Helicobacter pylori infection and gastric atrophy are well-known causes of gastric cancer.^1,2 Despite improved eradication and sanitation, H. pylori infection is still common in regions where endoscopic gastric cancer screening has been widely performed without eradication.^3,4H. pylori infection is responsible for 56.5% of infection-related cancer in Korea, followed by hepatitis B virus (23.9%), human papillomavirus (11.3%), and hepatitis C virus (6.0%).⁵ Although H. pylori-seropositivity in urban residents and in younger population is declining in Korea, the prevalence is still high in rural residents and in older population.⁶

Combined serum anti-H. pylori immunoglobulin G (IgG) and pepsinogen (PG) testing is cost-effective for gastric cancer screening in H. pylori-seroprevalent population.^7,8 ABC classification identifies gastric cancer risk based on the presence of gastric corpus atrophy (PG I level of ≤70 ng/mL and PG I/II ratio of ≤3.0) and H. pylori-seropositivity.⁹ Group A consists of seronegative subjects without gastric corpus atrophy, group B consists of seropositive subjects without gastric corpus atrophy, group C consists of seropositive subjects with gastric corpus atrophy, and group D consists of seronegative subjects with gastric corpus atrophy. Gastric secretory ability is unimpaired in groups A and B, and gastric cancer risk is lowest in group A.^10,11 Conversely, gastric secretory ability is impaired in groups C and D, and gastric cancer risk is highest in group D owing to previous H. pylori infection.^10,11

Asymptomatic group A subjects are often excluded from the candidates for annual endoscopic gastric cancer screening; however, recent studies show that gastric cancers are not uncommon in group A.^12–15 This may reflect the misclassification of seroreversed subjects with false negative findings in ABC classification system. In a recent Japanese study, the risk of gastric cancer in seroreversed subjects after eradication was similar to that in seropositive subjects.¹² Because there is still a risk of gastric cancer after successful eradication or spontaneous regression,¹⁶ it is important to distinguish pseudo group A seroreversed subjects among the group A seronegative subjects. In addition to the pseudo group A subjects with past infection, it is important to discriminate group A subjects with current H. pylori infection. The aim of this study was to determine the prevalence of the condition and characteristics of pseudo group A among group A subjects.

1. Study subjects

Korean subjects over 18 years of age who underwent serum anti-H. pylori IgG test and PG assay on the day of upper gastrointestinal endoscopy between January 2010 and June 2016 were analyzed. The subjects were included in the study, if they satisfied the criteria for group A (H. pylori-seronegative subjects without gastric corpus atrophy) as per ABC classification.⁹ Exclusion criteria were subjects with previous gastrectomy or renal failure. Those who did not answer the questionnaire on past eradication were also excluded. This observational cohort study was approved by the Institutional Review Board of Konkuk University Medical Center (KUH1010626). All subjects provided informed consent before the tests. The study was performed in prospective and retrospective manner by collecting test findings.

2. Questionnaires and endoscopic examination

The subjects were asked about their medical history including successful H. pylori eradication, gastrectomy, renal failure, hypertension, diabetes mellitus, coronary heart disease, and cerebrovascular attack. Questionnaires on social history included cigarette smoking (never, past, or current) and alcohol drinking (almost none, social, or heavy). Based on the National Institute for Alcohol Abuse and Alcoholism guideline, heavy drinking was defined as ≥8 drinks/wk for women and ≥15 drinks/wk for men. Recent drug intake within last 3 months were asked before the endoscopic examination.

After 12 hours of fasting, endoscopy was performed with the aid of either EG-2990i (Pentax, Tokyo, Japan) or GIF-H260 (Olympus, Tokyo, Japan). Endoscopic findings suggesting past infection were determined by the presence of gastric xanthoma (yellowish plaque), metaplastic gastritis (irregular whitish elevations and/or depressed patchy erythema), or advanced atrophy as described.^17,18 Advanced atrophy was defined as visible submucosal vessels extending up to the body (>closed-type 1 in Kimura-Takemoto classification) in this study, because the gastric cancer risk is increased from closed-type 2.¹⁹ Endoscopic images were reviewed by two gastroenterologists (H.K. and S.Y.L.).

3. Diagnosis for H. pylori-seronegativity

Serum anti-H. pylori IgG titer was measured by the Vidas H. pylori IgG assay (BioMérieux, Marcy-l’Etoile, France) till 2012, and by the Chorus H. pylori IgG (DIESSE Diagnostica Senese, Siena, Italy) thereafter. For the Vidas H. pylori IgG assay, seronegativity was defined as test value (TV) of < 0.75 arbitrary units (AU)/mL. Seropositivity was defined as TV of ≥1.00 AU/mL. For the Chorus H. pylori IgG assay, seronegativity was defined as TV of <8.0 AU/mL, and seropositivity was defined as ≥12.0 AU/mL. Sensitivity and specificity of the Vidas assay were 89.7% and 85.5%, and those of the Chorus assay were 100% and 75.4% in Korean subjects, respectively.²⁰

Group A subjects were further classified into four subgroups according to their anti-H. pylori IgG titer. For the Vidas H. pylori IgG assay, the lowest first-quartile, second-quartile, third-quartile, and highest fourth-quartile were defined as the anti-H. pylori IgG titer of (1) ≤0.2 AU/mL, (2) >0.2 and ≤0.4 AU/mL, (3) >0.4 and ≤0.6 U/mL, and (4) >0.6 AU/mL. For the Chorus H. pylori IgG assay, the quartiles were defined as the serology titer of (1) ≤5.0 AU/mL, (2) >5.0 and ≤6.0 AU/mL, (3) >6.0 and ≤7.0 AU/mL, and (4) >7.0 AU/mL, respectively.

4. Serum pepsinogen assay and ABC classification

Serum PG levels were analyzed using the latex enhanced turbidimetric immunoassay (HBi Co., Anyang, Korea).^10,21 Gastric corpus atrophy was defined as a combination of PG I level of ≤70 ng/mL and PG I/II ratio of ≤3.0.⁹ According to the ABC classification system, H. pylori (−)/gastric corpus atrophy (−) subjects were classified into group A. Subjects showing H. pylori (+)/gastric corpus atrophy (−), H. pylori (+)/gastric corpus atrophy (+), and H. pylori (−)/gastric corpus atrophy (+) were classified into groups B, C, and D, respectively.

5. Definition of pseudo group A

Pseudo group A owing to past infection was defined as seroreversed subjects after H. pylori infection, who showed normal findings on anti-H. pylori IgG and PG testings. If there was a history of successful H. pylori eradication, subjects were classified into the pseudo group. Subjects with endoscopic findings suggesting past infection (gastric xanthoma, metaplastic gastritis, or advanced atrophy) were also classified into the pseudo group A.^17,22 Unintended eradication or spontaneous regression was considered as a cause of seroreversion in the latter pseudo group A. Subjects without definite evidence of past infection were classified into true group A.

Current H. pylori infection was defined as positive test findings either in the urea breath test (UBT) or on Giemsa staining on the day of endoscopy. UBT was done using Heliview mass spectrometry (MediChems, Seoul, Korea) as described in our previous study.²⁰H. pylori infection was defined as different ¹³CO₂ concentration between the baseline and 30 minutes samples of >2.4. For Giemsa staining, gastric biopsied specimens were fixed in 95% ethanol. Thereafter, the samples were embedded in paraffin blocks, and were sectioned for hematoxylin & eosin and Giemsa stainings as described.²¹

6. Statistical analysis

Data were analyzed using PASW statistics version 17.0 (SPSS Inc., Chicago, IL, USA). A p-value of <0.05 was considered statistically significant. The κ-value was calculated to evaluate interobserver variability in the endoscopic findings. Differences between the two groups (i.e., true A and pseudo A) were analyzed using the t-test and chi-square test for continuous and categorical variables, respectively. Differences between three or four groups were analyzed by using the post-hoc analysis for continuous variables, and chi-square test with Bonferroni correction for categorical variables. Continuous variables were presented as mean±standard deviation, and were presented as proportion (%). Logistic regression analysis was done to verify independent risk factors for being pseudo group A (being misclassified into group A despite past infection). The values were presented as odds ratio and 95% confidence intervals. Receiver operating characteristic (ROC) curve analysis was done for significant continuous variables to determine cutoff values for being pseudo group A. Area under the curve and standard error were provided with sensitivity and specificity for each cutoff value.

1. Proportion of group A subjects and incidence of gastric neoplasm

Among 7,178 Korean adults who underwent serum PG assay and the anti-H. pylori IgG test on the same day of endoscopy, 274 were excluded owing to equivocal anti-H. pylori IgG test findings (Fig. 1). The subjects were classified into group A (n=2,620), group B (n=3,316), group C (n=898), or group D (n=70). The incidence of gastric adenoma and adenocarcinoma was 0.11% in group A, 0.33% in group B, 1.56% in group C, and 4.29% in group D (Supplementary Table 1). Among the group A subjects with gastric neoplasm, two were classified into pseudo group A. One subject showed advanced atrophy up to the lower body, and the other showed metaplastic gastritis on the antrum and body.

2. Proportion of pseudo group A subjects with past H. pylori infection

Among the 2,620 group A subjects, 448 (17.1%) were classified into the pseudo group owing to previous H. pylori eradication (Fig. 1). Additionally, 133 (5.1%) were also classified into the pseudo group owing to endoscopic findings suggesting past infection. Two gastroenterologists agreed that 139 subjects had endoscopic findings suggesting past H. pylori infection (either gastric xanthoma, metaplastic gastritis, or advanced atrophy); a total of 2,434 subjects had no such findings. Discrepancy between the two gastroenterologists was found in 47 subjects (1.8%). The κ-value was 0.846 (p<0.001).

3. True group A versus pseudo group A with past infection

The true group A subjects were younger (p<0.001) and had a lower predominance of male sex (p=0.001) than the pseudo group A subjects (Table 1). The true group A subjects were born more recently (mean year of birth: 1962.7±11.5) than the pseudo group A subjects (1957.0±9.8). Moreover, the pseudo group A subjects showed higher prevalence of hypertension (p=0.002), diabetes mellitus (p=0.017), aspirin use (p=0.037), and cigarette smoking (p<0.001) than the true group A subjects. The true group A subjects showed lower PG I and PG II levels (51.8±21.1 ng/mL and 8.9±4.5 ng/mL) than the pseudo group A subjects (53.6±25.0 ng/mL and 9.6±5.9 ng/mL). The serum anti-H. pylori IgG titers were also lower in the true group A than their counterparts (Table 2).

4. Comparisons between the two pseudo group A subjects with past infection

The pseudo group A subjects with endoscopic findings suggesting past infection were older (p<0.001) and had a higher predominance of male sex (p<0.001) than the pseudo group A subjects with eradication history. The pseudo group A subjects with eradication history were born more recently (mean year of birth: 1957.6±9.8) than the pseudo group A subjects with abnormal endoscopic findings (1954.9±9.6). The quartiles of the anti-H. pylori IgG titers also differed among the groups (Table 2). The pseudo group A subjects with eradication history showed highest anti-H. pylori IgG titers and PG II levels among all groups.

5. Differences according to the quartiles of the serum anti-H. pylori IgG titer

The subjects in the highest fourth quartile were older (p=0.001) and born earlier (p<0.001) than the subjects in the other quartiles (Table 3). Moreover, these subjects showed the highest mean PG II titer and the lowest PG I/II ratio among the subjects in all four groups. No difference was found among the groups with regard to smoking history (p=0.538), drinking history (p=0.634), recent intake of aspirin (p=0.378), antithrombotic drug intake (p=0.995), nonsteroidal anti-inflammatory drug use (p=0.136), presence of coronary heart disease (p=0.262), cerebrovascular disease (p=0.610), hypertension (p=0.480), and diabetes mellitus (p=0.329).

6. Group A subjects with current H. pylori infection

Among the 2,620 included subjects, 715 underwent the UBT or Giemsa staining on the same day of endoscopy (including 130 pseudo group A subjects with eradication history and 92 pseudo group A subjects with endoscopic findings suggesting past infection). Twenty subjects underwent both the UBT and Giemsa staining, and there was no discrepancy between the two tests in these subjects (all negative test findings). Other 599 subjects underwent Giemsa staining only, and 96 underwent UBT only. Positive test findings were found in 22 subjects (3.1%) (Supplementary Table 2). Among all variables, only the PG I and PG II levels differed between the H. pylori test-positive and -negative subjects (Table 4). The H. pylori-test findings were not affected by the anti-H. pylori IgG titer, age, year of birth, eradication history, and endoscopic findings suggesting past infection.

7. Independent risk factors for being pseudo group A owing to past infection

On multivariate analysis, age and year of birth were independent risk factors for being misclassified into group A despite past H. pylori infection (Table 5). The cutoff point for age was 50.5 years (sensitivity 72.1% and specificity 53.9%) on ROC curve analysis (p<0.001). The cutoff point for year of birth was 1959.5 (sensitivity 60.6% and specificity 61.2%, p<0.001) for being classified into pseudo group A (Fig. 2).

In this study, 22.2% of the group A subjects had past H. pylori infection, and were found to be misclassified into group A. The subjects aged over 50 years and those born before 1960 were included in the high-risk groups for being misclassified into pseudo group A, either by previous eradication or spontaneous regression of H. pylori. Moreover, 3.1% of the subjects had current infection; however, being classified into pseudo group A owing to past infection did not increase the risk of current infection. Positive H. pylori test findings were correlated only with the PG I and PG II levels. Taken together, past infection should not be neglected in older subjects, and current infection should be ruled out in subjects with increased gastric secretory ability when using the ABC classification in a seroprevalent population.

Seroreversion usually occurs within a few years after H. pylori eradication or spontaneous regression.²³ In addition to a rapid decrease in the anti-H. pylori IgG titer, the gastric secretory ability is often recovered after H. pylori regression.²⁴ Therefore, many seroreversed subjects with past H. pylori infection can be misclassified into group A as shown in this study. In addition to those subjects who underwent intended H. pylori eradication, subjects after unintended eradication or those who showed spontaneous regression also revealed normal serum PG assay findings after seroreversion. Nonetheless, the gastric cancer risk remains high in the group A subjects with past infection.¹² In our study, gastric neoplasm was found in three subjects in group A (0.11%), and two subjects showed endoscopic findings suggesting past infection. This is consistent with the findings of a recent study showing that most of the group A patients with gastric cancer had past H. pylori infection.²⁵

Older age was an independent risk factor for being classified into pseudo group A, with a cutoff point of 50.5 years in this study. Furthermore, group A Koreans born before 1960 had a higher risk. Mean year of birth was most remote (1955) in subjects with spontaneous regression, and was most recent (1963) in subjects without evidence of past infection. In subjects after successful eradication, mean year of birth (1958) was closer to that of subjects with spontaneous regression than subjects without evidence of past infection. Our findings are supported by the fact that more than 40% of group A subjects show atrophy on endoscopic findings, especially when they are older.²⁶ Owing to these reasons, the ABC classification system is less useful for an older, seroprevalent population.²⁷ Older subjects should be considered as having a high risk for gastric cancer owing to the high probability of past infection.

Another notable finding in this study is that 3.1% of the 715 group A subjects who underwent additional H. pylori test showed positive findings. Along with the seroreversed pseudo group A subjects with past infection, subjects with ongoing infection should not be neglected. False seronegative findings in infected Koreans may be due to non-availability of Korean serology assays at the clinic. The only Korean immunoassay, “Genesis H. pylori ELISA” from Green Cross Medical Science is available only for research purposes, and does not show quantitative results. Therefore, imported H. pylori IgG assays, which exhibited a similar diagnostic accuracy with the Genedia H. pylori ELISA in Koreans, were used in this study.²⁰ Nonetheless, 22 infected subjects were misclassified into the seronegative group. Because these subjects showed high mean PG I (75.3 ng/mL) and PG II (16.4 ng/mL) levels, this could be improved by adding other H. pylori tests to the group A subjects with increased PG levels.

The pseudo group A subjects with eradication history showed a higher serum anti-H. pylori IgG titer than those with abnormal endoscopic findings suggesting past infection. Nonetheless, classification into pseudo group A and its correlated variables (i.e., old age, male sex, and earlier year of birth) did not increase the risk of current infection. The lack of a correlation between the increased serology titer and current infection supports the finding that the IgG titers decrease slowly after eradication.¹⁴ Despite such limitations, the ABC classification using the PG assay is still useful in Koreans with eradication history, because atrophy and metaplastic gastritis can be improved after eradication.²⁸ Another Korean study showed that the PG I/II ratio of subjects becomes similar to that of normal controls after 5 years of eradication.²⁹ The gastric biomarker levels in subjects with eradication history seem to recover to the level in noninfected subjects after 3 years, although a longer duration is needed in subjects with severe corpus atrophy.^30,31

There are limitations in our study. First, eradication history was based on the answers to questionnaires. There might have been a recall bias; however, the answers are usually reliable because it is not easy to forget 7 to 14 days of antibiotic intake for H. pylori eradication by intention. Moreover, in a seroprevalent population, the risk of providing imprecise information on past eradication might be lower than the risk of unintended eradication or spontaneous regression. Therefore, we hypothesized that the proportion of misclassified pseudo group A subjects owing to recall bias is small, and would not affect the main findings of this study. Second, additional H. pylori tests were performed only in 715 subjects. Nevertheless, we could verify that positive H. pylori test findings are correlated with the PG I and PG II levels, and not with the pseudo group A-related variables. This further suggests that current infection is uncommon even in pseudo group A subjects with a high serology titer.

In conclusion, 22.2% of the group A subjects were misclassified despite the gastric cancer risk, and the efficacy of the ABC classification method is limited owing to these pseudo group A seroreversed subjects. Along with pseudo group A subjects with past infection, those with current H. pylori infection can be misclassified into group A. Ongoing H. pylori infection should be considered in subjects with increased PG I and PG II levels, regardless of their age, serology titer, eradication history, and endoscopic findings. To discriminate seroreversed subjects with a risk of gastric cancer, past infection should be assessed in subjects aged over 50 years, especially in those born before 1960. Furthermore, current infection should be ruled out in subjects with increased gastric secretory ability.

This work was supported by Konkuk University and National Research Foundation (2016R1D1A1B02008937).

No potential conflict of interest relevant to this article was reported.

Drafting of the manuscript: H.K., S.Y.L. Study concept and design: S.Y.L. Data acquisition, data analysis and interpretation, and statistical analysis: H.K., J.H.K., S.P.L. Administrative, technical support, and study supervision: J.H.K., I.K.S., H.S.P., C.S.S.