regions, individuals with a family history of GC, and certain racial and ethnic minorities. These populations exhibit a significantly elevated risk for GC, warranting targeted screening efforts. Second, endoscopy is the preferred method for screening and surveillance in individuals at increased risk for GC. It allows for direct visualization of the gastric mucosa, enabling prompt identification of precancerous lesions and biopsies for histological analysis, which is critical for effective risk stratification. Third, H. pylori eradication is essential for primary and secondary prevention of GC, given its significant role in the disease’s etiology. Opportunistic screening for H. pylori should be considered in high-risk individuals as well as their close contacts to mitigate GC risk. Fourth, in cases of suspected gastric atrophy or IM, a systematic biopsy protocol should be implemented to ensure accurate histological confirmation. This protocol should include a minimum of 5 biopsies from designated anatomical sites such as the antrum/incisura and corpus to support effective diagnosis and risk assessment. Fifth, individuals with confirmed gastric atrophy or GIM should undergo careful risk stratification. Surveillance strategies should be personalized based on the individual’s risk factors (e.g., family history), with endoscopic surveillance potentially every 3 years for those with severe conditions[11,21,28,30].

Importantly, most AGA recommendations regarding GIM surveillance are conditional rather than strong recommendations. The AGA does not advocate routine surveillance for all patients with GIM and explicitly discourages automatic endoscopic follow-up in individuals with low-risk features, such as complete, antrum-limited GIM without additional risk factors. Instead, the guideline emphasizes individualized risk assessment and shared decision-making, recognizing uncertainty in progression risk estimates and the absence of definitive evidence demonstrating mortality benefit from universal surveillance in Western populations. Surveillance decisions should therefore incorporate patient values, comorbidities, life expectancy, procedural risk, and local resource availability.

MAPS III: European Society of Gastrointestinal Endoscopy, European Helicobacter and Microbiota Study Group and European Society of Pathology Guideline 2025 on GIM screening and surveillance

Like the AGA (2020) recommendations, MAPS III (2025) was also done with some key differences compared to AGA (2020) and some modifications on the guidelines compared to MAPS II (2020). Only the important recommendations will be mentioned based on the level of evidence and importance.

Population and individual-level screening: First, endoscopic screening is suggested every 2-3 years in high-risk regions [age-standardized rate (ASR) > 20/100000] and every 5 years in intermediate-risk regions (ASR 10-20/100000), contingent on cost-effectiveness. Second, no screening is recommended in low-risk regions (ASR < 10/100000). Third, all first-time diagnostic endoscopies should include risk stratification and assessment for precancerous conditions, regardless of origin. Fourth, targeted screening is recommended for first-degree relatives of GC patients with H. pylori testing and eradication at age 20-30, in addition to endoscopic screening at age 45 or 10 years before the relative’s diagnosis. Fifth, discontinue screening in individuals > 80 years or with limited life expectancy[20,31,32].

Diagnostic endoscopy and histology: First, high-quality endoscopy using virtual chromoendoscopy (VCE) is recommended for diagnosis, staging, and surveillance. Second, VCE should guide biopsies, with random biopsies only if no visual lesions are detected. Third, a minimum of 4 biopsies (2 from antrum/incisura and 2 from corpus) in separate vials is recommended. Fourth, use validated classifications [e.g., Kimura-Takemoto, endoscopic grading of GIM (EGGIM)] for endoscopic staging. Fifth, histopathology should include dysplasia, atrophy, IM, and H. pylori status[20,33].

GIM is a common finding during upper gastrointestinal endoscopy, though it remains difficult to reliably diagnose via endoscopy alone, making histological confirmation by biopsy indispensable. Only a limited number of studies have correlated endoscopy and histological findings of GIM. Previous studies linked incomplete GIM subtypes to higher risk of GC, although IM subtype does not affect endoscopic detectability and histological GIM grading, supported by higher CDX2 expression, proved to be more relevant. Additionally, lower grade GIM (grade 1) where WLE showed inconclusive findings and required histologic confirmation.

Local and systemic inflammation also influence endoscopic detection, as its sensitivity for GIM improved in the presence of AG and active inflammation in the antrum, which was linked to the possible enhanced awareness of atrophy related lesions and inflammation induced mucosal nodularity. Finally, endoscopic diagnosis of GIM by WLE has limited value, and in the absence of endoscopic AG, histological confirmation of diagnosis remains necessary[34] (Table 1).

Table 1 Comparative summary of diagnostic technologies for gastric intestinal metaplasia.

Ref.	Design	Population	Main findings	Strengths	Critique
Yao et al[54]	Multicenter randomized controlled trial (training evaluation)	Asian and Western centers	E-learning improved detection of early GIM features (light-blue crest, white opaque substance)	International applicability; structured training intervention	No measurement of actual cancer prevention outcomes; inter-center variability possible
Yan et al[45]	AI model development study	Asian single-center cohort	CNN achieved area under the curves approximately 0.93 for GIM detection	High diagnostic accuracy; real-time feasibility	Model trained on homogeneous data; no external Western validation
Iwaya et al[56]	AI histopathological analysis study	Asian biopsy samples	Deep learning identified IM with > 95% sensitivity	Potential to automate pathology workload; reproducible	Retrospective; lack of prospective clinical deployment; reliance on digitized slides limits real-world application
Ligato et al[55]	AI corpus-focused endoscopic model	Western European cohort	CNN detected corpus IM with area under the curves approximately 0.89	Addresses Western underrepresentation; corpus-specific model	Small sample size; limited diversity across European centers; early-stage validation only

GIM: Gastric intestinal metaplasia; AI: Artificial intelligence; CNN: Convolutional neural network; IM: Intestinal metaplasia.

Surveillance of precancerous conditions: Surveillance of precancerous conditions should be done every 3 years for patients with extensive endoscopic findings (Kimura C3+/EGGIM 5+) or advanced histology [operative link on gastritis assessment (OLGA)/OLGIM III/IV)], every 1-2 years if the above findings are present plus a family history of GC, and every 3 years for patients with isolated IM plus one risk factor (family history, incomplete type, or persistent H. pylori). No surveillance for mild to moderate atrophy or IM limited to the antrum, unless other risk factors (family history, incomplete IM, persistent H. pylori) exist[20].

Although MAPS III provides structured surveillance intervals based on endoscopic and histologic staging, these intervals should not be interpreted as rigid mandates. The guideline acknowledges that surveillance intensity should be adapted according to patient age, comorbidity burden, competing mortality risks, and regional GC incidence, and that shared decision-making remains essential particularly when applying MAPS III recommendations in low-incidence Western populations.

Endoscopic resection and risk-based management: First, endoscopic submucosal dissection is recommended for differentiated lesions (dysplasia or intramucosal carcinoma, any size if non-ulcerated, ≤ 30 mm if ulcerated) and undifferentiated lesions ≤ 20 mm without ulceration. Second, post-resection management is considered based on histological risk: (1) Very low risk (curative): No further treatment; surveillance only; (2) Low risk: Staging completed; further treatment is typically unnecessary but should involve multidisciplinary team discussion; (3) Local risk: Surveillance/re-treatment recommended; and (4) High risk (noncurative): Staging and consideration for additional therapy, including surgery[20].

Role of H. pylori, additional preventive measures, and special populations: Observational studies have consistently suggested an inverse association between aspirin use and GC incidence, with several large pooled and population-based analyses reporting a duration- and dose-dependent reduction in risk. Recent meta-analyses and consortium-based studies indicate that prolonged aspirin exposure may be associated with a modest reduction in GC risk, particularly in high-incidence regions and among long-term users[35,36]. However, these findings are derived exclusively from observational data, and no randomized controlled trials have definitively demonstrated a causal chemopreventive effect of aspirin on GC.

Importantly, the interpretation of observational associations is limited by residual confounding and indication bias. Aspirin users often differ systematically from non-users with respect to healthcare engagement, comorbidity burden, socioeconomic status, and concurrent use of gastroprotective or cardioprotective therapies. Moreover, aspirin use is frequently prescribed for established cardiovascular indications, which may correlate with differences in surveillance intensity or competing mortality risks, thereby complicating causal inference[36]. These limitations preclude the routine recommendation of aspirin solely for GC chemoprevention.

From a clinical standpoint, current gastrointestinal and cardiovascular guidelines do not endorse aspirin for primary cancer prevention in the absence of a clear cardiovascular indication. Updated European guidance on the management of gastric premalignant conditions emphasizes surveillance and H. pylori eradication as the cornerstone preventive strategies, while explicitly noting the lack of randomized evidence supporting chemopreventive pharmacotherapy for GC[30]. In parallel, contemporary cardiovascular literature highlights that aspirin use in primary prevention is associated with a nontrivial risk of gastrointestinal bleeding, particularly in older adults, often offsetting potential benefits[37].

Accordingly, aspirin should not be recommended solely for the prevention of GC. Its use may be considered only in selected patients with established cardiovascular indications, after individualized assessment of bleeding risk and in accordance with cardiovascular prevention guidelines[38]. Until randomized trial data directly addressing GC outcomes become available, aspirin’s role in GC prevention should remain secondary and opportunistic rather than intentional.

It’s essential to take into consideration special populations with unique medical conditions, like patients with autoimmune gastritis, which requires endoscopic follow-up every 3 years; common variable immunodeficiency, which requires initial endoscopy, then surveillance based on findings; and finally hereditary syndromes, which require following specific syndrome guidelines or mucosal changes, whichever suggests a shorter interval[20,39].

Integrated risk-based clinical decision-making in GIM

Although GIM is widely recognized as a premalignant condition, its progression to GC is highly heterogeneous. Current international guidelines uniformly emphasize that surveillance decisions should be based on the cumulative interaction of multiple risk modifiers, rather than any single histologic or clinical variable in isolation. Importantly, no validated scoring system currently integrates all GC risk factors into a unified algorithm, and surveillance recommendations therefore rely on structured risk-based clinical judgment informed by guideline frameworks and epidemiologic evidence[20]. Key modifiers consistently associated with increased GC risk include the extent of IM, histologic subtype (complete vs incomplete), OLGIM stage, family history of GC, and persistent H. pylori infection[5,20]. When considered together, these variables allow stratification of patients into pragmatic risk categories that can guide surveillance eligibility and interval selection.

Patients with complete-type IM limited to the antrum, corresponding to OLGIM stage I-II, without a family history of GC and with documented eradication or absence of H. pylori are considered low risk. In this population, both the AGA and MAPS III guidelines do not recommend routine endoscopic surveillance, given the low absolute risk of progression and lack of evidence demonstrating a mortality benefit from surveillance in low-incidence regions[5,20]. Management should instead focus on H. pylori eradication, lifestyle modification, and shared decision-making.

High-risk patients are characterized by one or more of the following features: Extensive IM involving both antrum and corpus, incomplete-type IM, advanced histologic staging (OLGIM III-IV), persistent H. pylori infection, or a first-degree family history of GC. Robust evidence from prospective cohort studies and meta-analyses demonstrates that OLGIM III-IV stages are associated with a markedly increased risk of GC, with risk ratios exceeding 10 compared with lower stages[22,23].

In such patients, surveillance endoscopy is justified. MAPS III recommends surveillance every 3 years for extensive or advanced-stage disease and every 1-2 years when additional high-risk modifiers particularly family history are present[20]. While the AGA adopts a more conservative stance, it acknowledges that surveillance may be appropriate in these high-risk settings following individualized risk assessment and patient-centered discussion[5,20].

An intermediate-risk group includes patients with antrum-limited IM accompanied by a single additional risk modifier, such as incomplete histologic subtype, family history of GC, or uncertain H. pylori status. Although absolute cancer risk in this group is lower than in patients with extensive disease, multiple guideline documents support consideration of surveillance at approximately 3-year intervals, particularly when patient preferences, comorbidities, and regional cancer incidence are taken into account[5,20].

This integrative, risk-based approach reconciles differences between international guidelines by translating individual risk modifiers into practical surveillance strategies without imposing rigid algorithms unsupported by current evidence. Surveillance decisions should remain individualized, incorporating patient age, comorbidities, life expectancy, procedural risk, and local healthcare resources. Importantly, international guidelines should be interpreted as flexible frameworks rather than prescriptive rules, particularly in Western, low-incidence populations where overt surveillance may expose low-risk patients to unnecessary procedures without proven survival benefit[30].

Key differences between AGA (2020) and MAPS III (2025) guidelines on GIM screening and surveillance

The AGA (2020) and MAPS III (2025) guidelines both offer updated recommendations on the screening and surveillance of GIM but differ in several important aspects. The AGA emphasizes a more individualized, risk-based approach, focusing primarily on high-risk groups such as first-generation immigrants from high GC incidence regions, racial and ethnic minorities, and individuals with a family history of GC, while MAPS III adopts a broader public health framework by recommending population-based screening intervals depending on the regional incidence of GC. In terms of diagnostic endoscopy, the AGA recommends standard high-quality endoscopy with systematic biopsies, whereas MAPS III places strong emphasis on the use of VCE for targeted biopsy and staging, with a minimum of four biopsies placed in separate vials. For surveillance, AGA suggests personalized intervals, typically around every 3 years for high-risk patients, whereas MAPS III provides a more detailed stratification based on endoscopic (EGGIM/Kimura-Takemoto) and histologic (OLGA/OLGIM) staging, recommending surveillance every 1-3 years depending on the severity and additional risk factors. Additionally, MAPS III gives more explicit guidance on post-endoscopic resection management and advocates the use of validated staging systems, while the AGA focuses more broadly on primary and secondary prevention strategies, particularly H. pylori eradication and lifestyle interventions. Overall, MAPS III offers a more structured algorithmic approach, while the AGA guidelines are more flexible and centered on individualized clinical judgment.

These differences reflect not only divergent epidemiologic contexts but also contrasting philosophical approaches to guideline development. MAPS III adopts a population-based preventive framework suitable for moderate- to high-incidence regions, whereas the AGA framework is intentionally conservative, emphasizing conditional recommendations and individualized decision-making in the absence of robust Western outcome data. As such, surveillance strategies should not be transferred wholesale across regions; rather, clinicians should contextualize guideline recommendations within local cancer risk, healthcare infrastructure, and patient preferences.

OLGA/OLGIM STAGING SYSTEMS ON THE RISK ASSESSMENT OF GC

The OLGA system was developed to stage chronic gastritis, while the OLGIM system later emerged as a simplified, more consistent method for staging IM. Both systems are based on the Sydney system and rely on histopathological analysis of biopsy samples obtained during gastroscopy. They help quantify the extent of atrophic and metaplastic changes in the gastric mucosa and stratify the associated risk of progression to GC[23].

A recent prospective cohort analysis by Benites-Goñi et al[22] evaluated eight studies (n = 12526) and found a strong association between higher OLGA (III-IV) and OLGIM (III-IV) stages and increased GC risk, with risk ratios of 32.31 and 12.38, respectively, and absolute risk increases of 4%-5%. The findings remained robust in high-incidence regions and after excluding autoimmune gastritis cases. Even intermediate stages (OLGA II and OLGIM II) were associated with elevated risks of high-grade dysplasia and GC, supporting the utility of these systems in GC surveillance planning. Botezatu et al[40] assessed OLGA and OLGIM staging in 142 Moldovan patients undergoing endoscopy. Patients were grouped by presence of IM or dysplasia. A moderate, significant correlation was observed between chronic AG types and OLGA stages, with a weaker but still significant correlation with OLGIM stages. As OLGA and OLGIM stages increased, pepsinogen I levels and the pepsinogen I/II ratio decreased, reinforcing their value as diagnostic tools in this population[40]. Similarly, Yue et al[23] conducted a meta-analysis of eight studies (n = 2700) and confirmed that OLGA/OLGIM stages III-IV are significantly associated with increased GC risk. These findings validate the role of OLGA and OLGIM systems in identifying high-risk individuals who require intensified surveillance.

THE ROLE OF H. PYLORI ERADICATION IN GC PREVENTION

H. pylori infection is a known key player in the long-term progression of chronic gastritis to AG, which is often accompanied by GIM, dysplasia, and ultimately GC[41]. Several mechanisms through which H. pylori infection can lead to metaplasia and dysplasia are described, primarily via the activation of immune cells and release of pro-inflammatory cytokines leading to the development of a chronic inflammatory state. This long-lasting inflammation promotes DNA methylation and reactive oxygen species production that contribute to cellular damage[26].

Early eradication of H. pylori at the stage of AG or GIM may prevent precancerous lesions from acquiring molecular alterations that may be described as “point of no return”. Current surveillance practices of patients with precancerous lesions fail to adequately reflect actual cancer risk, reinforcing the need for improved guidelines. Additionally, population screening for H. pylori through serology, endoscopy, or molecular marker, appears more cost-effective than surveillance of established precancerous lesions[41].

Biological rationale for eradication and the concept of a “window of opportunity”

The eradication of H. pylori has emerged as a pivotal strategy for GC prevention, aiming to disrupt the sequential pathological cascade from chronic gastritis through AG, IM, dysplasia, and ultimately adenocarcinoma[42,43]. Mechanistically, eradication interrupts chronic inflammatory signaling, curtails oxidative DNA damage, and mitigates the accumulation of oncogenic epigenetic modifications such as aberrant methylation[44]. However, the efficacy of intervention is contingent on timing: Once extensive, incomplete-type IM or dysplasia is established, the mucosa acquires irreversible genetic and epigenetic changes, limiting the preventive effect of bacterial clearance[44].

A conceptual model of gastric carcinogenesis highlights critical intervention points. Early H. pylori eradication during chronic non-AG can dramatically reduce cancer risk[26,44]. Eradication at the atrophic stage may halt progression but rarely induces complete histological recovery, while intervention post-IM or dysplasia, although still beneficial, becomes progressively less impactful[43]. Thus, the concept of a “window of opportunity”, a finite period during which mucosal reversibility remains biologically plausible, underscores the urgency of early detection and treatment.

Clinical evidence for eradication efficacy and importance of timing

In a pivotal randomized controlled trial, Yan et al[45] demonstrated a 43% reduction in GC incidence over 26.5 years, with maximal benefit confined to individuals lacking baseline IM or dysplasia. The refined meta-analysis by Wu et al[42] crucially advanced the field by confirming that H. pylori eradication significantly reduces the risk of developing GC, with the greatest benefit observed in individuals who have previously undergone endoscopic resection for gastric neoplasia and a more modest but still significant benefit in otherwise healthy adults. Differences among these studies arise from baseline histologic variability, eradication timing, regional GC incidence, and eradication protocol standardization.

Geographic disparities further complicate interpretation. Most large trials derive from East Asia, where GC incidence is high and screening infrastructures are robust. Although Lee et al[46] demonstrated a greater than 25% reduction in GC incidence after mass eradication in Taiwan, which is rare in Western populations, raising critical questions about global applicability.

Eradication in secondary prevention: Post-endoscopic resection

Following endoscopic resection of early gastric neoplasia, residual mucosal fields often retain molecular alterations that predispose to metachronous cancers. H. pylori eradication in this context constitutes a powerful secondary preventive strategy. A randomized trial by Chey et al[47] confirmed that eradication halved the risk of metachronous cancers, while Kim et al[48] refined these findings by emphasizing the criticality of timing: Eradication performed within one year post-resection conferred superior protection compared to delayed intervention[47,49]. A meta-analysis by Khan et al[50] further corroborated these observations, demonstrating that eradication significantly reduced the incidence of metachronous lesions and preneoplastic transformations after initial endoscopic therapy. Importantly, most evidence again derives from East Asia, where GC risk and mucosal background differ from Western populations, underscoring a pressing need for validation in diverse cohorts.

Limitations of eradication as a standalone strategy

While H. pylori eradication remains foundational, its limitations are increasingly recognized (Table 2). Eradication becomes less effective once extensive IM or dysplasia dominates the mucosa, particularly when incomplete-type IM is present, as these molecular alterations rarely regress even after bacterial clearance[50]. Incomplete IM has been associated with persistent genomic instability and field cancerization, reinforcing the need for intensified surveillance strategies beyond bacterial eradication alone.

Table 2 Comparative analysis of key studies on Helicobacter pylori eradication and gastric cancer prevention.

Ref.	Design	Population	Main findings	Strengths	Critique
Yan et al[45]	Randomized controlled trial	East Asian, high-risk cohort; 26.5-year follow-up	43% reduction in GC incidence; greatest benefit in early mucosal stages	Long-term follow-up; stratified analysis based on baseline histology	Single geographic region; changes in healthcare practices over decades may confound findings
Wu et al[42]	Updated meta-analysis	International with timing-specific focus	Eradication significantly reduces the risk of developing GC, with the greatest benefit observed in individuals who have previously undergone endoscopic resection for gastric neoplasia	Clarified time-dependence of eradication benefit; detailed sub-analyses	Publication bias risk; heterogeneous eradication protocols not standardized
Lee et al[46]	Prospective community intervention	Taiwanese general population	> 25% reduction in GC incidence post-eradication	Real-world mass intervention; extended follow-up	No randomized comparator group; confounded by improvements in screening and healthcare access
Kim et al[49]	Nationwide retrospective cohort	Korean patients post-endoscopic submucosal dissection	Early eradication (< 1 year) decreases metachronous cancer risk	Large sample size; timing stratification	Retrospective nature; surveillance bias; limited generalizability outside East Asia

GC: Gastric cancer.

Antibiotic resistance further complicates eradication success, with clarithromycin resistance exceeding 20% in multiple regions[44], demanding alternative regimens. Recent evidence suggests that H. pylori infection and its eradication significantly alter the composition of the gastrointestinal microbiota, although the long-term consequences of these shifts for gastric carcinogenesis remain to be fully clarified[51]. However, the clinical significance of these microbiome alterations is still incompletely understood and warrants further longitudinal investigation. Furthermore, the feasibility of large-scale eradication programs remains constrained in low-resource settings, limiting global impact[52].

Studies have shown that H. pylori eradication is advised at all stages of gastric mucosal pathology. However, its clinical effectiveness depends largely on the histologic stage at which treatment is initiated. For instance, if eradication is achieved before the development of IM or dysplasia, the associated reduction in future GC risk is more pronounced. On the other hand, in patients with established GIM, a complete histologic regression is rarely achieved. This is even more challenging in advanced cases like those with extensive involvement or incomplete-type metaplasia.

Although eradication may attenuate active inflammatory activity and limit further mucosal metaplasia, a residual risk of malignancy persists. This is attributed mainly to persistent molecular and epigenetic alterations within the gastric mucosa that are not reversible even following bacterial clearance. Therefore, in this clinical setting, eradication should be regarded as a risk-reducing intervention rather than a curative strategy.

Furthermore, a clear distinction must be made between primary prevention and secondary prevention. In the first one, eradication aims to prevent the development of premalignant lesions, whereas in the latter one, eradication is intended to reduce the risk of metachronous GC following endoscopic or surgical treatment of neoplasia. While eradication has been shown to significantly decrease the incidence of metachronous cancer rates after endoscopic resection, its protective effect in patients with untreated, established GIM is comparatively limited and does not eliminate the need for ongoing endoscopic surveillance.

Accordingly, while H. pylori eradication constitutes a fundamental component of GC prevention strategies, it should not be considered sufficient as a standalone preventive strategy in patients with IM. In fact, metaanalysis of 16 randomized controlled trials involving 15027 patients found that H. pylori eradication was associated with a 45% reduction in relative risk of GC compared with no eradication (risk ratio: 0.55; 95% confidence interval: 0.46-0.67) in patients with IM or dysplasia. Therefore, optimal care requires integration of eradication therapy with risk-stratified surveillance protocols based on the histological subtype of metaplasia, its extent, as well as additional patient-specific risk factors[53].

ADVANCES IN DIAGNOSIS AND RISK STRATIFICATION OF GIM

Although AI-assisted endoscopic and histopathologic tools have shown promising diagnostic performance for GIM, it is critical to emphasize that these technologies remain investigational. Most available evidence is derived from retrospective analyses, single-center studies, or highly selected datasets, often from high-incidence regions. Consequently, AI-based systems should currently be viewed as adjunctive decision-support tools under evaluation rather than established components of routine clinical practice. Robust prospective, multicenter validation across diverse populations is required before widespread clinical adoption can be justified.

Optical technologies: Transforming real-time diagnosis

WLE, the traditional tool for gastric surveillance, has limited sensitivity and relies heavily on random biopsies, often missing subtle mucosal changes associated with IM[54]. Narrow-band imaging (NBI), with or without magnification, has significantly improved real-time optical diagnosis by enhancing contrast between mucosal and vascular patterns. Magnifying endoscopy with NBI can detect characteristic markers like the light-blue crest and white opaque substance, which correlate with histologic IM[54].

However, variability among studies highlights the role of operator expertise, equipment quality, and regional mucosal phenotypes. Findings from East Asian high-risk populations may not fully apply to Western settings due to differences in H. pylori prevalence and gastric mucosal backgrounds[54]. The EGGIM system provides a structured, endoscopy-based risk stratification, with scores ≥ 5 aligning with high-risk OLGIM stages[40]. Still, broader validation in diverse populations is needed to confirm the global utility of these optical grading systems.

Importantly, many studies evaluating AI-assisted optical diagnosis of GIM are limited by retrospective design, enrichment with high-quality or representative images, and exclusion of technically challenging cases. Furthermore, training datasets are frequently population-specific, predominantly derived from East Asian cohorts with high GC incidence, which may limit generalizability to Western populations characterized by different mucosal patterns, lower disease prevalence, and variable endoscopist expertise. These limitations underscore the risk of spectrum bias and highlight the need for prospective validation in real-world endoscopy settings before AI-driven optical diagnosis can be reliably integrated into clinical workflows.

AI in enhancing diagnostic accuracy and risk prediction

AI has emerged as a promising investigational tool in gastrointestinal diagnostics; however, current evidence supporting its use in GIM detection and risk stratification remains largely exploratory (Table 2). Importantly, Ligato et al[55] highlighted that AI model performance deteriorates when externally validated across heterogeneous populations. Asian-trained AI systems often underperform when applied to Western mucosal phenotypes, where inflammation patterns and IM distribution differ[55]. This demographic bias underscores the critical need for diversified training datasets.

In histopathology, Iwaya et al[56] demonstrated that AI could detect IM on digitized biopsy slides with > 95% sensitivity, potentially streamlining pathology. However, retrospective validation and reliance on digital platforms remain barriers to real-world clinical deployment. While AI promises consistency, scalability, and objectivity, regulatory hurdles surrounding model explainability, liability in diagnostic errors, and integration into existing clinical workflows remain unresolved, stalling widespread adoption.

Despite encouraging accuracy metrics, the clinical readiness of AI systems for GIM should not be overstated. Most models have been developed and tested retrospectively, often without external validation or with limited validation across heterogeneous clinical environments. Performance frequently declines when algorithms are applied to populations or imaging platforms different from those used during training. Moreover, prospective evidence demonstrating that AI-assisted diagnosis improves patient outcomes, alters surveillance strategies, or reduces GC mortality is currently lacking. Until such data are available, AI should be regarded as a complementary tool to expert endoscopic and histopathologic assessment rather than a replacement.

Future research priorities include prospective, multicenter trials with standardized imaging protocols, predefined clinical endpoints, and diverse patient populations. Such studies are essential to determine not only diagnostic accuracy but also clinical utility, cost-effectiveness, and impact on surveillance decision-making before AI-based tools can be incorporated into guideline-driven practice.

CHALLENGES AND CONTROVERSIES

The 2020 AGA guidelines on GIM have introduced new perspectives in clinical practice. These guidelines have encouraged a more personalized approach to managing this premalignant condition[57]. Despite their potential to improve patient care, significant challenges and controversies remain. Regional inconsistencies, uncertainty around follow-up initiation and discontinuation, cost and adherence issues, and medico-legal implications are the main challenges that clinicians face when navigating these new treatment landscapes[43,57]. One of the most significant hurdles in the management of GIM is the inconsistent adoption of guidelines across regions. Despite the publication of the 2020 AGA guidelines, many healthcare institutions continue to follow older protocols for GIM management, leading to regional discrepancies in how patients are treated. This issue is not just theoretical - research indicates that patient outcomes can vary depending on which guidelines healthcare providers follow[57]. Another challenge for both conditions is the lack of clarity around when to start or stop treatment follow-up. With GIM, the 2020 AGA guidelines suggest individualized surveillance based on a patient’s specific risk factors, but this flexibility leads to wide variations in practice. For example, some gastroenterologists may begin surveillance for a patient with early-stage GIM, while others may choose to delay it. This issue becomes even more complicated when determining when to discontinue surveillance, especially for low-risk patients[57]. Cost-effectiveness is also a major concern. Surveillance for GIM, although recommended for high-risk individuals, comes with a financial burden for both patients and healthcare systems[57]. Lastly, there are medico-legal and health policy issues that complicate the integration of these guidelines into everyday practice. Clinicians are often hesitant to adopt nontraditional surveillance and follow-up strategies when it comes to GIM, because if surveillance is missed or delayed, there is the risk of legal repercussions if a patient’s condition progresses to cancer[43,57].

A persistent area of uncertainty concerns the management of patients with low-risk GIM. Current evidence does not support routine surveillance in individuals with complete, antrum-limited IM and no additional risk factors, yet clinical practice often exceeds guideline recommendations due to medico-legal concerns and patient anxiety. In such cases, shared decision-making is essential, with transparent discussion of uncertain cancer risk, lack of proven survival benefit from surveillance, potential procedural harms, and healthcare costs. Explicit documentation of these discussions may help reconcile guideline flexibility with real-world clinical pressures.

FUTURE DIRECTIONS

The discovery of new biomarkers is valuable for developing molecular techniques to improve early detection and monitoring of diseases. Recent research has identified several biomarkers and molecular pathways involved in the progression from IM to GC. Key findings include the involvement of extracellular matrix-receptor interaction genes (like COL1A1, COL1A2, FN1, and THBS2) in GC, while peroxisome proliferator-activated receptor signaling-related genes (such as FABP1, APOC3, and APOA1) play a role in IM. A transcription factor-hub gene network highlighted androgen receptor, transcription factor 4, spalt-like transcription factor 4, and estrogen receptor 1 as crucial regulators. Additionally, hsa-miR-29 and other microRNAs were implicated in GC development. Some genes (PTGR1, ALDOB, and SULT1B1) were downregulated in GC but upregulated in IM, suggesting potential tumor-suppressive roles. Survival analysis showed that most extracellular matrix genes predicted poor outcomes in GC, while integrin subunit beta 1 was associated with a better prognosis. These insights could guide future targeted therapies and prognostic tools for GC[58].

A study investigating biomarkers involved in the progression from GIM to GA identified several promising candidates, including gene expression profiles (retinol binding protein 2, CD44), key signaling pathways, and molecular interaction networks. Potential therapeutic targets for drug repurposing such as epidermal growth factor receptor, Src kinase, paxillin, Jun, breast cancer type 1 susceptibility protein, tumor protein p53, mouse double minute 2 homolog, and CD44 were also highlighted. These findings provide valuable insights for risk stratification and chemoprevention strategies in patients with GIM at risk of developing GA[59].

Advances in endoscopic imaging have enhanced detection of preneoplastic gastric lesions, overcoming limitations of WLE. NBI improves mucosal visualization with higher sensitivity but lower specificity. Chromoendoscopy highlights subtle changes and defines lesion margins, aiding in early GC staging. Probe-based confocal laser endomicroscopy provides real-time, high-resolution imaging for accurate diagnosis and targeted treatment, reducing the need for multiple biopsies[26].

AI, particularly foundation models (FMs), is transforming medical imaging for upper gastrointestinal cancers by offering scalable, automated image analysis. Trained on large, diverse datasets, FMs can be fine-tuned for tasks like interpreting endoscopy and pathology images, improving diagnostic accuracy and efficiency. The integration of multi-modal data combining imaging, genomics, and clinical records, promises more personalized diagnoses and treatment plans. Additionally, FMs could automate medical report generation, streamlining workflows and enhancing clinical decision-making[60].

A study aimed to identify factors that increase the risk of developing GC in patients who were followed for over three years after H. pylori eradication. By comparing endoscopic, histological, and biochemical findings, the researchers found that patients who later developed GC were typically older at the time of eradication, had lower serum pepsinogen I/II ratios, more severe endoscopic mucosal atrophy, greater inflammation and IM in the gastric corpus, and higher activity in the antrum. Therefore, severe endoscopic atrophy and a low serum pepsinogen I/II ratio at any follow-up point should prompt regular endoscopic surveillance in post-H. pylori eradication patients to monitor for potential GC development[61].

A study identified male sex, age ≥ 60, smoking, and low vegetable intake as key risk factors for GC. Using these, patients with non-neoplastic gastric lesions were stratified into risk groups. High-risk patients (with ≥ 3 factors) had a four-year cumulative GC incidence comparable to the one-year rate in patients with neoplastic lesions. The findings support integrating individualized risk stratification into clinical guidelines, recommending closer endoscopic surveillance (every 3-5 years) for high-risk individuals with mild precursor lesions[62].

A study of H. pylori-negative patients with gastric precancerous lesions found no significant differences in overall microbiota diversity but identified important taxonomic shifts. Lautropia mirabilis was enriched, while Limosilactobacillus reuteri, Solobacterium moorei, Haemophilus haemolyticus, and Duncaniella dubosii were depleted in both IM and dysplasia compared to chronic gastritis. Prevotella jejuni and Parvimonas were enriched in IM. These microbiota changes suggest a shift toward a pro-inflammatory environment that may promote lesion progression, highlighting potential microbial targets for prevention strategies[1].

Genetic polymorphisms related to inflammatory cytokines are often associated with susceptibility to gastric disease owing to differences in the expression levels that interfere with chemotaxis and amplify the immune response. A study found that in male patients, those carrying the interleukin 1B-511 CC genotype had a significantly higher prevalence of moderate to severe AG in the gastric corpus than CT or TT carriers, independent of age, alcohol use, and H. pylori virulence factors. Among females, the interleukin-8-251 AA genotype was independently linked to a higher risk of moderate/severe corpus AG compared to AT or TT genotypes (21.4% vs 6.0%; adjusted hazard ratio: 3.799). Additionally, in males, the interleukin-8-251 TT genotype was associated with a greater risk of moderate/severe IM in the corpus, while the interleukin-10-592 CA and CC genotypes correlated with more severe monocyte infiltration in the antrum[48].

CONCLUSION

GIM represents a pivotal step in the GC progression pathway, linking CG to malignancy. Despite significant advances in diagnostic technologies, biomarker discovery, and international guideline development, challenges in the standardized management and surveillance of GIM persist. Variations between the AGA and MAPS III guidelines highlight the need for a unified, risk-based global approach tailored to regional cancer incidence and healthcare resources. Early detection through high-definition endoscopy, the eradication of H. pylori, and precise risk stratification using systems like OLGA and OLGIM have emerged as crucial strategies. While AI-based technologies offer substantial promise for enhancing GIM detection and risk stratification, they currently remain adjunctive tools under investigation and should not yet be considered standard of care pending prospective, multicenter validation demonstrating clinical benefit. The future of GIM management lies in the integration of multimodal risk assessment, combining endoscopic, histological, molecular, and AI-driven data, to personalize surveillance and optimize outcomes. Addressing medico-legal, adherence, and cost-effectiveness barriers remains essential to translating these innovations into clinical practice. Ultimately, advancing global collaboration, refining follow-up protocols, and embracing precision medicine approaches will be key to mitigating the burden of GC through effective prevention and early intervention at the stage of GIM. Importantly, international guidelines for GIM should be interpreted as flexible frameworks rather than prescriptive rules, with surveillance decisions guided by individualized risk assessment and shared decision-making, particularly in Western, low-incidence populations.