Helicobacter pylori is an important cause of non-cardia gastric cancer and is the most prevalent carcinogenic infectious agent, worldwide, especially in East Asia [1,2]. The development of intestinal-type gastric cancer has long been hypothesized to involve a stepwise transformation from normal gastric epithelium to malignancy. During this transformation, intestinal epithelial properties are acquired at an intermediate stage. This stepwise progression from a normal gastric epithelium to cancer (malignant tumor) is called the Correa pathway; chronic infection of the gastric mucosa leads to this stepwise progression from atrophic gastritis to intestinal metaplasia [3]. Chronic gastritis is inevitable during the early stages of the Correa pathway, resulting in atrophy of the epithelial cells comprising the normal gastric fundic glands, including the parietal cells. Under these conditions, cells with intestinal characteristics can emerge, leading to intestinal metaplasia. Most H. pylori-infected individuals H. pylori remain asymptomatic. Gastric cancer caused by H. pylori infection is dependent upon bacterial virulence, host genetic polymorphisms, and environmental factors. Most H. pylori strains possess a cytotoxin-associated gene A (cagA) pathogenicity island encoding the oncogenic CagA protein, which affects the expression of cellular signaling proteins through the type IV secretion system [4]. However, we found that when CagA was injected into cells via the type-IV secretion system, it was usually underwent autophagy. However, in CD44v9-positive cells, which possess the properties of cancer stem cells, autophagic degradation is suppressed owing to oxidative stress resistance and is the mechanism by which CagA accumulates, leading to gastric carcinogenesis [5]. Furthermore, we showed that when H. pylori infection-associated gastritis progresses and CAPZA1-overexpressing cells develop atrophic gastritis [6], the expression of CD44v9 is also induced. Cancer stem-like cells are generated when CagA is injected into CAPZA1-positive cells [5]. This means that once gastritis progresses to a certain stage, beyond the point of generation of CAPZA1- and CD44v9-positive cells, the risk of gastric cancer does not disappear, even after bacterial eradication.

Following endoscopic submucosal dissection (ESD) for early gastric cancer (EGC), the gastric mucosa, which has the highest risk for gastric cancer development, retains a high risk of metachronous recurrence. However, bacterial eradication can reduce this risk by approximately one-third [7,8]. Regardless, even after H. pylori eradication, some risk remains. This is because, at the time of H. pylori eradication, the precancerous lesions may have adequately progressed toward a cancerous phenotype. In one study, we enrolled 88 patients who underwent ESD for EGC between 2008 and 2010 and observed recurrence over three years. The expression level of CD44v9 in primary EGC tissues was evaluated using immunohistochemistry, and recurrence rates were compared between the CD44v9-positive and -negative patient groups. The recurrence rate of EGC was significantly higher in the CD44v9-positive group than that in the CD44v9-negative group (hazard ratio [HR], 21.8; 95% confidence interval [CI], 5.71–83.1). The degree of gastric atrophy was also identified as a significant marker of multiple recurrences (HR, 4.95; 95% CI, 1.30–18.8). Therefore, CD44v9 expression may determine EGC recurrence [9].

These observations imply that H. pylori infections should be eradicated as early as possible, before gastric preneoplastic lesion progression [10]. The infection should also be eradicated before precancerous changes, such as the appearance of CD44v9-positive cells, occur [11]; endoscopic surveillance remains necessary, even after bacterial eradication.

Based on the global distribution of gastric cancer incidence rates, the Asian side of the Eurasian continent has an extremely high incidence of gastric cancer. This includes the regions of East Asia (China and South Korea), including Japan; Southeast Asia (Vietnam, Laos, Cambodia, and Myanmar); Central Asia; Türkiye; and the three Baltic states (Estonia, Latvia, and Lithuania). These countries are among the top 25 countries in terms of age-standardized gastric cancer incidence and mortality rates. Japanese men have the world’s highest incidence of gastric cancer (48.1 cases/100000), and Japanese women rank fourth (17.3 cases/100000) [12]. In South Korea, the incidence of gastric cancer among men is 39.7 cases/100000 (third highest rate, worldwide), and among women it is 17.6 cases/100000 (also third highest, worldwide). Although the incidences of gastric cancer in Japan and South Korea are high, the associated mortality rates are low (Japan: 12.4 for men, 4.8 for women; South Korea: 8.8 for men, 4.2 for women) [12]. Other countries with high rates of gastric cancer also have high mortality rates (Mongolia: gastric cancer incidence: 47.2 for men [second highest, worldwide], 20.7 for women [highest, worldwide]; gastric cancer mortality rate: 36.5 for men, 15.2 for women) [12]. Japan and South Korea are the only countries with diverging incidence and mortality rates because they have implemented effective gastric cancer nation-wide screening systems. Thus, patients with gastric cancer are diagnosed at the early stage. Moreover, the systems for diagnosing gastric cancer risk and early-stage gastric cancer in these countries are believed to be more developed than those in other countries. Before the launch of the South Korean screening system, the total number of new gastric cancers in 1999 was 20900 (20.5% of the total cancers), and, prior to 2000, the 5-year survival rate was less than 50% (43.9% in 1993–1995 and 47.3% in 1996–2000) [12].

To reduce gastric cancer mortality and enhance early-stage detection, the Japanese government enacted the 1983 Health Care for the Elderly Law and began providing annual barium X-ray screenings for gastric cancer to adults aged 40 and over. In 2007, as part of the Revision of the Guidelines for Cancer Prevention-Focused Health Education and Cancer Screening Implementation, the Japanese Ministry of Health, Labor, and Welfare also recommended annual barium X-ray contrast fluorography for people aged 40 and over. Unfortunately, in 2008, the number of eligible individuals receiving this preventive screening test was only 10.2%. Following the documentation that endoscopic gastric cancer screening was effective in reducing gastric cancer deaths, the 2015 Amendments to the Guidelines for Cancer Prevention-Focused Health Education and Cancer Screening Implementation recommended that adults aged 50 years and older should choose between barium X-ray contrast fluorography and upper gastrointestinal endoscopy every 2 years. To date, endoscopic gastric cancer screening has been performed for over 20 years, usually during voluntary health checkups and medical examinations.

Nationwide gastric cancer screening has been performed in Japan and South Korea for several decades. Recently, Sun et al. [13] evaluated the impact of these screening programs on gastric cancer mortality. Briefly, they used a flexible synthetic control method to estimate the effect of a screening program on age-standardized gastric cancer mortality in individuals aged 40 years or older. The analysis used mortality data from the World Health Organization and country-level covariates from the World Bank and the Global Burden of Disease Study. They compared outcome trends after the intervention with the counterfactual trends of synthetic controls and estimated average post-intervention rate ratios (RRs) and associated 95% CIs. In South Korea, they found that the average post-intervention RR was 0.83 (95% CI, 0.71–0.96) for gastric cancer mortality; 15 years after the start of nationwide screening, the RR reached 0.59. In Japan, the average RR for gastric cancer mortality was 0.97 (95% CI, 0.88–1.07), although sensitivity analyses revealed a potential bias in the Japanese results [13]. Hence, although there was a clear benefit to nationwide gastric cancer screening in South Korea, the effectiveness of Japanese screening program was unclear [13].

Although gastric cancer mortality rates are decreasing in Japan, an analysis comparing the results with those from a synthetic control group failed to demonstrate the effectiveness of gastric cancer screening in Japan in reducing gastric cancer mortality rates. One possible reason for this discrepancy is the differences in gastric cancer screening methods between Japan and South Korea. After enacting the 1983 Health Care Act for the Elderly, only barium X-ray contrast fluorography has been recommended for preventive examination method of gastric cancer, in Japan. However, the number of eligible individuals who underwent this preventive examination in 2008 was only 10.2% (3916203 individuals). Following the revision of the Guidelines for Cancer Prevention-Focused Health Education and Cancer Screening in 2015, upper gastrointestinal endoscopy (esophagogastroduodenoscopy [EGD]) has been recommended for gastric cancer screening. However, as of 2015, only 19% of local governments in Japan had introduced endoscopic examinations. In South Korea, as of 2011, 72.55% of the people screened chose endoscopic examinations over barium X-ray examinations [13,14].

The primary gastric cancer prevention strategy focuses on the diagnosis and eradication of H. pylori infection. However, in Japan and South Korea, where the incidences of gastric cancer are high, endoscopic or barium X-ray contrast fluorography for identifying asymptomatic individuals were adopted as a secondary prevention strategy before the era of H. pylori eradication [14]. In contrast, in countries with low gastric cancer incidences, the identification of high-risk patients with precancerous lesions (e.g., intestinal metaplasia) through targeted surveillance is accepted [14]. Since 2022, Japanese preventive health checks (barium X-ray contrast fluorography or upper gastrointestinal endoscopy) have been available to citizens aged 50 years and over (with no upper age limit) once every 2 years. However, South Korea conducts endoscopic examinations every 2 years for citizens aged 40–74 years and strongly recommends endoscopy over barium X-ray contrast fluorography. In South Korea, there is an upper age limit owing to the absence of sufficient evidence regarding the effectiveness of screening for people aged 75–84 years. South Korea has also clearly indicated that screening will not be conducted on people aged 85 and over [14]. Thus, the recent differences in gastric cancer mortality rates between Japan and South Korea, as reported by Sun et al. [13], are likely related to differences in EGC detection rates that have resulted from differences in screening methods.

EGD (upper gastrointestinal endoscopy) is a highly effective gastric cancer screening test because it facilitates the early diagnosis and detection of gastric cancer and indirectly diagnoses H. pylori infection and H. pylori-infected gastritis, which are responsible for most gastric cancer developments. Thus, this screening method provides for secondary prevention of gastric cancer through early diagnosis and leads to H. pylori eradication, which is a component of primary prevention.

In addition to diagnosing gastric cancer, especially EGC, background gastritis (a risk factor for gastric cancer) should be observed and diagnosed during upper gastrointestinal endoscopy. The Kyoto classification of gastritis is useful for endoscopically assessing gastric cancer risk [15]. Recently, we conducted a study focused on endoscopic findings [15] adopted in the Kyoto score of the Kyoto classification of gastritis.

The Kyoto classification of gastritis [15] was created by Professor Ken Haruma in Kawasaki Medical University to evaluate H. pylori infections and gastric cancer risk. He organized the endoscopic findings that could be objectively and easily determined and integrated them with tissue diagnoses, especially the updated Sydney System (USS) [16]. Relative to H. pylori infection, patients are classified as “non-infected,” “currently infected,” or “previously infected.” The Kyoto classification uses atrophy, intestinal metaplasia, enlarged folds, and nodular gastritis as endoscopic findings indicative of the risk of gastric cancer (Fig. 1). It also adopts diffuse redness as a finding that distinguishes between current infection and post-eradication, thus accounting for the effect of eradication on suppressing gastric cancer. Thus, to create the Kyoto score, the scores determined for these two aspects observed during the endoscopy are added together [15].

According to the Kimura–Takemoto classification, gastric mucosal atrophy is represented by a discolored area with low mucosal height that is graded as C1, C2, C3, O1, O2, or O3, depending on the extent of atrophy (oral-side shift of endoscopic atrophic border). Under white light, intestinal metaplasia is observed as a gray-white, raised spot. Map-like redness is a patchy, slightly depressed redness that is thought to be a manifestation of intestinal metaplasia caused by changes in the gastric environment after H. pylori eradication. The presence of enlarged folds indicates that the folds of the greater curvature of the corpus were not flattened by air insufflation and are 5 mm or thicker. Nodular gastritis is characterized by a small, dense, and uniform granular protrusion centered around the pyloric antrum, similar to those in goose flesh. Diffuse redness is a continuous, uniform redness that spreads, mainly in the mucosa of the greater curvature of the corpus, without evidence of atrophy. The regular arrangement of collecting venules (RAC) is an endoscopic image in the mucosa of the gastric corpus. In microcirculatory terms, collecting venules are defined as venules that collect blood flow from the capillaries further collect blood in the deep layer of the mucosa before flowing into the small, submucosal veins. This clear visibility is the typical image of a normal gastric mucosa that is not infected with H. pylori.

Recently, a succession of reports has revealed that image enhancement technologies, such as linked-color imaging, narrow band imaging, texture and colour enhancement imaging, dyes, and artificial intelligence (AI), are highly useful in diagnosing gastric tumors and H. pylori-infected gastritis. Among these, the emergence of endoscopic diagnoses using AI is interesting because it has the potential to significantly change the future direction of screening. The AI-based endoscopic diagnostic system will coexist with endoscopists’ endoscopic gastric cancer screening, and contribute to improving the skills of endoscopists by detecting risk lesions, etc. However, the introduction of such advanced technology has the potential to improve the accuracy of endoscopic diagnosis in primary care. Although mucosa at high risk of developing gastric cancer can be evaluated using semi-quantitative scoring systems, such as the Kyoto classification of gastritis, the overall image is extremely heterogeneous. Thus, this is an area where the validity of each diagnosis is likely to vary depending on situational factors that can be interpreted by the endoscopist. However, an AI-enhanced endoscopic diagnostic system that applies deep learning may solve the problem of subjective interpretations. AI has been improving owing to the development of deep learning methods, availability of big data, and advances in computer processing devices. In the medical field, AI has been used for image recognition, primarily in radiological and pathological diagnoses. In the field of gastrointestinal endoscopy, AI-based computer-aided detection/diagnosis (CAD) systems have been applied in certain areas, such as for the detection and diagnosis of colonic polyps. However, to date, their implementation in actual clinical settings has been limited. Accurate detection or diagnosis of gastric cancer is challenging, and its performance varies greatly depending on the skill of each endoscopist [30]. Early diagnosis of gastric cancer is particularly difficult owing to the heterogeneity of the images acquired during screenings. Although CAD systems for gastric cancer are being actively developed, their use for diagnosing gastric cancer remains difficult because they require a large number of images. Furthermore, the training image data must be of a sufficiently high quality for proper CAD training [30]. Recently, AI systems for upper gastrointestinal endoscopy have been developed and trained using large numbers of high-quality images; therefore, these types of systems may be more reliable.

The introduction of AI could reduce some notable shortcomings associated with current screening procedures, such as the presence of gastric cancer in the absence of obvious symptoms, difficulty in identifying lesions by assessing images using only the white light, and the lack of experienced endoscopists. Thus, AI may be able to significantly improve the accuracy of screening. AI-assisted endoscopy has great potential as a more accurate and sensitive method for the early detection, differentiation, and prediction of gastric cancer depth of invasion [31]. The sensitivity and accuracy of the neural network-trained AI system is said to be comparable to that of non-expert endoscopists. Further, they can also shorten the time required for image evaluation, thereby significantly reducing the burden on endoscopists [32]. In addition, AI-enhanced systems may be able to detect and provide feedback on endoscopists’ examination processes in real-time, thus standardizing their operations [32]. AI may also have great potential for training novice endoscopists and could be useful for introducing endoscopic screening systems into countries where such screening systems have not yet been established [32]. This technology could be of great help in detecting early-stage gastric cancer [33]. The anticipated superior diagnostic capabilities of AI-assisted technologies will be particularly beneficial for detecting early-stage gastrointestinal cancer, an area in which the diagnostic ability and accuracy of endoscopists varies [33].

Prior to 2013, the annual number of deaths due to gastric cancer, in Japan, was approximately 50000. However, since 2013, when eradication therapy for H. pylori gastritis was approved for coverage by health insurance companies, the number of deaths from gastric cancer have declined. In 2015 the number of deaths was 46681 (30810 men and 15871 women); in 2020, the total was 42319 (27771 men and 14548 women) and, in 2023, the total was 38767 (25323 men and 13444 women) [34]. Thus, 2023 marked the first time that the annual number of deaths due to gastric cancer dropped below 40000, marking a 20.3% decrease in the number of deaths within 10 years.

Eradicating H. pylori does not immediately affect mortality rates. However, societal interest has now shifted to gastric cancer prevention, as indicated by more people undergoing gastric cancer screening. To achieve this, the quality of gastric cancer screening must be improved. The application of AI to upper gastrointestinal endoscopy may improve the ability to diagnose early-stage gastric cancer or mucosa at risk of developing gastric cancer. This technology may also enable endoscopists to better identify patients who are candidates for endoscopic resection, ultimately leading to H. pylori eradication [35]. However, the benefits of AI on the upper gastrointestinal endoscopy have not yet been fully demonstrated. Thus, prospective trials are required to confirm the accuracy and feasibility of real-time routine endoscopy. In any case, endoscopists should not exist independently of AI but should coexist with and prosper because of these technological advances. By combining the expertise of endoscopists with AI, AI systems will become smarter.