Different etiological role of Helicobacter pylori (Hp) infection in carcinogenesis between differentiated and undifferentiated gastric cancers: A nested case-control study using IgG titer against Hp surface antigen

Abstract

Background. The present study epidemiologically clarified the different roles of Helicobacter pylori (Hp) infection in carcinogenesis between two major histological types of noncardia gastric cancer (ncGC), intestinal (=differentiated) and diffuse (=undifferentiated), by analyzing IgG antibody titer against Hp surface antigen on the data set of a nested case-control study in a large cohort study conducted in Japan. Methods. A total of 36 745 subjects aged 40 to 69 years in the Japan Health Center-based prospective study who responded to the baseline questionnaire and provided blood were followed over 15 years; 350 ncGC cases (differentiated=242, undifferentiated=108) matched to controls were used. Using baseline blood samples, plasma IgG titer was measured using ELISA. The level of IgG titer >10 U (cut-off value) was classified into three grades in groups of equal number: low, middle, and high. Results. IgG titer of Hp was significantly (p<0.01) higher in undifferentiated cases than in differentiated ones. Among the three grades of Hp IgG titer, the high titer was more closely associated with risk of undifferentiated ncGC (odds ratio (OR) for high=7.8, 95% confidence interval (CI)=2.4–24.9 vs. OR for low=6.4, 95% CI=2.1–19.6), while the low titer was a better predictor of differentiated ncGC (OR for high=3.2, 95% CI=1.6–6.4 vs. OR for low=5.9, 95% CI=3.0–11.6, trend p < 0.05). The high titer group had the lowest risk to develop differentiated ncGC with <7 years (OR=3.2, 95% CI=1.3–7.7), whereas the high titer group demonstrated the highest risk for undifferentiated ncGC developing (OR=11.6 95% CI=2.3–59.1). Conclusion. Our study epidemiologically confirmed that atrophic changes caused by Hp infection determine the development of differentiated-type ncGC, and the inflammation itself induced by Hp infection promotes the development of undifferentiated-type ncGC.

Helicobacter pylori (Hp) infection is one of the major risk factors for developing noncardia gastric cancer (ncGC), which includes two major histological types, ‘differentiated’ and ‘undifferentiated’ [1–3]. Differentiated-type ncGC is linked to the atrophic gastric mucosa with intestinal metaplasia, and the undifferentiated type arises from the gastric mucosa without atrophy or intestinal metaplasia [4]. Although the background mucosa and origins are different between the differentiated and undifferentiated types, accumulating evidence seemingly demonstrate that Hp infection plays similar roles in the carcinogenesis of ncGC [2], [3], [5–7]. Since the risk of Hp infection for ncGC (odds ratio) is almost the same, the pepsinogen level, which is a useful biomarker to determine the atrophy of gastric gland [8], demonstrates similar behavior between the differentiated and undifferentiated types, particularly the ratio of Pepsinogen I (PGI) to Pepsinogen II (PGII) [9].

In most epidemiological studies, Hp infection is qualitatively determined by serological test, and no quantitative analysis of the density of Hp has been conducted. Eradication of Hp significantly and promptly reduces the level of conventional immunoglobulin G (IgG) antibody titer against the surface protein of Hp in humans, demonstrating that monitoring serum IgG levels after treatment provides an early indicator of the efficacy of therapy in eradication of Hp infection [10–12]. These evidences imply that IgG titer could be a marker for the density of Hp; a low density of Hp is linked to mucosal atrophy and a high density is associated with severe inflammation [13].

Extensive pathological evidences indicate that the sequence of histological changes caused by chronic consistent inflammation by Hp is the determinant in the development of the differentiated-type ncGC [4], whereas the pathological and molecular mechanisms that Hp infection promotes to develop undifferentiated-type ncGC are poorly understood. It is speculated that inflammation itself could be involved in carcinogenesis of undifferentiated-type. We hypothesized that the low density of Hp is a risk for differentiated ncGC, and the high density of Hp is a risk for undifferentiated-type ncGC, and thus the behavior of the IgG titer is different in the two types of ncGC. In this study, we tried to confirm our hypothesis by analyzing the IgG titer based on the data from a large case-control study nested within a prospective cohort conducted in Japan, in which ORs were similar (5.8 for differentiated type vs. 5.1 for undifferentiated type) [7].

Materials and methods

The study population and study design were described in detail as elsewhere [7]. Briefly, a population-based cohort of 61 009 men and 62 567 women was established and followed from January 1, 1990 to December 31, 2004. Among 97 644 eligible subjects of the JPHC Study cohort (46 803 men and 50 841 women), 13 467 (29%) men and 23 278 (46%) women donated their blood samples at baseline and were included in the study. Newly diagnosed cases of cancer were collected through two data sources, one from local major hospitals and the other from population-based registries. Cases of gastric cancer were extracted from the cancer registry for the JPHC study based on site (International Classification of Diseases for Oncology (ICD-O) code C160-169; [14]). Among 1 681 cases with a histologically proven diagnosis, plasma at baseline was obtained from 512 cases.

Until quite recently in Japan, the upper third of the stomach has been called the “cardia” on the basis of the guidelines for gastric cancer classification [15]. Because it seemed difficult to distinguish the so-called “cardia,” which is located mainly in the esophagogastric junction, from the upper third of the stomach, we combined tumors at these sites into 1 group for analysis in this study (ICD-O code C160-161). A tumor located on the lower side of the stomach was classified as a distal gastric cancer (ICD-O code C162-167). Those subsites that could not be classified because of an undifferentiated lesion (ICD-O code C168) or those with no information (ICD-O code C169) were categorized as an unclassified subsite. Histologic classification was based on one of the author's (S. Sasazuki) reviews, in consultation with a pathologist, of the record reported by each hospital. The subdivisions were made on the basis of the classification derived by Lauren [16].

For each case, we selected one control matched for gender, age (±3 years), study area, blood donation date (±2 months), and fasting time at blood donation (±5 hours). The original data set included 511 sets of 511 cases and 511 controls [7], excluding one case with incomplete data and the matched control. Of 511 cases and controls, 368 cases involved the distal portion (noncardia gastric cancer). In this study, 350 cases including 242 differentiated types (intestinal type) and 108 undifferenciated types (diffuse type) were analyzed, because 18 cases had no information or were unclassified histological types.

IgG antibody titers to Hp were measured with a direct ELISA kit (E Plate “Eiken” Hp Antibody; Eiken Kagaku Co. Ltd. Tokyo, Japan). IgG titers were determined by optical density reading in relation to a standard curve. The level of IgG titer was classified into negative cases (<10) and three positive grades (low 10–32.8; middle: 32.8–54.6; high: 54.6 − ). Negative was determined because of the recommended kit cut-off value. Each threshold value was determined for the same number of cases in all 3 groups.

Chi-square test, one-way ANOVA, or U- test was used to determine statistical significance for difference. Matched odds ratio (OR) and 95% confidence intervals (CIs) were calculated using conditional logistic regression analysis. All statistical analysis was performed using SAS version 9.1.2.

Results

Baseline characteristics are shown in Table I. The cases of differentiated ncGC were 3 years older than those of undifferentiated type. PGI, PGII and the ratio of PGI and PGII levels were significantly higher [33.2 vs. 22.8 (p < 0.001), 14.7 vs. 11.5 (p < 0.001), and 2.2 vs. 1.9 (p < 0.05), respectively) in the cases of undifferentiated type than of differentiated type.

Table I.  Baseline characteristics of case and control subjects.

		Differentiated		Undifferentiated
		case n = 242	control n = 242	case n = 108	control n = 108
age	mean (SD)	57.5 (7.1)**^2)	57.5 (7.1)	54.3 (6.6)	54.3 (6.7)
Sex	men/women	186/56**^2)	186/56	51/57	51/57
Pepsinogen I (PGI)	mean (SD)	22.8 (14.5)**^2)	27.9 (18.5)	33.2 (18.8)	30.0(14.5)
Pepsinogen II (PGII)	mean (SD)	11.5 (6.9)**^2)	11.2 (7.0)	14.7 (7.3)	10.7 (5.8)
Ratio of PGI to PGII	median (IQR)	1.9 (1.5)*^2)	2.5 (2.2)	2.2 (1.1)	3.0 (1.8)
Hp IgG titer
entire period^1)	median (IQR)	36.0 (31.0)**^2)	34.1 (47.9)	46.2 (33.5)	39.2 (46.5)
< 7 years^1)	median (IQR)	34.7 (32.6)	30.7 (48.0)	37.7 (35.7)	33.4 (46.1)
> = 7 years^1)	median (IQR)	38.5 (34.1)*^2)	38.5 (48.0)	51.1 (39.1)	48.3 (51.1)

SD = standard deviation IQR= intraqurtail range.

¹⁾period between on-set and blood donation.

²⁾compared to undifferentiated cases **p < 0.01, *p < 0.05.

The median (intraquartile range; IQR) of IgG titer was also shown in Table I by stratification with duration between blood donation and cancer. IgG titer was significantly (p < 0.05) lower in the differentiated ncGC cases than in the undifferentiated ones with onset ≥7 years after blood donation, while there was no significant difference in IgG titer between histological types in the cases with onset <7 years after blood donation.

Spearman's correlation coefficients between IgG titer and pepsinogens were calculated by stratification with histological type, and the results are shown in Table II. In the cases and controls, both PGI and PGII were significantly correlated with Hp titer, and the correlation coefficient between IgG titer and PGII showed a higher correlation, compared to that between IgG titer and PGI. Control cases showed a higher correlation between Hp titer and PGII (Spearman's r = 0.61 or, 0.69). In cases of differentiated-type ncGC, however, the correlation coefficient value was almost half of that (r = 0.26). In the undifferentiated cases, the association between Hp titer and PGI was closer (r = 0.35) than that of differentiated cases. Interestingly, although the control cases showed a significant negative association (differentiated: r = − 0.48; undifferentiated: r = − 0.61) between IgG titer and ratio of PGI to PGII, there was no significant association in either type of cancer cases (differentiated: r = − 0.002 or undifferentiated: r = 0.02).

Table II.  Spearman's correlation coefficient between IgG titer and Pepsinogens.

	Differentiated		Undifferentiated
	case n = 242	control n = 242	case n = 108	control n = 108
Pepsinogen I (PGI)	0.20** (0.07, 0.32)	0.16 *(0.03, 0.28)	0.35 **(0.19–0.50)	0.29** (0.11, 0.47)
Pepsinogen II (PGII)	0.26**(0.15, 0.38)	0.61**(0.53, 0.70)	0.38** (0.22–0.55)	0.69** (0.58, 0.80)
Ratio of PGI to PGII	−0.002 (−0.14, 0.13)	−0.48**(−0.59, −0.37)	0.02 (−0.17–0.21)	−0.61** (−0.75, −0.47)
PGI and PGII^1)	0.64** (0.56, 0.73)	0.60**(0.52, 0.68)	0.79**(0.72–0.87)	0.66** (0.54, 0.77)

( ):95%confidence interval.

1) Spearman's correlation coefficient between PGI and PGII.

*P < 0.05, **p < 0.01.

In Table III, the odds ratio (OR) for gastric cancer associated with IgG titer was shown by stratification with histological types and duration between blood donation and cancer. There was a distinct difference between differentiated and undifferentiated types in the duration between blood donation and cancer. The OR for differentiated GC was higher in the onset 7 years later than within 7 years, whereas Hp-positive subjects demonstrated no significant risk for the undifferentiated type after 7 years.

Table III.  Odds ratios (ORs) for association of noncardia gastric cancer.

		Differentiated					Undifferentiated
period of on-set	Hp titer	case	control	OR	95%CI		case	control	OR	95%CI
Entire period	negative	17	63	reference			4	26	reference
	low	92	56	5.9	3.0	11.6	30	22	6.4	2.1	19.6
	middle	73	59	4.4	2.2	8.5	33	29	5.9	1.9	18.5
	high	60	64	3.2	1.6	6.4	41	31	7.8	2.4	24.9
		p for trend among positive cases < 0.05					5]p for trend among positive cases = 0.706
<7 year
	negative	13	43	reference			2	18	reference
	low	56	35	5.8	2.5	13.6	21	13	9.2	2.0	41.9
	middle	47	38	3.9	1.8	8.7	20	18	7.8	1.6	38.2
	high	34	34	3.2	1.3	7.7	21	15	11.6	2.3	59.1
		p for trend among positive cases = 0.395					p for trend among positive cases = 0.811
> = 7year
	negative	4	20	reference			2	8	reference
	low	36	21	6.5	2.0	20.5	9	9	3.5	0.6	20.5
	middle	26	21	5.7	1.7	19.7	13	11	3.9	0.7	20.4
	high	26	30	3.4	1.1	11.0	20	16	4.5	0.8	23.4
		p for trend among positive cases <0.05					]p for trend among positive cases = 0.758

In differentiated cancer cases, subjects with the lowest group of IgG titers demonstrated the highest risk (OR = 6.5 95% CI = 2.0–20.5,) for ncGC onset with ≥7 years, and the decreasing trend as IgG titer increased (trend p < 0.05). This trend was also seen during the period <7 years, although it did not show significance. On the other hand, in the undifferentiated type, subjects in the high titer group showed the highest risk (OR = 7.8 95% CI = 2.4–24.9) during the entire period. In particular, there were few subjects with seronegativity when the onset was within 7 years.

Discussion

In our large-scale nested case-control study, the behavior of Hp IgG titer in association with the risk for GC was significantly different between the two histological types. Subjects with a high IgG titer had a low risk for developing differentiated-type ncGC among Hp-positive ones in the near future, but a high risk for undifferentiated-type ncGC. The low titer groups, on the other hand, revealed a high risk for differentiated-type ncGC. These results were consistent with our hypothesis.

In the present study, the methodological advantages include the prospective design, the large sample size, the long-term (15 years) follow-up, and the accurate classification of location and tumor histology. Estimates of the association between Hp serological evaluation and ncGC risk depend on study design. Hp density is decreased in the circumstances such as exist in the gastric mucosa with atrophic changes and intestinal metaplasia, linked to ncGC [17]. Consequently, the titer of Hp antibodies shows a decreasing trend as the stage of gastric mucosa becomes more advanced, and the accuracy of retrospective case-control studies could be limited in the interpretation of the results. As in our study, long-term prospective studies are more accurate than retrospective case-control studies.

Previously, we demonstrated that the status of inflammation affected the ratio of PGI to PG II, and Hp titer was correlated with PGII more closely than PGI [18]. Consequently, subjects with high titers showed a low ratio of PGI to PGII. In this study, a similar correlation between Hp titer and PGs was observed among controls. These findings indicate that the severe inflammation of the gastric mucosa decreases the ratio of PGI to PGII. The undifferentiated cases showed a low ratio of PGI to PGII and the highest PGI and Hp titer. In addition, there were few cases with seronegativity in the undifferentiated cases within 7-year onset. The progression of mucosal atrophy decreases the IgG titer, resulting in false-negative serological test for Hp [19]. Taken together, undifferentiated cases were characterized by the presence of severe inflammation in the gastric mucosa without atrophy. The differentiated-type ncGC, on the other hand, showed the lowest PG I and ratio of PGI to PGII; the low titer was linked to the risk, indicating the presence of mucosal atrophy.

In this study, we also found an interesting association between IgG titer and the ratio of PGI to PGII, which was significantly and inversely correlated with Hp titer among control subjects. The correlation was absent among ncGC cases. The mechanism by which the correlation between the ratio of PGI to PGII and Hp titer was modified among cancer cases was as follows. The cytokines induced by Hp play crucial roles in evoking inflammation on the gastric mucosa [20]. Many studies reported the possible association between polymorphisms of proinflammatory cytokines such as IL-1beta, IL-2, IL-8, and TNF-alpha and the risk of ncGC [21–24]. Altered inflammatory cytokine profiles on the gastric mucosa could be prone to be ncGC [25]. The expression of PGI and PGII could be altered by some specific inflammatory cytokines among GC cases. The cancer-specific cytokine profiles might thus modify the association between IgG titer and the ratio of PGI and PGII among cancer cases.

Most studies on the polymorphism of cytokines showed that the association in atrophic gastritis or hypochlorhydria was related to differentiated-type ncGC [26]. A recent study has reported that characterization of immune responses in gastric cancer patient is different between two types [27]. The number of B cells was significantly increased, and the number of T cells significantly decreased in differentiated- compared to undifferentiated-type tumors, suggesting that cytokines induced by Hp may differ between histological types.

The association between Hp seropositivity and ncGC risk depended on the time of onset. The seropositivity of Hp showed an increased risk for differentiated ncGC in the long-term, compared to the short-term. Although few undifferentiated cases with seronegativity developed within 7 years, no association was found among the cases with onset more than 7 years after blood donation. These results suggest that the undifferentiated-type ncGC might develop more rapidly after infection of Hp than the differentiated-type ncGC.

CagA-positive strain causes more extensive inflammation of the gastric mucosa. Some studies have shown that the seropositivity against Cag A protein reflects a closer association between Hp and GC [28]. Although we also examined for an association between IgG titer against CagA and ncGC, none was found (unpublished data). Several reasons seem possible. It is accepted that antibodies to CagA persist longer after eradication of Hp, and titer of CagA could not have been associated with the density of Hp. In addition, as we previously reported, the titer of CagA shows ethnic differences [29]. The antigen of CagA is specific to a given ethnic population. Further studies are therefore needed to clarify the relationship between CagA titer and GC risk.

The present study used an ELISA kit to measure IgG titer for antibody against Hp surface. The low titer group demonstrated the highest risk for differentiated ncGC. We consider that the false-negative subjects could have more risk for differentiated ncGC, and that it is important to identify the group. Previous work showed that the band related to Urease B was detected by immunoblotting in the sera of a subject with lower value than the recommended cut-off value of the kit [30]. Therefore, the immunoblot assay may be useful to identify the false-negative subjects and determine the cut-off value.

Among the 97 644 eligible subjects of the JPHC Study cohort, in this study the subjects were 36 745 (38%) men and women who participated in provided blood samples and the survey, and they were ones who participated in the health checkup. When we examined for differences between non-participants and participants in the health checkup survey, especially women had a different socioeconomic status and a favorable lifestyle profile, such as smoking less, participating in more physical exercise, and eating more green vegetables or fruits [31]. This finding means that some caution is needed in generalizing or interpreting the results.

In summary, the etiological roles of Hp infection in carcinogenesis differ with the histological type. Our study epidemiologically confirmed that the histological changes caused by chronic consistent inflammation by Hp is the determinant for developing differentiated-type of GC, and the severe inflammation itself evoked by Hp infection promotes development of the undifferentiated-type of ncGC.