Helicobacter pylori Protein–Specific Antibodies and Risk of Colorectal Cancer

This article is featured in Highlights of This Issue, p. 1925

Infection with Helicobacter pylori, which induces chronic inflammation in the gastric mucosa, is the strongest known risk factor for gastric cancer. Recent studies suggest that H. pylori infection may also increase the risk of colorectal cancer, and although these findings have been inconsistent, two recent meta-analyses found significant 40% to 50% increased odds for colorectal cancer among individuals with evidence of a current or past H. pylori infection (1, 2). A primary mechanism by which H. pylori infection might increase the risk of colorectal cancer is via an increase in the release of gastrin (3), a peptide hormone whose main role is to stimulate gastric acid secretion, but which also functions as a mitogen (4). Chronic gastritis that results from persistent H. pylori infection is associated with hypergastrinemia, as H. pylori infection-related gastritis reduces acid secretion, which in a negative feedback loop induces the production of high-gastrin levels (3). H. pylori infection could also increase risk of colorectal cancer through mechanisms related to those secondary to chronic infection and/or alteration of the bacterial flora that comprise the gastrointestinal microenvironment (4–6). Other evidence that colorectal and gastric cancers may share some aspects of a common etiology include the facts that colorectal cancer has consistently been found to be the most common synchronous cancer among patients with gastric cancer (7, 8); second primary gastric cancers are increased following colorectal cancer diagnosis (9); and, correspondingly, second primary colorectal cancers are increased following gastric cancer diagnosis (10).

The majority of the previously published studies examining the association between H. pylori and colorectal cancer risk did not take into account H. pylori strain type. Because more than 50% population of the world is infected with the bacteria and only a small percentage of those individuals develop cancer, it is important that both host and bacterial factors are considered when assessing associations with disease. Furthermore, H. pylori has colonized the stomach of humans for more than 50,000 years and has evolved over time to become highly genetically diverse (11). Although the authors of the most recent study of the association of H. pylori and colorectal cancer risk did stratify by presence of the H. pylori protein and known gastric cancer virulence factor cytotoxin-associated antigen (CagA), the association found with CagA-positive strains was not substantially stronger than that with H. pylori seroprevalence alone (significant ORs of 1.20 and 1.18, respectively; ref. 12). In addition, this case–control study collected blood samples at hospitalization for cancer treatment and, thus, the measurement of H. pylori status among cases may have been hindered by presence of the tumor and/or the initiation of cancer treatment. Finally, as the authors discuss themselves, they did not inspect the association of other H. pylori virulence factors, such as VacA, HcpC, and GroEL, with colorectal cancer risk. Only two other studies to our knowledge have investigated CagA status and colorectal cancer risk; in the Alpha-Tocopherol, Beta-Carotene Study cohort of Finnish male smokers, no association was found with H. pylori seropositivity, and there was only a suggestion of an increase in risk for colorectal cancer among persons seropositive to CagA [OR, 1.17; 95% confidence interval (CI), 0.74–1.84; ref. 13], whereas in a small hospital-based case–control study in Israel composed of 67 cases of colorectal adenocarcinoma and 45 individuals hospitalized for transesophageal echocardiography as controls, CagA seropositivity was linked to a more than 10-fold increase in risk for colorectal cancer (OR, 10.6; 95% CI, 2.7–41.3; ref. 14).

In addition, none of the published studies of H. pylori and colorectal cancer risk have included a substantial number of African Americans, the racial/ethnic group with the highest incidence of and mortality from colorectal cancer in the United States (15). We recently found an exceptionally high prevalence of H. pylori infection among African American participants in the Southern Community Cohort Study (SCCS), composed of individuals recruited from community health centers (CHC) in the southeastern United States that serve primarily low-income and uninsured persons (16). In fact, both African Americans and Whites in the SCCS have seroprevalences that rival developing countries (89% and 69%, respectively) and African Americans are significantly more likely than Whites to be infected with 8 different H. pylori proteins suspected to be gastric cancer-associated virulence constituents, including CagA (OR, 6.4; 95% CI, 4.5–9.1) and VacA (OR, 2.3; 95% CI, 1.5–3.5; ref. 16).

The present study sought to assess the association of H. pylori and colorectal cancer risk by including H. pylori protein–specific analyses of the bacteria in a prospective, nested case–control design among an understudied population with a high prevalence of H. pylori, the SCCS.

As previously described in detail (17), the SCCS is a prospective cohort study that enrolled approximately 86,000 men and women from 12 southeastern states from 2002 to 2009. Participants, ages 40 to 79, were primarily (86%) recruited from CHCs, institutions offering basic health and preventative services mainly to the medically uninsured and underinsured as well as by mail (14%). Individuals recruited from CHCs participated in in-person, comprehensive computer-assisted interviews that obtained information on demographic and lifestyle factors as well as regular diet, personal and family medical history, and health services use. The SCCS was reviewed and approved by Institutional Review Boards at Vanderbilt University and Meharry Medical College (Nashville, TN). Written informed consent was obtained from all study participants. The individuals recruited by mail completed an article version of the same baseline questionnaire. Individuals recruited at the CHCs were also asked to donate venous blood samples (20 mL), which were then immediately refrigerated. The samples were then shipped overnight to Vanderbilt University to be centrifuged the next day, and then stored at −80°C. For the assay of H. pylori protein antibodies as examined in the present project, a serum sample for each study subject was aliquoted into 50 μL portions.

Incident colorectal cancer cases (International Classification of Diseases–Oncology, ICD–O-3; codes C180–189, C199, and C209) diagnosed after entry into the SCCS were identified through linkage with state cancer registries from the 12-state study area and/or from National Death Index mortality records. A total of 188 colorectal cancer cases were identified as occurring after biospecimen collection and before the end of 2011, among those SCCS participants recruited at CHCs who donated a blood sample.

For each case, 2 controls were chosen, matched on age (±2 years), race (African American, White, or other), sex, menopausal status (for women), CHC site, date of sample collection (±6 months), and availability of serum. Matching with these criteria to identify 2 controls was successful for 87% of cases. For 10% of cases, the requirement for the control being recruited from the same CHC as the case was relaxed to a CHC in the same state and for the remaining 3% of cases, the requirement for matching on CHC location was dropped altogether. Of 376 controls chosen in this manner, 3 were excluded because of missing baseline data, 1 was missing serum, 1 was a duplicate, and 1 was not matched, resulting in 370 controls and 188 cases for the present study.

Because previously described, H. pylori multiplex serology uses a glutathione S-transferase capture immunosorbent assay combined with fluorescent bead technology (Luminex) to detect antibody levels to a constellation of IgA, IgM, and IgG values to 15 H. pylori proteins (UreA, Catalase, GroEL, NapA, CagA, CAgM, Cagδ, HP0231, VacA, HpaA, Cad, HyuA, Omp, HcpC, and HP0305; refs. 18, 19). These proteins were chosen for the assay based on known surface exposure and immunogenicity in two-dimensional immunoblot analyses (UreA, Catalase, NapA, CagA, HP231, VacA, and HpaA), serologic association with gastric cancer (GroEl, Cad, HyuA, Omp, and HcpC) and/or gastric ulcer (HP305 and CagM), and specific recognition in H. pylori-positive sera (Cagδ and CagM). All sera are analyzed once within a single assay day. Antigen-specific cutoff points were calculated [mean median reporter fluorescence intensity (MFI) plus three times SD, excluding positive outliers] in 17 H. pylori-negative sera previously classified for H. pylori status run within the same experiment. Defining H. pylori seropositivity as reactivity with at least 4 proteins has shown good agreement (κ = 0.70) with commercial serologic assay, resulting in 89% sensitivity and 82% specificity (18). For quality control, 18 samples from one-pooled sample were included in the assay. The determination of seropositivity for all of the H. pylori proteins was strongly consistent and the interassay coefficient of variation (CV) ranged from 10% to 21% (notably, the CagA, CV was 13% and the VacA, CV was 15%).

To assess the individual associations of seropositivity to each of the 15 H. pylori proteins with colorectal cancer risk, conditional logistic regression was used in separate models for each protein to determine ORs and 95% CIs. To determine the possibility of confounding by baseline characteristics, Pearson χ² test was used to assess differences between cases and controls, between individuals seropositive and seronegative for H. pylori, and between individuals seropositive and seronegative for VacA. No baseline characteristic was found to be associated with both colorectal cancer risk and H. pylori or VacA status. Smoking status was the only baseline factor associated with colorectal cancer risk, but it was not associated with H. pylori or VacA status. Potential confounders associated with H. pylori or VacA status, including education, income, regular use of aspirin, vitamin supplement use, and colorectal cancer screening history, were not associated with colorectal cancer risk in this population, and like smoking status, inclusion of these variables in the model did not alter the main effects by 10% or more. Race was strongly associated with H. pylori status, but this demographic characteristic could not confound the association as controls were matched to cases by race, and conditional logistic regression ensured that the association between H. pylori and colorectal cancer compared risk among race-concordant sets.

To examine a potential dose–response relationship between the level of antibodies and colorectal cancer risk, we created quartiles of antibody levels (MFI) for each of the 15 H. pylori proteins assessed on the basis of the distribution among control subjects. Conditional logistic regression was used again to determine the association of increasing quartile of H. pylori protein–specific antibody level and colorectal cancer risk with indicator variables representing antibody levels, using the lowest quartile as the reference category. To test for a linear trend across antibody quartiles of each protein, a continuous variable was created with the values of 0, 1, 2, and 3 for the 4 quartiles. The Bonferroni correction for multiple testing was considered but rejected as the 15 H. pylori proteins assessed are significantly correlated with one another.

To further explore the relationship between H. pylori status and risk of colorectal cancer, we assessed the association in separate models for colon cancer (C180–189) and rectal cancer (C199 and C209), for cancer of the right/ascending colon (C180–184) and cancer of the left/descending colon (C185–187), for cancers diagnosed at local, regional, and distant stages, individually, and for cancers by age at diagnosis (separately, in tertiles and dichotomously comparing <60- to 60+-year-olds). For these models, we again conducted conditional logistic regression, and for analyses by stage adjusted for a marker of socioeconomic status (high school graduate, yes vs. no) and for a history of colorectal cancer screening (ever vs. never). We examined effect modification of the association of H. pylori status and colon cancer risk through models stratified by race and sex, and assessment of an interaction term using the likelihood ratio test.

Baseline characteristics of colorectal cancer cases and controls were similar in most respects, including markers of socioeconomic status and body mass index (Table 1). Although the great majority of individuals in this population (87.1%) were seropositive for H. pylori (i.e., were positive to at least 4 of the proteins), individuals seronegative for H. pylori were significantly more likely to be White (rather than African American), have obtained a high school or greater education, have a household annual income above $15,000, be a regular aspirin user, and use vitamin supplements. The same proportion of colorectal cancer cases and controls (87%) was seropositive to H. pylori overall, leading to no difference in risk (OR, 1.03; 95% CI, 0.59–1.77).

Seropositivity to 5 of the 15 H. pylori proteins investigated—VacA, HP231, HP305, NapA, and HcpC—was associated with a statistically significant 60% to 80% increase in odds of colorectal cancer risk (Table 2). The association of seropositivity to 9 of the other 10 H. pylori proteins was also in the direction of increased risk, the one exception being CagA (OR, 0.99; 95% CI, 0.64–1.53). When examining the associations with colon cancer alone, the association of these 5 proteins became even stronger, resulting most significantly in a more than 2-fold increase in risk for individuals seropositive to VacA (OR, 2.24; 95% CI, 1.22–4.11). The association between CagA seropositivity and colon cancer moved in the positive direction, although it did not reach statistical significance (OR, 1.26; 95% CI, 0.74–2.16). Overall H. pylori seropositivity was associated with a nonsignificant increase in risk for colon cancer (OR, 1.47; 95% CI, 0.74–2.93). No association was found between overall H. pylori seropositivity or seropositivity to specific H. pylori proteins and rectal cancer risk. Analyses separating cancers of the right colon from those of the left colon revealed that the associations with seropositivity to individual proteins seemed stronger for right-sided tumors than left-sided tumors (Supplementary Table S1). The association of overall H. pylori seropositivity became stronger and close to significance, for cancers of the right colon only (OR, 2.58; 95% CI, 0.94–7.05; P = 0.06).

When examining the association of VacA-positive H. pylori seroprevalence with colorectal cancer in separate models by stage, we found no associations with local cancer (OR, 1.06; 95% CI, 0.48–2.37) or regional cancer (OR, 2.09; 95% CI, 0.88–4.92), but significant excess risk with distant colorectal cancer (OR, 5.67; 95% CI, 1.51–21.37). Similarly, when comparing the association of VacA-positive H. pylori seroprevalence by age at diagnosis, significance was found only for those at the youngest end of the spectrum (diagnosed at <55 years; OR, 3.48; 95% CI, 1.45–8.38), and the association diminished as age at diagnosis increased from 55 to 64 years (OR, 1.42; 95% CI, 0.69–2.93) and 65 years or older (OR, 1.34; 95% CI, 0.60–2.97). Stratification by race and sex in separate models did not reveal effect modification by these factors (data not shown).

A significant positive dose–response association by quartile of antibodies for risk of colorectal cancer was found for 7 of 15 H. pylori proteins assessed (VacA, HP231, HP305, NapA, HcpC, Cad, and GroEL; Table 3). When examining these trends with colon cancer only, the dose–response trends for each of these 7 H. pylori proteins were strengthened (although statistical significance was lessened, because of smaller numbers), statistical significance in trend was reached for an additional protein (UreA), and a suggestion of a dose–response association was found for CagA (P for trend = 0.06). No dose–response associations between antibodies to H. pylori proteins and risk of rectal cancer were found.

In this prospective study of a predominantly low-income population with a high prevalence of H. pylori infection, seropositivity to 5 H. pylori-specific proteins, most notably VacA, was associated with a 60% to 80% increase in risk of colorectal cancer, and this risk increased with increasing quartile of protein-specific antibody levels. The dose–response association of VacA antibody levels was particularly strong with late-stage and early-onset colon cancers. There was no significant association with overall H. pylori seroprevalence or for rectal cancer separately.

VacA is a multifunctional toxin that targets the mitochondria through the formation of pores in the epithelial cell membrane, entering host cells through vacuoles in the cytoplasm, where it induces apoptosis as well as blocks T-cell proliferation and induces cell-cycle arrest (20). In vivo experiments have found that expression of VacA plays an important role in the colonization of the host (21). Although most strains of H. pylori possess the VacA gene and express the associated protein, less than half produce the most active form (22). The association of increased levels of antibodies to VacA with colorectal cancer risk may be related to gastrin, a peptide hormone that induces gastric acid secretion and stimulates protein expression of COX-2 (23). Laboratory studies have found that administration of VacA-positive strains of H. pylori (as well as CagA-positive H. pylori strains) induces an inflammatory response and increases gastrinemia (24–26), and that after H. pylori eradication, serum gastrin levels decrease significantly (27). An association between high-gastrin levels and colorectal cancer risk has been suggested in several studies of human populations (28–30), although the association only reached significance in one study (30). In the most recent study of the gastrin–colorectal cancer association, mean levels of gastrin did not differ between cases and controls, although significantly higher levels were seen among patients with lymph node metastasis compared with patients without (P = 0.03; ref. 29).

With the exception of the present study, no other studies to our knowledge have examined colorectal cancer risk with H. pylori subtypes other than CagA-positive strains. About gastric cancer, serum antibodies to both CagA and VacA have been found to be associated with increased risk of disease as well as a stronger inflammatory response (31). Some studies have even found stronger associations for gastrointestinal disease among individuals with VacA antibodies than those with CagA antibodies (32–34). However, overall, the findings for the association of VacA antibodies and gastrointestinal pathology are inconsistent, with studies that find positive associations with disease (32, 34, 35) and those that find null associations (31, 36).

Although in the current study, we found a null association of CagA with colorectal cancer risk, the association of colon cancer risk with CagA-positive strains (OR, 1.26; 95% CI, 0.74–2.16) is similar to what has been seen in other studies, although it did not reach significance in our analysis. It is possible that differences in risk associated with CagA-positive H. pylori might emerge if one could differentiate CagA isoforms [i.e., phylogeographic origin of the strain and/or Glu-Pro-Ile-Tyr-Ala (EPIYA) motifs], as has been done in studies of gastric cancer (37–39), although this requires the use of gastric specimens, rather than blood samples that are collected in most prospective studies. In addition, the present study found no evidence of an interaction by VacA and CagA strain (data not shown), whereby a VacA–CagA interaction has been suggested in gastric cancer etiology (40).

In the present study, the strongest associations with increased level of VacA antibodies were found for colon cancer and specifically cancer of the right colon as well as for colon cancer diagnosed at earlier ages and presenting at later stage. These results conflict with a case–control study of overall and CagA-positive H. pylori and colorectal cancer risk in Germany that included 1,712 cases and 1,669 controls and found stronger associations with cancer of the left colon and rectum and with early-stage tumors (12). However, VacA status was not assessed and the cases were recruited during their hospital stay or even after surgery, which could affect the measurement of H. pylori antibodies.

Our findings of significant, strong dose–response associations of VacA antibodies with cancer of the right colon and cancers diagnosed at late stage and early onset provoke further questions. Because our findings are new and the interactions observed not specified a priori, and because our sample is of only moderate size, the results need replication in additional prospective studies before causal inferences could be considered. Also, as seropositivity to the 5 H. pylori proteins associated with colorectal cancer risk was highly correlated, we did not have the power to investigate the independent effects of each, although the results suggested that the findings seem to be primarily due to the association of VacA. There are mechanistic pathways, including gastrin production, gastric atrophy, and the expression of COX-2, that might explain causality, but the presence of high levels of VacA antibodies could also be an indicator of a different underlying factor that we have not measured, such as baseline inflammatory state of the host. However, because of the high prevalence of H. pylori in the population we studied, and even in the general U.S. population for whom it is estimated that 53% of African Americans and 22% of Whites are infected with H. pylori (41), the potential for a causal relationship needs to be explored in depth, as should it exist, opportunity for targeted cost-effective cancer prevention through H. pylori eradication therapy in high-risk populations could be explored.

No potential conflicts of interest were disclosed.

Conception and design: M. Epplein, M. Pawlita, R.M. Peek Jr

Development of methodology: A. Michel

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Pawlita, A. Michel, Q. Cai, W.J. Blot

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Epplein, M. Pawlita

Writing, review, and/or revision of the manuscript: M. Epplein, M. Pawlita, A. Michel, R.M. Peek Jr, Q. Cai, W.J. Blot

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Epplein, A. Michel

The Southern Community Cohort Study is funded by a grant from the National Cancer Institute (R01 CA092447). The development of H. pylori multiplex serology was funded in part by the Joint Initiative for Innovation and Research of the German Helmholtz Association. Analysis of the samples in this study was funded by a CTSA award (No. UL1TR000445) to Vanderbilt from the National Center for Advancing Translational Sciences. The serum/plasma sample preparations were conducted at the Survey and Biospecimen Core, which is supported in part by the Vanderbilt-Ingram Cancer Center (P30 CA68485). R.M. Peek Jr is supported by DK R01 58587, CA R01 77955, and CA P01 116087. The project described was supported by the National Center for Research Resources, Grant UL1 RR024975-01, and is now at the National Center for Advancing Translational Sciences, Grant 2 UL1 TR000445-06. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

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.