Helicobacter pylori Infection and Gastric Cancer Risk: Evaluation of 15 H. pylori Proteins Determined by Novel Multiplex Serology

Infection with Helicobacter pylori is a major cause of gastric cancer (GC). The association likely has been underestimated in the past due to disease-related clearance of the infection. On the other hand, only a minority of the infected individuals develop GC, and better risk stratification is therefore highly desirable. We aimed to assess the association of GC with antibodies to 15 individual H. pylori proteins, determined by novel multiplex serology, to identify potentially relevant risk markers. This analysis was based on 123 GC cases aged 50 to 74 years and 492 age-matched and sex-matched controls from Saarland, Germany. Eight of the antibodies were significantly associated with noncardia GC and seven of them were significantly related to GC at any site. More pronounced associations were observed for noncardia GC; adjusted odds ratios (95% confidence intervals) ranged from 1.60 (1.01–2.54) for HyuA to 5.63 (3.20–9.91) for cytotoxin-associated antigen A (CagA). A dose-response relationship was found between the number of seropositivities and GC (P < 0.001). The seropositivities of CagA and GroEL were found to be independent predictors of GC, which were strongly related to GC risk in a dose-response manner (P < 0.001). In conclusion, GroEL was identified as a new independent risk marker that may contribute to enhanced quantification of H. pylori–related GC risk. [Cancer Res 2009;69(15):6164–70]

Despite a worldwide decline in incidence, gastric cancer (GC) is still the second most common cause of cancer-related mortality (1). Helicobacter pylori infection has been identified as a major cause of GC, especially noncardia GC (2–4). However, the association likely has been underestimated in the past due to disease-related clearance of the infection (5). According to recent studies, H. pylori infection might even be a close to necessary condition for noncardia GC (6). Nevertheless, only a minority of infected individuals develop the disease, and identification of virulence factors of the infection for risk stratification is highly desirable.

Seropositivities for antibodies against the cytotoxin-associated antigen A (CagA) and vacuolating toxin (VacA) have long been established as important virulence markers of H. pylori (7, 8). In the past few years, additional serologic markers of H. pylori infection that might serve as potential predictors for the development of gastric diseases were identified by immunoproteomics studies (9–12). However, very few of them have been confirmed in population-based epidemiologic studies (13, 14). Recently, we have assessed the associations of seropositivity for antibodies to 15 H. pylori immunogenic proteins, measured with a newly developed multiplex serology method using chronic atrophic gastritis (CAG, an established important precursor of intestinal GC; ref. 15), in a large population-based study from Germany (16). Helicobacter cysteine–rich protein C (HcpC) and the chaperonin GroEL were identified as new independent virulence factors, and in combination with the established virulence factors, CagA and VacA, were strongly associated with CAG. In the present study, we aimed to assess the relationship of seropositivity for these 15 H. pylori–specific antigens with GC in order to verify potentially relevant predictive markers for gastric carcinogenesis.

For this analysis, we combined the data from the ESTHER study and the VERDI study, two large population-based studies conducted in the entire state of Saarland, located in the southwest of Germany. The details of the study designs have been described elsewhere (6, 17, 18). Briefly, 2,043 patients aged 50 to 74 years with first diagnoses of various forms of cancer were recruited in the clinical arm of ESTHER between January 2001 and December 2003, and 908 patients below age 80 with a first diagnosis of breast, colorectal, or gastric cancer were recruited in VERDI between October 1996 and February 1998. In addition, 9,953 participants aged 50 to 74 years were recruited in the community arm of ESTHER during a general health check-up by their general practitioners in 2000 to 2002. The studies were approved by the ethics committees of the medical faculty of the University of Heidelberg and of the medical board of the state of Saarland. Written informed consent was obtained from each participant.

In total, 153 out of 2,951 (2,043 plus 908) recruited cancer patients were GC cases with histologically verified diagnosis, and 6 of them were excluded because of missing baseline data or serum samples. For this analysis, we included 123 GC cases aged 50 to 74 years (100 with noncardia GC and 23 with cardia GC), and 492 control subjects from the community arm of ESTHER (frequency-matched on sex and 5-year age group) without a history of GC or CAG.

Questionnaires. Standardized questionnaires were completed by every participant, providing information on sociodemographic characteristics, medical history, health status, family history, and lifestyle factors.

Serologic examinations. Serum samples were obtained from all participants in this study and stored at −80°C. Thirty-nine samples from GC cases were collected at diagnosis or before treatment (operation or chemotherapy), 77 within 3 months after starting treatment and 7 within 4 to 10 months after starting treatment. Initially, the H. pylori infection was assessed by a commercial ELISA for the presence of IgG antibodies against H. pylori (H. pylori screening ELISA; ravo Diagnostika). Infection status was classified according to the instructions of the manufacturer; borderline results were treated as negative.

Novel H. pylori multiplex serology was used to determine the serostatus of 15 specific H. pylori antigens as listed in Table 2. The details of this method have been described elsewhere (16).³

In brief, the multiplex method was based on a glutathione S-transferase capture immunosorbent assay combined with fluorescent-bead technology (19). The quantity of bound antibodies was determined as the median reporter fluorescence intensity (MFI) of at least 100 beads per bead set per serum. The determination of the serostatus of the specific antibodies was based on MFI. Validation of the cutoff points and correlations between seropositivity for H. pylori–specific antigens have been described in detail elsewhere (16).

Cases and controls were characterized with respect to age, sex, education, family history of GC, alcohol consumption, smoking status, and H. pylori serostatus (defined by H. pylori screening ELISA). Differences between cases and controls in these variables were assessed with χ² test. The analyses were repeated after excluding cases with cardia GC.

Associations between GC and serostatus for the 15 H. pylori proteins (and also for the previously conducted H. pylori screening ELISA) were assessed by means of odds ratios (OR) and 95% confidence intervals (95% CI) using unconditional multiple logistic regression. Age, sex, education, family history of GC, alcohol consumption, and smoking status were controlled for in all models. Next, in order to investigate the combined effect of such identified significant predictors, the risk of GC was then assessed according to the number of seropositivities for pertinent antibodies in each sample. The analyses were repeated according to H. pylori serostatus defined by screening ELISA.

Colinearity regarding serostatus for these significant predictors in the multiple logistic regression analyses was checked (the largest condition index exceeding 30 was regarded as indicating multicolinearity) and stepwise backward multiple logistic regression analysis was then used to identify those serum markers (coded as binary variables) that were independently associated with GC risk. The significance level for serum markers to stay in the model was 0.05. Dose-response relationships between the thus identified independent predictors and GC were evaluated using quartiles of antibody levels (MFI) in controls to categorize the serostatus of each antibody, and the group of subjects with the lowest quartile level was regarded as the reference. Finally, cross-classification of H. pylori serostatus, determined by the seropositivity of H. pylori screening ELISA and of the identified independent predictors, was used to predict the development of CAG. All statistical analyses were carried out using SAS statistical software, release 9.1.

Table 1 shows the distribution of sociodemographic and lifestyle factors and H. pylori serostatus defined by H. pylori screening ELISA among the 123 GC cases (23 with cardia GC and 100 with noncardia GC) and 492 age-matched and sex-matched healthy controls. Significant differences were observed for education level and H. pylori seropositivity, both for cases with any form of GC and for cases with noncardia GC compared with healthy controls (P < 0.05).

Table 2 depicts the prevalence of seropositivity for the 15 H. pylori–specific antibodies analyzed by multiplex serology. The associations between GC and seropositivity for each of the 15 antibodies are shown in terms of adjusted OR (95% CI), controlling for age, sex, education, family history of GC, alcohol consumption, and smoking status. When looking at all GC cases, associations were significant for seven antibodies, with adjusted ORs (95% CI) ranging from 1.62 (1.06–2.48) for Cag16 to 3.93 (2.44–6.34) for CagA. The associations for these antibodies increased when cardia GC cases were excluded from the analyses and eight of them were statistically significant, with adjusted ORs (95% CI) ranging from 1.60 (1.01–2.54) for HyuA to 5.63 (3.20–9.91) for CagA.

As shown in Table 3, both overall GC risk as well as noncardia GC risk strongly increased with the number of seropositivities for the eight significant predictors in a dose-response manner (P < 0.001). We repeated the analyses according to H. pylori serostatus defined by screening ELISA. The positive dose-response relationship was very strong among screening ELISA negative subjects, and weaker and not statistically significant in positive subjects. Similar results were observed even after excluding CagA from the analyses (Supplementary Table S1).

No major colinearity was observed regarding the serostatus of these eight H. pylori proteins. In the multiple regression analysis for all GC and noncardia GC, the largest condition indices were 6.60 and 7.60, respectively. Stepwise backward selection identified seropositivities for CagA and GroEL as independent predictors of GC and noncardia GC (Table 4). In additional analyses excluding covariates (i.e., education, family history of GC, alcohol consumption, and smoking status) from the model, similar results were observed. Strong, significant dose-response relationships between antibody levels (categorized by quartiles of MFI) and GC were observed for both CagA and GroEL (Table 5).

As shown in Table 6, when seropositivities for CagA and GroEL were combined with H. pylori seropositivity defined by the screening ELISA, much stronger associations with GC and noncardia GC were observed compared with the results based on screening ELISA only. Intriguingly, the increase in risks appeared particularly pronounced in subjects who were H. pylori ELISA–negative but tested positive for both CagA and GroEL. Finally, in additional analyses based on joint classification by serostatus of CagA and GroEL, a very strong increase in GC risk was observed in the presence of both markers (Supplementary Table S2).

In this population-based study among older adults from Germany, 7 of the 15 serum markers for H. pylori proteins measured by multiplex serology were shown to be significantly associated with GC. The associations were more pronounced for noncardia GC and 8 of them were statistically significant. The risk increased with the number of seropositivities. Seropositivities for CagA and GroEL were identified as independent risk predictors, which were strongly related to GC risk in a dose-response manner.

H. pylori infection is well recognized as a key cause triggering the development of GC, especially noncardia GC (2, 4). However, in associations between H. pylori infection and GC, heterogeneity was observed among previous studies according to the age of the participants, type and stage of GC, study base (population-based versus hospital-based), and virulence of the infecting strains (3, 8, 20). Apart from these potential explanations, disease-related clearance of H. pylori likely contributes to such heterogeneity as well. There is evidence that with the development of advanced gastric disease, the organism can be lost from the stomach due to the changed internal environment. Levels of circulating antibodies will decrease and finally disappear spontaneously subsequent to the loss of infection, so that patients with GC may be H. pylori–seronegative even though they have been infected in the past (21, 22). Consistent with this hypotheses, in our study, much lower H. pylori seropositivity (defined by H. pylori screening ELISA) was observed for GC patients at severe stages (III–IV; 53.8%) compared with those at less severe stages (0–II; 78.3%). Therefore, estimates of the association between H. pylori infection and GC may have often been underestimated in the past.

Considering the fact that some H. pylori proteins may have stronger antigenicity and their antibodies may persist much longer after the clearance of infection, seropositivity for such antigens is suggested to be a useful supplementary risk predictor of GC by serving as a marker of past infection (23, 24). There is evidence that antibodies to CagA can persist longer after the eradication of H. pylori infection and estimates of the association between exposure to H. pylori and GC were highly increased when CagA serostatus was considered combined with the screening ELISA (25). To our knowledge, the present study is the first epidemiologic study on GC risk simultaneously investigating antibodies to 15 different H. pylori proteins using multiplex serology. In our previous study, significant associations between seropositivities for all these 15 specific antibodies and CAG were observed, and HcpC and GroEL were identified as new independent virulence factors (16). Possibly due to the different persistence periods of H. pylori antibodies after disease-related clearance of infection and the smaller sample size of patients with GC, in the present study, only 8 of the 15 antibodies were significantly associated with GC and noncardia GC. The seropositivities of CagA and GroEL were identified as independent predictors of GC.

Our results on CagA are consistent with previous studies showing that the seropositivity of CagA predicts higher risk of GC compared with screening ELISA (8). There are two plausible explanations for this finding. First, the difference in seropositivity of CagA (75.6%) and of the screening ELISA (66.1%) in GC cases suggests that CagA may have a much stronger antigenicity compared with some other immunogenic proteins, and thus, antibodies against it persist longer after clearance of the infection (23–26). Second, as shown in Table 6, lower proportions of CagA-positive strains were observed among H. pylori–infected controls compared with H. pylori–infected GC cases. This observation is consistent with our findings with respect to CAG (16) and other previous studies (27–30), in which CagA was suggested to be a virulence factor and people infected with CagA-positive strains were more likely to develop gastric diseases.

The findings for GroEL are of particular interest, given that seropositivity for this protein, which is widely expressed in most H. pylori strains, was even higher than the seropositivity for CagA (31, 32). GroEL belongs to the family of molecular chaperones which are required for the proper folding of many proteins in the bacteria (33). Additionally, it was reported that GroEL may play a role in gastrointestinal homeostasis due to its ability to bind to components of the gastrointestinal mucosa and to aggregate H. pylori (34–36). In our analyses, a similar seropositivity of the screening ELISA (53.3%) and GroEL (56.9%) was observed in controls, but a much higher positivity for GroEL (83.7%) was seen in GC cases compared with the screening ELISA (66.1%). These patterns suggest that antibodies against GroEL might persist longer after disease-related loss of H. pylori infection. GroEL may thus be a particularly suitable marker of either current or past infection, and may thus be particularly helpful to overcome the underestimation of H. pylori–related GC risk due to disease-related clearance of infection. However, the possibility of cross-reactivity causing false-positive results for GroEL needs to be kept in mind because chaperonin proteins are highly conserved among bacterial species (37). Although an analysis of 180 animal sera with defined serostatus for H. typhlonius, H. bilis, and H. hepaticus infection did not indicate any serologic cross-reactivity with antibodies directed against H. pylori proteins analyzed by multiplex serology (16).

In the interpretation of our data, some limitations have to be considered. First, due to the case-control study design, the possibility of disease-associated changes in H. pylori infection markers has to be taken into account. In particular, disease-associated clearance of H. pylori infection and subsequent loss of antibodies are of concern. Although the latter could not be observed directly, they are strongly suggested by the observed strong positive dose-response relationship between the number of assessed risk markers and GC risk in otherwise seronegative participants. On the other hand, the possibility of disease development–related activation of H. pylori cannot be excluded. Second, for the majority of cases, blood samples were collected after starting treatment (operation and/or chemotherapy). The associations of specific antibodies with GC might therefore have been underestimated due to treatment-related loss of H. pylori infection. However, results were very similar in additional sensitivity analyses conducted among subgroups of cases whose blood samples were taken before or after the initiation of therapy (data not shown). Third, despite the overall large sample size of the study, restricted case numbers limited the power and precision of assessments of associations.

In conclusion, seropositivity for 7 of the 15 H. pylori–specific antibodies measured by multiplex serology were significantly associated with GC in this study population, and more pronounced associations were observed for noncardia GC. CagA and GroEL were found to be independent predictors of GC risk. To our knowledge, for the first time, our studies provide evidence for the seropositivity of GroEL, which may be a very useful marker for both current and past H. pylori infection, to be highly predictive of both CAG and GC. Furthermore, ideally longitudinal studies with repeat serologic measurements and large sample sizes are needed to confirm and further elucidate the role of novel markers for enhanced quantification of H. pylori–related GC risks.

No potential conflicts of interest were disclosed.

Grant support: The ESTHER study baseline examination was supported by a grant from the Baden-Wuerttemberg State Ministry of Science, Research and Arts. The VERDI study was supported by the German Cancer Foundation (Deutsche Krebshilfe project 70-1816). The work of L. Gao was supported by a scholarship from the German Academic Exchange Service (DAAD).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.