Until 2018, gastric cancer was the most common cancer in Korea. In 1999, the total number of new gastric cancer cases was 20900 (20.5% of total cancers), and the 5-year survival rates were less than 50% before 2000 (43.9% in 1993–1995; 47.3% in 1996–2000) [1,2]. To detect gastric cancer at an earlier stage and reduce its mortality, the Korean government implemented the National Cancer Screening Program for Gastric Cancer (NCSP-GC), in 1999, initially for medical aid beneficiaries; in 2002, the program was expanded to all adults aged 40 years or older [3]. Each NCSP-GC participant underwent either endoscopy or an upper gastrointestinal series (UGIS) every two years until 2017. However, in 2015, the Korean Guideline for Gastric Cancer Screening no longer recommended UGIS as a primary screening modality for the NCSP-GC [4]. Thus, endoscopy became the only primary screening modality in 2018; an UGIS could be selected if a participant was unable to undergo endoscopy [3]. The number of endoscopic screenings performed as part of the NCSP-GC increased from approximately 0.22 million in 2002 to over 7.84 million in 2022 (Fig. 1) [5]. In this review, the effectiveness, performance, and limitations of the endoscopic screening component of the NCSP-GC are reviewed, and the strategies for overcoming limitations are discussed.

Mortality due to gastric cancer has decreased markedly since 1999. The age-standardized mortality rate decreased from 23.9/100000 persons in 1999 to 5.9/100000 persons in 2021 [2]. In addition, for the 2017-2021 period, the 5-year survival rate was 77.9%, a marked increase of 34% compared with that in 1993– 1995 due to the NCSP-GC [1]. A meta-analysis of case-control and cohort studies from Korea, Japan, and China reported that endoscopic screening reduced gastric cancer mortality (risk ratio for gastric cancer mortality, 0.60; 95% confidence interval [CI], 0.49–0.73) [6]. Two studies have reported on the effectiveness of endoscopic screening for decreasing gastric cancer mortality in Korea [7,8]. In a nested case-control study using the NCSP-GC database, ever-screened individuals had a 21% reduction in the risk of mortality from gastric cancer (odds ratio [OR], 0.79; 95% CI, 0.77–0.81) compared with never-screened individuals [7]. The mortality reduction was shown for patients who received endoscopy (OR, 0.53; 95% CI, 0.51–0.56), but not for those who received UGIS (OR, 0.98; 95% CI, 0.95–1.01) [7]. Similar results were reported in a community-based prospective cohort study wherein patients who participated in endoscopic screening demonstrated a significant gastric cancer mortality risk reduction of 42% (hazard ratio, 0.58; 95% CI, 0.36–0.94) [8]. Unlike the data from Japanese studies that reported gastric cancer mortality reduction following radiographic screening using UGIS (approximately 50% gastric cancer mortality reduction) [9,10], a similar benefit was not shown in Korean studies [7,8]. In 2015, the Korean guideline no longer recommended UGIS as a primary modality for the NCSP-GC, but recommended this modality only for participants who were unable to undergo endoscopy [4].

In 2021, gastric cancer was diagnosed at the localized (66.0%) and regional (18.6%) stages [1] that could be treated curatively using endoscopic resection, gastrectomy, or gastrectomy with adjuvant chemotherapy. With the increased numbers of endoscopic screenings, the proportion of patients diagnosed with localized gastric cancer increased from 54.6% in 2009 to 66.0% in 2021 (Fig. 2). In a study using the 2002–2007 NCSP-GC database, the ORs for gastric cancer detection at the localized stage were higher in patients who received endoscopic screening (OR, 2.10; 95% CI, 1.90–2.33) than in those who received UGIS screening (OR, 1.24; 95% CI, 1.13–1.36) when compared with never-screened patients [11].

The NCSP-GC, especially the endoscopic screening component, has contributed to a reduction in gastric cancer-related mortality since its implementation, in 2002, for Korean adults aged ≥40 years. In addition, approximately two-thirds of patients with gastric cancer are now diagnosed at the localized stage owing to endoscopic screening and can be managed with less invasive and curative treatments, including endoscopic resection or laparoscopic gastrectomy.

In 2022, the NCSP-GC participation rate was 63.8%, with 92.5% of the participants undergoing endoscopic screening (number of endoscopic screenings, 7840437) (Fig. 1). Gastritis was the most common endoscopic diagnosis between 2013 and 2022, accounting for more than 85% of all diagnoses (Table 1). Despite an increase in the total number of endoscopies, the number of endoscopic diagnoses of suspected gastric cancer, early gastric cancer, and advanced gastric cancer has decreased from 10136 (0.21%) in 2013 to 7389 (0.09%) in 2022. The number of biopsies performed as part of the NCSP-GC has also increased with the increase in endoscopic screenings (Table 2). The detection rate and numbers of pathologically diagnosed gastric cancer cases have not increased (0.83%, 9861 cases/1184477 biopsy cases in 2013; 0.44%, 10664 cases/2421186 biopsy cases in 2022). Over 75% of the biopsy cases were normal or pathologically diagnosed as gastritis.

Upon endoscopic screening, approximately 95% of participants had abnormal findings, most of which were gastritis. Considering the approximately 30000 new gastric cancer cases every year, approximately one-third of the annual total number of these cancers might be confirmed histologically following endoscopic screening.

Despite the benefits of endoscopic screening for early gastric cancer detection and mortality reduction, the proportion of participants with abnormal findings observed during endoscopic screening have been approximately 95%, much higher than those reported for NCSP screenings for colon, breast, cervix, and lung cancers (Supplementary Table 1 in the onlineonly Data Supplement). Most abnormal findings identified during endoscopic screening are gastritis (>85%); less than 5% of screenings are normal (Table 1). A multicenter study involving 40 Korean health check-up centers reported that normal endoscopic findings were observed in 14% of 25536 patients [12]. In this study, among individuals diagnosed with endoscopic gastritis, the distribution of endoscopic gastritis was 35.1% superficial gastritis, 26.6% erosive gastritis, 30.4% atrophic gastritis, and 8.0% intestinal metaplasia [12]. Considering the proportion of erosive, atrophic gastritis, or intestinal metaplasia findings (65%), a considerable number of participants who undergo endoscopic screening might visit secondary or tertiary referral hospitals to seek a second opinion. This might be associated with unnecessary additional endoscopies and the overuse of medical resources.

Participation rates in the program have increased since its initial implementation (from 7.4% in 2002 to 63.8% in 2022) (Fig. 1), largely due to the program’s well-organized endoscopic screening component. However, opportunistic gastric cancer screenings remain an important part of endoscopic screening in Korea. A national survey study conducted on NCSP-GC participants reported that the rates of opportunistic screening for gastric cancer increased from 10.0% to 24.9% between 2009 and 2021, and the combined proportion of NCSP-GC and opportunistic screenings was 75% in 2022 [13]. The number of endoscopic screenings in 2022 was 1.66-fold higher than that in 2013 (about 4.73 million endoscopies in 2013 and 7.84 million in 2022) whereas the number of histologically confirmed gastric cancer cases identified as part of the NCSP-GC in 2022 was similar to the number identified in 2013 (9861 cases in 2013 and 10664 cases in 2022) (Tables 1 and 2). These findings suggest that systematic evaluations of endoscopies performed as part of the NCSP-GC are needed, although many undetected cases might be confirmed as gastric cancer in subsequent evaluations, based on the abnormal findings observed during NCSP-GC endoscopies.

In 2018, the Quality Improvement Committee of the Korean Society for Gastrointestinal Endoscopy announced several quality indicators for endoscopy in the NCSP [14]. The indicators included statements about endoscopist qualifications and about endoscopy processes, facilities, equipment, and outcomes. Whereas there were several specific statements about quality indicators for colonoscopy performance (image recording, bowel preparation, cecal intubation, withdrawal time ≥6 minutes), the only statement regarding performing endoscopies was that at least eight standard images should be recorded [14]. These might be due to the lack of studies investigating endoscopy quality indicators.

A meta-analysis of 22 cohort studies reported that the pooled rate of missed gastric cancer cases was 9.4% [15]. The observation time during an endoscopy is one of the most important factors associated with gastric cancer detection. A retrospective study from Singapore showed that endoscopists who performed endoscopies for more than 7 minutes per procedure detected significantly more biopsy-proven, high-risk gastric lesions including gastric cancer, lymphoma, dysplasia, or intestinal metaplasia (adjusted OR, 2.50; 95% CI, 1.52–4.12) than those who performed endoscopies for less then 7 minutes per procedure [16]. Based on the results of this study, the European guideline recommends at least 7 minutes of observation time, from intubation to extubation, for endoscopy procedures [17]. A Japanese study also reported that endoscopists taking 5–7 minutes of observation time (OR for detection of neoplastic lesions, 1.90; 95% CI, 1.06–3.40) and those spending more than 7 minutes of observation time (OR, 1.89; 95% CI, 0.98–3.64) had higher upper gastrointestinal (UGI) neoplastic lesion detection rates than those spending less than 5 minutes of observation time [18]. A Korean study involving 111962 individuals also showed that longer observation times (≥3 minutes) were associated with higher UGI neoplasm detection rates (adjusted OR for detection of gastric adenoma or cancer, 1.52; 95% CI, 1.17–1.97) [19]. In a Chinese study, observation times of more than 5 minutes were associated with the detection of benign and malignant UGI lesions during endoscopy [20]. Currently, no observation time recommendations are included in the endoscopic screening guideline for the NCSP-GC. However, based on previous studies, at least 3–5 minutes of observation are required to decrease the rate of missed gastric cancer cases.

One of the key performance indicators suggested by the European Society of Gastrointestinal Endoscopy is accurate photo documentation of anatomical landmarks and abnormal findings [21]. The anatomical landmarks included the duodenum, major papilla, antrum, angle, corpus, retroflex of the fundus, diaphragmatic indentation, upper end of gastric folds, squamocolumnar junction, and the distal and proximal esophagus (at least 10 images) [21]. The recording of at least eight clear, standard images is recommended for NCSP-GC endoscopic screening [14]. In a Korean study that included 5970 individuals who underwent endoscopies performed by eight endoscopists, the numbers of endoscopic images were positively associated with observation times, biopsy rates, and detection of clinically significant gastric lesions [22]. The 2019 World Endoscopy Organization position statement recommended taking a set of 28 anatomic site images from the hypopharynx to the duodenum as part of a complete endoscopy [23]. Compared with the recommendations from guidelines and statements of diagnostic endoscopy outside Korea, the performance indicators for NCSPGC screening endoscopies include minimal items and only eight photo documentation images [14]. This might be partly associated with the low cost of endoscopic screenings in Korea. Endoscopies performed according to other standard photo documentation guidance documents, might help ensure sufficient observation times to yield complete examinations without blind spots and detect gastric cancer and other abnormal findings. To improve the quality of the NCSP-GC screening endoscopies, appropriate costs need to be reimbursed. Further studies are needed to evaluate the association between standard photo documentation recommendations and the detection of gastric lesions.

Endoscopies using artificial intelligence (AI) models have been actively investigated, particularly for colonoscopies. A systematic review and meta-analysis of five randomized controlled studies reported that the pooled adenoma detection rate was significantly higher in colonoscopies involving AI models than in the control group (risk ratio for adenoma detection rate, 1.44; 95% CI, 1.27–1.62) [24]. Likewise, endoscopies performed in conjunction with AI models have been evaluated relative to the detection and characterization of UGI lesions, including Barrett’s neoplasia, esophageal cancer, Helicobacter pylori infection, and gastric cancer [25]. The use of AI models has also been tested to determine whether improved endoscopy quality results. A randomized controlled trial showed that endoscopies performed in conjunction with a real-time AI system reduced the blind spot rate (5.86% vs. 22.46%) and increased the completeness of photodocumentation (90.64% vs. 79.14%) compared with conventional endoscopy [26]. Another study reported that endoscopies performed in conjunction with an AI model could detect early gastric cancer with minimal blind spots (accuracy for early gastric cancer detection, 92.5%; accuracy for gastric location recognition, 90%) [27]. Currently, various types of AI models, with different outcomes, have been developed for use during endoscopy. Thus, in the future, AI systems will have a role in helping endoscopists perform high-quality endoscopic screenings; further validation of these AI models is required in real-time endoscopic screening practices.

Endoscopic screenings performed as part of the NCSP-GC have contributed to early gastric cancer detection and reduced gastric cancer mortality. Annually, approximately one-third of the total number of new gastric cancer cases (10000 cases) are histologically confirmed following NCSP-GC endoscopic screening. Participant rates in the NCSP-GC endoscopic screenings have increased, with approximately 7.84 million screening endoscopies performed as part of the program in 2022. However, the standard quality indicators of NCSP-GC endoscopic performance remain insufficient in the NCSP-GC guidelines. To improve the quality of endoscopy performance and to detect more gastric cancers as part of the NCSP-GC, performance indicators (e.g., observation time and photo documentation) need to be included. In the future, endoscopies performed in conjunction with AI models will help endoscopists perform high-quality, standard endoscopic screenings as part of the NCSP-GC.