In 1994, the International Agency for Research on Cancer classified H. pylori as a class I human carcinogen.1 A large number of epidemiologic studies have further characterized the association by select tumor (e.g., subsite, histologic type, tumor stage) and demographic characteristics (e.g., gender, age at diagnosis), although the role of potential confounders (e.g., smoking) has been considered in only some studies.2, 3, 4, 5, 6, 7 The role of infection of H. pylori, particularly more virulent strains that possess cytotoxin-associated gene A (CagA strains), in the etiology of junctional (gastric cardia and esophageal adenocarcinomas) cancers remains controversial.4, 7, 8 The purpose of our study was to investigate the subsite-specific risks of H. pylori and CagA seropositivity in a well-characterized case-control study of esophageal and gastric adenocarcinomas in Los Angeles County.9 In 3 recent studies, the risk of distal gastric cancer in association with H. pylori infection appeared substantially stronger in smokers than in nonsmokers;10, 11, 12 both H. pylori and smoking conferred independent effects on risk in 2 studies.11, 12 The effects of H. pylori and CagA seropositivity were examined separately for never, former and current smokers in this study population.
Infection with Helicobacter pylori (H. pylori), especially CagA+ strains, has been associated with an increased risk of noncardia gastric adenocarcinoma. The relationship with junctional cancer (adenocarcinomas of the esophagus and gastric cardia combined) has not been adequately investigated, although some studies have reported a reduced risk associated with H. pylori and CagA seroseropositivity. We investigated this question in a subset of cases and controls from a recently completed, large population-based case-control study of gastric and esophageal adenocarcinomas in Los Angeles County. Using established antigen-specific ELISAs, serum IgG antibodies to H. pylori whole-cell antigens (Helico-G) and CagA were measured in population controls (n = 356) and patients with incident esophageal adenocarcinoma (n = 80), gastric cardia cancer (n = 87) or distal gastric cancers (noncardia gastric adenocarcinoma) (n = 127). After controlling for demographic characteristics (age, gender, race, birthplace, education), smoking and body mass index, seropositivity for H. pylori was associated with a statistically significant increased risk of distal gastric cancer (adjusted odds ratio [OR] = 1.85, 95% confidence interval [CI] = 1.03, 3.32) but the risk of junctional cancer was not increased (adjusted OR = 1.26, 95% CI = 0.82, 1.94). The risk of junctional cancer was not changed when CagA and H. pylori were both considered, but the risk of distal gastric cancer was further increased. Subjects who were seropositive for both CagA and H. pylori compared to those who were seronegative for H. pylori showed a risk of 2.20 (95% CI = 1.13, 4.26) for distal gastric cancer and 0.86 (95% CI = 0.47, 1.59) for junctional cancer. Although tests for interaction between smoking and H. pylori were not statistically significant for junctional or distal gastric cancers, risk for both tumor types tended to be higher among current smokers who were also H. pylori seropositive. In conclusion, we find no evidence that infection with CagA+ strains of H. pylori reduces risk of esophageal and gastric cardia adenocarcinoma in this population. Our findings confirm the positive association between risk of distal gastric cancer and infection with H. pylori infection, especially CagA+ strains. © 2002 Wiley-Liss, Inc.
MATERIAL AND METHODS
Data for this analysis were derived from a population-based case-control study of gastric and esophageal adenocarcinomas conducted in Los Angeles County from 1993–1999.9 In brief, cancer patients were identified by the Los Angeles County Cancer Surveillance Program (CSP), the population-based cancer registry covering more than 9 million residents of Los Angeles County. Men and women who were aged 30–74 years when diagnosed with histologically confirmed distal gastric adenocarcinomas from 1992–1994 or esophageal and gastric cardia adenocarcinomas from 1992–1997 were considered eligible. We did not conduct an independent review of the subsite of the gastric cancer or the histologic classification but relied on the information in pathology reports collected by the CSP to document each cancer. Junctional cancers included esophageal (C15.0–C15.9) and gastric cardia (C16.0) adenocarcinomas combined and noncardia gastric cancers were grouped as distal gastric cancers.
Control subjects were individually selected from the neighborhood of residence of the case patient to whom they were matched on gender, race and date of birth (± 5 years). Control subjects must not have had a diagnosis of gastric or esophageal cancer. A total of 942 cases (222 esophageal adenocarcinomas, 277 gastric cardia adenocarcinomas and 443 distal gastric adenocarcinomas) and 1,356 control subjects were interviewed in the parent case-control study. All of the control and 671 of case interviews were conducted with the subjects themselves. All participants were interviewed using a structured questionnaire that asked about lifetime smoking habits (i.e., cigarettes, pipes, cigars, chewing tobacco), lifetime use of all types of alcoholic beverages (i.e., beer, wine, hard liquor), body size characteristics (i.e., height and weight at 20 and 40 years and 1 year before diagnosis for cases or comparable date for controls), lifetime occupational history and usual adult diet history. In addition, information was collected on personal and family history of cancer, use of selected medications and history of select nonmalignant conditions of the gastrointestinal tract that were diagnosed by a physician.
Living self-respondent case patients whose condition permitted and control subjects were asked to donate a blood specimen at the completion of the interview. The specimens were returned to the laboratory on ice, processed and then stored in −70°C freezers at the University of Southern California Norris Comprehensive Cancer Center until they were shipped in sealed containers on dry ice to one of us (PH) in the United Kingdom. To minimize bias in case-control comparisons due to any temporal changes in laboratory results, each analytic batch of specimens represented case patients and control subjects in the same proportion throughout. The laboratory personnel conducting these assays were blinded as to the case-control status and all demographic characteristics such as race, age and gender of the samples.
H. pylori and CagA serum IgG antibodies were measured on 294 case patients (80 esophageal, 87 gastric cardia and 127 distal gastric adenocarcinomas) (blood specimens of 8 esophageal, 17 gastric cardia and 20 distal gastric adenocarcinomas were not tested because the specimens were collected after the completion of H. pylori testing). H. pylori and CagA serum IgG antibodies were measured on 356 participants, approximately a 50% random sample of controls who donated blood specimens (n = 725). Measurements were conducted in the laboratories of P. Hawtin and J. Crabtree using published methods. The Helico-G kit, a validated enzyme-linked immunosorbent assay (ELISA), with sensitivity and specificity levels of 94.0% and 85.5%, respectively, was used to measure H. pylori IgG antibodies.13 Each sample was analyzed in duplicate after 1:100 dilution. Borderline results (i.e., ELISA units from 0.7–1.0) were considered H. pylori positive as analyses that excluded borderline results yielded essentially the same associations as we have presented below (6 esophageal adenocarcinoma, 10 gastric cardia cancers, 23 distal gastric cancers and 41 controls were classified as borderline). Serum samples were assayed for CagA IgG antibodies by ELISA using a purified recombinant fragment of CagA (kindly provided by A. Covacci, Chiron Vaccines, Siena, Italy) as described previously.14 CagA seropositivity was determined by reference to a standard curve of CagA-positive serum assayed on each plate. The cutoff used to determine CagA seropositivity was the mean + 2 standard deviations from a series of 30 serum samples previously shown to be CagA negative by Western blot.15 This ELISA has been shown to have a sensitivity of 95% and a specificity of 97% in noncancer patients (data not shown). Subjects were first classified as H. pylori positive (HP+) or H. pylori negative (HP−) and among those who were HP+, as CagA positive (CagA+) or CagA negative (CagA−). Because some investigators16, 17, 18, 19 have identified a small group of subjects who tested negative for H. pylori but positive for CagA, we also classified participants into 4 groups based on both H. pylori and CagA serologic results: HP-CagA−, HP+CagA−, HP-CagA− and HP-CagA+.
We used χ2 tests to assess the influence of demographic (e.g., age, sex, race, birthplace, educational level) and lifestyle factors (e.g., body mass index [weight in kg divided by the square of height in meters], smoking and alcohol on the prevalences of H. pylori and CagA seropositivity among control subjects. Unconditional logistic regression methods were used to calculate odds ratios (ORs), as estimates of the relative risk, and corresponding 95% confidence intervals (CIs) for junctional cancers (esophageal and gastric cardia adenocarcinomas) and distal gastric cancers in relation to H. pylori and CagA status. All unconditional models included age (<40, 40–49, 50–59, 60–69, 70+), sex (male/female), race (White, African-, Latino, Asian-American), birthplace (U.S. born, non-U.S. born), education (less than high school, high school, some college, college or more), cigarette smoking status (never, former, current smoker) and body mass index (referred to as adjusted ORs hereafter). In the parent case-control study, cigarette smoking was a significant risk factor for each of the 3 tumor types whereas high body mass was a risk factor primarily for junctional cancers.9 All reported significance levels (p-values) are 2-sided.
The participants with measured H. pylori and CagA status were very similar to the case patients and control subjects in the parent case-control study in terms of age, sex, race, birthplace, education, and smoking status (Table I). Specifically, among the cases, self-respondent cases with H. pylori measurements did not differ from self-respondent cases without H. pylori measurements. Not surprisingly, cases for whom next-of-kin interviews were conducted had more advanced stage disease than self-respondent cases. Among the controls, those who did not donate blood specimens did not differ from controls who donated blood specimens. In addition, the controls for whom H. pylori were tested did not differ from those controls who donated bloods but were not selected for H. pylori testing.
|Esophageal adenocarcinomas||Gastric cardia||Distal gastric||Controls|
|NOK1n = 66||Self interview, no HP/CagA n = 76||Self interview, HP/CagA tested3n = 80||NOK1n = 85||Self interview, no HP/CagA2n = 105||Self interview, HP/CagA tested3n = 87||NOK1n = 120||Self interview, no HP/CagA2n = 196||Self interview, HP/CagA tested3n = 127||No blood n = 631||Blood donated, no HP/CagA4n = 369||Blood donated, HP/CagA tested n = 356|
|< High school||25.8||21.1||18.8||29.4||15.2||13.8||50.0||36.7||41.7||19.2||16.8||19.4|
|College graduate or higher||19.7||28.9||30.0||17.6||27.6||29.9||13.3||19.4||18.9||34.1||34.7||32.9|
Table II shows the prevalences of H. pylori and CagA seropositivity by various demographic and lifestyle characteristics among control subjects. The prevalence of H. pylori seropositivity was similar in men and women and in different age groups but the seropositivity rate was higher among non-whites than whites (p = 0.0002), individuals born outside of the U.S. compared to those who were U.S.-born (p = 0.0004) and those who had less than high school education relative to those with more education (p = 0.01). H. pylori seropositivity rates did not differ by smoking status, drinking habits or current body mass index. Similar patterns of associations were observed between CagA seropositivity and demographic and lifestyle factors (Table II).
|Characteristics||n||% H. pylori +||% CagA +*||H. pylori (HP) and CagA combined|
|% HP− CagA−||% HP+ CagA−||% HP− CagA+||% HP+ CagA+|
|< High school||69||81.2||51.8||13.0||39.1||5.8||42.0|
|Quartile 1 (lowest)||107||58.9||33.3||34.6||39.3||6.5||19.6|
|Quartile 4 (highest)||85||67.1||43.9||28.2||37.6||4.7||29.4|
The control subjects (64.6%) and case patients with junctional cancers (64.7%) showed comparable H. pylori seropositivity rates but case patients with distal gastric cancer displayed higher rates (81.1%) (Table III). The adjusted OR for distal gastric cancer associated with H. pylori seropositivity was 1.85 (95% CI = 1.03–3.32), whereas the relative risk of junctional gastric cancer was 1.26 (95% CI = 0.82–1.94). Among subjects who were seropositive for H. pylori, CagA seropositivity was lower among case patients with junctional cancers (27 of 108; 25.0%) than control subjects (87 of 230; 37.8%), whereas distal gastric cancer patients showed higher CagA seropositivity (55 of 103; 53.4%) (Table III). The risk of distal gastric cancer was statistically significantly elevated in association with seropositivity of both H. pylori and CagA compared to those who were H. pylori seronegative (adjusted OR = 2.20, 95% CI = 1.13–4.26). In contrast, the relative risk of junctional cancer was not influenced by seropositivity of H. pylori and CagA (adjusted OR = 0.86, 95% CI = 0.47–1.59).
|No. of controls1||Esophageal adenocarcinoma||Gastric cardia||Junctional cancer||Distal gastric cancer|
|No. of cases1||Adjusted OR2 (95% CI)||No. of cases1||Adjusted OR2 (95% CI)||Adjusted OR2 (95% CI)||No. of cases1||Adjusted OR2 (95% CI)|
|H. pylori (HP)|
|Positive||230||49||1.01 (0.58–1.77)||59||1.43 (0.83–2.45)||1.26 (0.82–1.94)||103||1.85 (1.03–3.32)|
|HP+ CagA−||143||34||1.14 (0.63–2.07)||47||1.70 (0.97–2.98)||1.43 (0.91–2.24)||48||1.62 (0.86–3.06)|
|HP+ CagA+||87||15||0.82 (0.37–1.81)||12||0.80 (0.36–1.81)||0.86 (0.47–1.59)||55||2.20 (1.13–4.26)|
|HP+ CagA−||143||34||1.14 (0.62–2.12)||47||1.96 (1.07–3.59)||1.52 (0.95–2.43)||48||1.88 (0.92–3.83)|
|HP− CagA+||19||3||1.03 (0.25–4.14)||6||2.19 (0.74–6.43)||1.47 (0.58–3.70)||7||1.87 (0.57–6.18)|
|HP+ CagA+||87||15||0.82 (0.36–1.84)||12||0.92 (0.40–2.13)||0.91 (0.49–1.72)||55||2.58 (1.22–5.43)|
Table III also shows the risk patterns separately for esophageal (n = 80) and gastric cardia (n = 87) adenocarcinomas. We found no evidence of any influence of H. pylori or CagA status on the risk of esophageal adenocarcinoma as all the point risk estimates were around 1.00. The risk of gastric cardia cancer was elevated among H. pylori seropositive subjects relative to those who were seronegative (adjusted OR = 1.43, 95% CI = 0.83–2.45), but this elevated risk was observed primarily among individuals who were CagA seronegative (adjusted OR = 1.70, 95% CI = 0.97–2.98). The estimated relative risk was below 1.0 for individuals who were positive for both H. pylori and CagA (adjusted OR = 0.80, 95% CI = 0.36–1.81) (Table III).
We repeated analyses of junctional cancers restricted to white subjects (139 cases and 221 controls). In this subgroup, the relative risk of junctional cancers associated with H. pylori seropositivity was 1.26 (95% CI = 0.80–1.97). Compared to individuals who were seronegative for H. pylori, those who were H. pylori seropositive and CagA seronegative showed a relative risk of 1.36 (95% CI = 0.84–2.20), whereas those who were seropositive for both H. pylori and CagA showed a relative risk of 0.88 (95% CI = 0.43–1.81).
A small proportion of control subjects (15.1%, 19 of 126) who tested seronegative for H. pylori were found to be seropositive for CagA. When we examined risk patterns using seronegatives for both H. pylori and CagA as the baseline comparison group, the risk of esophageal adenocarcinoma remained close to unity for all comparisons. Subjects who were seropositive for H. pylori but seronegative for CagA showed a statistically significant increased risk for gastric cardia cancer (adjusted OR = 1.96, 95% CI = 1.07–3.59), while those who were seropositive for both markers showed a statistically significant increased risk for distal gastric cancer (adjusted OR = 2.58, 95% CI = 1.22–5.43).
Table IV shows the risk of distal and junctional gastric cancers associated with the joint effects of smoking and status of H. pylori and CagA. Although the global tests for interaction between H. pylori/CagA and smoking status were not statistically significant for each of the 3 tumor types, the influence of H. pylori/CagA seropositivity on risk varied by smoking status. The lack of any effect of H. pylori/CagA status on risk of esophageal adenocarcinoma was observed in never, former and current smokers. For distal gastric cancer, the risk associated with seropositivity for both H. pylori and CagA was comparable for never (adjusted OR = 1.77, 95% CI = 0.68–4.62), former (adjusted OR = 2.19, 95% CI = 0.83–5.80) and current smokers (adjusted OR = 1.69; 95% CI = 0.51–4.62), but current smokers who were H. pylori seropositive but CagA seronegative showed the highest risk (adjusted OR = 3.82, 95% CI = 1.32–11.04). Compared to never smokers who were H. pylori seronegative, current smokers who were H. pylori seropositive irrespective of CagA status showed significantly elevated risk of distal (adjusted OR = 2.70, 95% CI = 1.03–7.11) and junctional cancer (adjusted OR = 2.43, 95% CI = 1.12–5.26).
|Smoking status||H. pylori (HP) and CagA status||Esophageal adenocarcinoma||Gastric cardia||Junctional cancer||Distal gastric cancer|
|P interaction, smoking and HP status||2 degrees of freedom||0.66||0.41||0.84||0.48|
|P interaction, smoking and HP/CagA status||4 degrees of freedom||0.67||0.70||0.93||0.12|
To our knowledge, this is the largest population-based case-control study of gastric and esophageal adenocarcinoma and association with H. pylori infection. Before discussing our findings, it is important to mention some of the limitations of our study and the extent to which the results may have been affected. First, H. pylori infection status was only assessed among some 87% of case participants and a 50% random sample of control subjects who donated blood specimens. As shown above (Table I), self-respondent cases with H. pylori measurement did not differ from self-respondent cases without H. pylori measurement. With the exception for stage of cancer, self-respondent cases did not differ from cases for whom next-of-kin interviews were conducted. Control subjects who did not donate blood specimens were also similar to those who donated specimens. Thus, there is some reassurance in that cases and controls included in this analysis on H. pylori are comparable to those not included although there is no absolute guarantee that some unmeasured difference exists that could impact the generalizability of our results.
The patterns of prevalence of H. pylori infection among control subjects (Table II) are also consistent with the known descriptive epidemiology of H. pylori in terms of associations with various demographic and lifestyle characteristics.20 Second, blood specimens from cases were obtained after their cancer diagnosis. Since loss of infection occurs in the presence of gastric atrophy and malignancy, our observed association between distal gastric cancer and H. pylori infection is likely an underestimation of the true effect. Loss of infection is less likely to occur in the gastric cardia and lower esophagus and thus our risk estimates should not have been biased.4 Third, we did not conduct an independent review of the tumor type and location classification but relied on histology and subsite classification that were routinely reported to our tumor registry. To the extent that some misclassification of tumor type exists,21 any differences in risk patterns by tumor subsite should be even stronger in the absence of such misclassification. It should be noted that only 1 previous study22, 23 on H. pylori infection included data separately for esoghageal, gastric cardia and distal gastric adenocarcinomas but the numbers of each tumor type were small. Fourth, the accuracy of our assays of H. pylori and CagA status is an important consideration. Whilst it is ideal to have validation data from every population, or possibly every ethnic group, in which a diagnostic test is used, we and most other investigators do not have that information for either H. pylori or CagA but have to rely on validation results obtained in populations other than the one under study. The CagA ELISA test we used in our study has been validated in multiple populations in the United Kingdom, Poland and Japan14, 24 and has been found to be very accurate when compared to Western blotting results. Although the joint prevalence of H. pylori and CagA seropositivity was lower among controls in Los Angeles than in some studies,7, 25 our results are similar to those reported in a large U.S. study using resources from the NHANES III study26 and a recent study from Germany.12 Expectedly, some subjects showed H. pylori antibody levels that were “borderline,” i.e., they displayed values that were in between the “negative” and the “positive” cutpoints. Because the assay gives a continuous measure, we believe excluding such subjects may be potentially more misleading than retaining them. Instead, we conducted a sensitivity analysis and considered cutoff points from 7.0–10.0 and found that the risk estimates for junctional cancers and distal cancers remained largely similar.
Our findings on H. pylori and distal gastric cancers are compatible with the published literature, although the magnitude of risk (adjusted OR = 1.85) we observed is lower than the pooled risk estimate (OR = 3.0) calculated for noncardia gastric cancers in a meta-analysis of prospective studies.4 Our results support other studies conducted in eastern19, 27 and western populations,12, 25, 28, 29, 30 demonstrating a further increase in the risk of distal gastric cancers of the intestinal type in association with seropositivity of CagA. Similar to several recent studies,10, 11, 12 the effect of H. pylori infection on risk of distal cancers appeared stronger among current smokers in our study. However, the numbers within each category of H. pylori and smoking status were modest in this and previous studies. Larger sample sizes are needed to determine with certainty the role of H. pylori/CagA status and smoking for each of the tumor type of interest.
In contrast, we found little evidence of a role of H. pylori or CagA on the risk of junctional cancers, particularly esophageal adenocarcinoma. Studies that have examined the role of H. pylori/CagA in the etiology of gastroesophageal diseases have yielded inconsistent results. Studies from western populations have found variable H. pylori-associated risk estimates for gastric cardia and/or esophageal adenocarcinoma; ORs were above 1.0 in 5 studies,31, 32, 33, 34, 35 below 0.8 in 4 studies7, 23, 36, 37 and between 0.9 and 1.0 in 3 other studies.22, 38, 39 The reduced risk was statistically significant in 3 studies;7, 23, 36 CagA was measured in one of these studies.7 In contrast, among studies conducted in China and Japan, all19, 27, 40, 41, 42 but one43 have found elevated risks (RR ranged from 1.3–11.3) in association with H. pylori infection. The result was statistically significant in one study.19 Small sample size has been a serious limitation of most studies in the literature; only 2 of the 19 studies with data on gastric cardia/junctional cancers had as many as 100 cases.7, 19 Most of the other studies had fewer than 30 cases22, 23, 27, 31, 33, 35, 37, 38, 40, 41 and 6 had from 30–55 cases.18, 34, 36, 39, 42, 43 In addition, some studies included only cardia cancers or those located in the upper one-third of the stomach;18, 19, 22, 33, 35, 37, 39, 40, 41, 43 others included both esophageal and gastric cardia adenocarcinomas7, 36, 38 and only one included esophageal adenocarcinoma.23
In conclusion, in this population-based study of gastric and esophageal adenocarcinoma in Los Angeles County, our results from a cross-sectional analysis do not support the hypothesis that H. pylori, especially CagA seropositive strains, has contributed to the rising incidence of esophageal/gastric cardia adenocarcinomas in western countries. Studies with adequate sample sizes of confirmed gastric cardia and esophageal adenocarcinomas will be needed to definitively settle the controversy over the role of H. pylori and junctional cancers. The interrelated influences of H. pylori and CagA and history of smoking on the risk of these tumor types also warrant further investigation.
Incident cancer cases for our study were collected by the USC Cancer Surveillance Program (CSP), which is supported under subcontract by the California Public Health Foundation, subcontract 050-F-8709, which is supported by the California Department of Health Services as part of its statewide cancer reporting program, mandated by Health and Safety Code Sections 210 and 211.3. The CSP is also part of the National Cancer Institute's Division of Cancer Prevention and Control Surveillance, Epidemiology, and End Results Program, under contract number N01CN25403. Work in the laboratory of JEC is supported by Yorkshire Cancer Research and the European Commission (contract ICA4-CT-1999-10010). We are grateful to all the study participants for their contributions. We thank Ms. A. Fung, Ms. I. Rivera, Mr. T. Stirton and Ms. J. Yashiki for their help with data collection and Ms. D. Sosnowska for her help with the laboratory measurements.