infection with helicobacter pylori strains - cancer research

(CANCER RESEARCH 55. 2111-2115. May 15. 1995]

Infection with Helicobacter pylori Strains Possessing cagA Is Associated with anIncreased Risk of Developing Adenocarcinoma of the Stomach1

Martin J. Blaser,2 Guillermo I. Perez-Perez, Harry Kleanthous, Timothy L. Cover, Richard M. Peek, P. H. Chyou,

Grant N. Stemmermann, and Abraham NomuraDepartments of Medicine Â¡M.J. B., G. I. P-P., T. L. C.. R. M. P.I and Microbiology and Immunology fM. J. B.Â¡,Vanderbilt University School of Medicine, Nashville, Tennessee37232; Medical Sen-ice, Department of Veterans Affairs Medical Center. Nashville, Tennessee 37212 Â¡M.J. B., T. L CJ: OraVax, Inc.. Cambridge, Massachusetts 02139Â¡H.K.I; and Japan-Hawaii Cancer Study, Honolulu, Hawaii 96X17 Â¡P.H. C., G. N. S., A. N.Â¡

ABSTRACT

To determine whether infection with a Helicobacter pylori strain possessing cagA is associated with an increased risk of development of ade-nocarcinoma of the stomach, we used a nested case-control study based ona cohort of 5443 Japanese-American men in Oahu, Hawaii, who had a

physical examination and a phlebotomy during 1967 to 1970. We matched103 //. priori-infected men who developed gastric cancer during a 21-yearsurveillence period with 103 //. /y/ori-infected men who did not develop

gastric cancer and tested stored serum specimens from patients andcontrols for the presence of serum IgG to the cagA product of H. pyloriusing an ELISA. The serum IgG assay using a recombinant CagA fragment had a sensitivity of 94.4% and a specificity of 92.5% when used in aclinically defined population; serological results were stable for more than7 years. For men with antibodies to CagA, the odds ratio of developinggastric cancer was 1.9 (95% confidence interval, 0.9-4.0); for intestinal

type cancer of the distal stomach, the odds ratio was 2.3 (95% confidenceinterval, 1.0-5.2). Age <72 years and advanced tumor stage at diagnosis

were significantly associated with CagA seropositivity. We conclude thatinfection with a cagA -positive //. pylori strain in comparison with acagA-negative strain somewhat increases the risk for development of

gastric cancer, especially intestinal type affecting the distal stomach.

INTRODUCTION

There is increasing evidence that persistent infection with Helicobacter pylori is a risk factor for the development of gastric adenocar-cinoma (1, 2), especially of the distal stomach (3-5). The evidencecomes mainly from epidemiological investigations (6-9), includingnested case-control studies (4, 5, 10), and molecular and pathological

studies support its biological plausibility (11).However, although H. pylori infection is highly prevalent in pa

tients with gastric cancer, most H. pv/ori-infected persons never

develop these neoplasms (12). A logical next step is to identify otherfactors that more precisely determine risk among H. /nYon-infectedpersons. H. pylori strains are highly diverse (13-15), and individuals

may harbor more than one strain (14, 15). Thus, it is appropriate toisolate particular characteristics of H. pylori strains that might affectrisk of gastric cancer development.

Most presently known phenotypic characteristics of H. pylori areconserved among virtually all strains. However, approximately 60%of isolates possess a gene, cagA, which encodes a high molecularweight protein (CagA) of variable size (M, 120,000-140,000) (16,17). Immunoblot studies suggest that persons infected with cagA+3

strains have higher degrees of gastric inflammation and epithelial cell

Received 1/24/95; accepted 3/16/95.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 with18 U.S.C. Section 1734 solely to indicate this fact.

1Supported by NIH Grants R01CA 58834 and 33644, R29DK45293, and

5T32DK07673, and the Medical Research Service of the Department of VeteransAffairs.

2 To whom requests for reprints should be addressed, at Division of Infectious

Diseases. Vanderbilt University School of Medicine, A-3310 Medical Center North, 1Ihl21st Avenue South, Nashville, TN 37232-2605.

3 The abbreviations used are: cagA+, fa#4-positive; cagA ', ffl^A-negative; IL, inter-

leukin; OR, odds ratio; CI. confidence interval.

damage than do persons from whom cagA strains have been isolated(18). Persons infected with cagA+ H. pylori strains have enhanced

expression of IL-la, IL-lÃŸ,and IL-8 in gastric biopsies compared touninfected persons or patients infected with cagA~ strains (19). Since

both intensity of inflammation and epithelial damage may be involvedin the pathogenesis of gastric cancer (1), it is reasonable to examinethe importance of cagA in this context. The presence of serumantibodies to the cagA product is strongly associated with pepticulcÃ©ration(18, 20, 21), as well as with inflammation.

We benefit from our previous prospective study to determinewhether H. pylori infection was a risk factor for the development ofgastric cancer among American men of Japanese ancestry in Hawaii(5). In this study, 111 patients who developed gastric cancer over a21-year period were identified. Serum specimens from these patients

(and from matched controls) had been obtained an average of 13 yearsprior to the diagnosis of gastric cancer.

We now report the development and validation of a new serologicalassay based on a recombinant fragment of cagA, and we explore therole of cagA as an additional risk factor for gastric cancer amongpersons from this cohort who were infected with H. pylori on entryinto the study.

MATERIALS AND METHODS

Selection of Patients for Validation Study. To determine the utility of theCagA serological assay, we studied sera from 181 persons in Nashville, TN;Los Angeles, CA; and Syracuse, NY, whose H. pylori status had been definedpreviously (19, 22-24). Uninfected persons (n = 115) were those who under

went endoscopy and had biopsies that did not reveal H. pylori infection byrapid urease test or by histological examination; all patients also had negativeserology (in a standardized ELISA) for serum IgG directed to H. pylori (25).The 115 uninfected patients were divided arbitrarily into a reference (n = 35)and a test (n = 80) group. Infected persons were those from whom H. pylori

was isolated from culture of gastric biopsies; these 66 patients included thosewith duodenal ulcÃ©ration(n = 14), gastric ulcÃ©ration(n = 6), gastritis alone(n = 36), or other diagnoses (n = 10). For each infected patient, a single H.

pylori isolate was evaluated to determine cagA status by colony hybridizationwith a gene-specific probe, as described previously (16, 24). On the basis of thehybridization assay, 36 patients were defined as being infected by a cagA*strain and 30 patients by a cagA~ strain. All sera had been stored at -20Â°C

until used.Selection of Patients for Cancer Study. All the patients in this study were

part of the Japan-Hawaii Study cohort as described previously (5). During the21-year period from 1968 to 1989, 109 cases of pathologically confirmed

gastric carcinoma had been identified, and previous serological testing indicated that 103 of these patients had been infected with H. pylori at the time oftheir original serum submission in the 1960s (5). Of the 103 matched controlsfor these cases, 83 had been infected with H. pylori at the time of originalserum submission. For each of the remaining 20 H. pylori-infected cases, 3

additional controls matched on age at examination and date of serum collectionwere identified following the previous criteria (5). Serology was done on codedsamples to determine whether H. pylori infection was present. Each of the 20control sets had at least 1 potential control with H. pylori IgG antibodypositivity. If more than 1 of the 3 controls for each case was positive, then asuitable control was selected randomly. Thus, in total, the study consisted of

2111

Research. on November 16, 2018. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

H. pylori i-agA AND GASTRIC CANCER RISK

103 H. pylori-infected men who developed gastric cancer and their 103

(83 + 20) matched controls, who also were infected with H. pylori but did notdevelop gastric cancer during the study period.

Preparation of the Recombinant CagA Antigen. A 1.7-kilobasc BamHlfragment containing base pairs 1921-3648 of cagA cloned in pMC3 (16) was

subcloned into the BamHl site of pETlSb (Novagen, Madison WI) downstream of the T7 promoter. This plasmid (pORV220) was used to transformEscherichia coli host strain BL21, a Ã€DE3 lysogen containing the T7 RNApolymerase gene under control of the lacUVS promoter (26). Addition ofisopropyl-ÃŸ-D-thiogalactopyranoside to a growing culture of the lysogen in

duces TV polymerase to transcribe the target DNA on the recombinant plasmid.This vector allows transcriptionally regulated expression of a fusion proteinconsisting of the CagA fragment with a NH,-terminal histidine tag. The fusion

protein consists of 600 residues (24 vector + 576 insert) and has a predictedmolecular mass of 66.4 kilodaltons. The protein was purified from lysates ofinduced broth cultures with the use of a nickel-chelating resin and eluted with

imidazole as described (27).Serological Methods. The presence of serum IgG antibodies to H. pylori

was determined by ELISA with the Pyloristat kit (Biowhittaker, Walkersville,MD) as described previously (5). For the CagA ELISAs, optimal concentrations of antigen, patient serum, and anti-human IgG conjugate were determined

by checkerboard titrations. For orv220, the optimal antigen concentration was5 (xg/ml, and 100-/J.I aliquots were loaded into wells in a 96-well microtiter

plate. The optimal dilution of human serum was 1:100, and horseradishperoxidase-conjugated goat anti-human IgG was used at a dilution of 1:4000.

Other details of the serological methods were exactly as described in similarassays (23, 25).

Assessment of Stability of Antibody Levels. To determine whether serumantibodies to the cagA product persist over the course of chronic H. pyloriinfection, we evaluated paired serum specimens from 36 healthy epidemiologists who were part of a cohort studied previously for clinical and epidemio-

logical features associated with H. pylori infection (28). On average, thespecimens were obtained 7.59 years apart, and knowledge of H. pylori statuswas available from serological studies using a conserved H. pylori antigen(28). These sera were also examined for antibodies to the cagA product withthe use of the methods developed in the previous sections.

Statistical Analysis. The I test for paired samples was used for the comparison of means, and McNemar's test was used for the comparison of the

distribution of various characteristics between patients and control subjects.The OR for stomach cancer, based on the results of the CagA assay, wasderived from conditional logistic regression methods (29). Tests for lineartrend in the logit of risk were derived from conditional logistic regressionmodels through the use of grouped cagA test results (coded by quartile as 1, 2,3, or 4). All models of conditional logistic regression were fitted by the use ofiterative maximum likelihood methods and a special application of the proportional hazards regression model (30). The estimate of the attributable riskof gastric carcinoma related to a cagA+ strain was based on the method of

Walter (31). A receiver-operator-characteristic curve was constructed to sum

marize the sensitivity and specificity estimates (32).

RESULTS

Concordance of Findings with pEM3 and orv220. Previousstudies of CagA serology used a recombinant antigen that was purified from E. coli cell lysates by sequential column chromatography(21). Because purification of this antigen was tedious and prolonged,overlapping cagA gene fragments were expressed as fusion proteins inan alternate prokaryotic expression system. Preliminary studies indicated that orv220 had the greatest utility of several of the candidateantigens based on CagA fragments (data not shown). We first askedhow this truncated protein compared with the antigen purified from E.coli strains transformed with pEM3, which encodes the entire cagAopen reading frame. By linear regression analysis, serum IgG resultsfor 41 persons (19 infected and 22 uninfected) correlated closelybetween assays with the use of the 2 different antigens (r = 0.96;

P < 0.001). With the use of the same threshold based on the meanvalue for uninfected persons + 2 SD, the sera from 21 (95%) of the

22 uninfected persons failed to react with orv220; the exception wasweakly positive. For the 19 H. pylori-infected persons, the results with

orv220 were exactly the same as for the pEM3 antigen (12 positiveand 7 negative). Thus, serological reactivity with the 66.4-kilodalton

CagA fragments encoded by orv220 was nearly identical with thatdetected with a larger CagA fragment.

Diagnostic Accuracy of CagA Serology with the Use of orv220.We next asked whether CagA serology based on serum IgG couldaccurately reflect the cagA status of H. pylori strains with whichpatients were infected. To establish the assay, sera from 35 personsknown not to be infected with H. pylori were used for reference, andafter multiple runs, thresholds based on mean values plus intervals ofSD were established. Concurrently, sera were tested from a secondgroup of 80 uninfected persons, 36 persons known to be infected witha cagA + strain and 30 persons from whom the only H. pylori isolateobtained was cagA~. Not surprisingly, absorbance ratios for the

uninfected persons were nearly identical to those for the referencegroup (Fig. 1). In contrast, the values for the persons infected withcagA+ strains were significantly higher (Student's / test, P < 0.001;

one-tailed test). Among the 30 sera obtained from persons from whomthe only isolate was a cagA" strain, a bimodal distribution was

observed. For 27 of the sera, the values were similar to those in the 2groups of uninfected persons, but for 3 sera the values were muchhigher and similar to those for patients infected with cagA+ strains.

To establish a threshold for use in diagnostic assays, we examinedthe accuracy of several absorbance ratio cutoffs comparing uninfectedpersons and those known to be infected with cagA + strains (Table 1).

Overall, the highest accuracy was obtained when the mean value forthe 35 uninfected (referent) patients + 3 SD was used, with sensitivityof 94.4% and specificity of 92.5%. We used this type of threshold inall future studies. On the basis of receiver-operator-characteristic

curves, there was a high level of discrimination of cagA status with theuse of orv220 (Fig. 2).

Stability of Anti-CagA Response. Having paired sera obtained7.6 years apart on the average from patients in previous studies (28),we examined the stability of the anti-CagA response. Among 36

participants, 27 showed evidence of H. pvlori infection on initial(mean absorbance, 3.70 Â±0.26) and follow-up (mean absorbance,

4.14 Â±0.24) periods with the use of the ELISA based on conservedH. pylori antigens (25, 28). In contrast, among 9 uninfected persons,initial (mean absorbance, 0.21 Â±0.04) and follow-up (mean absorb-

1.20 n

C.O

0.80

0.40

0.00

Oo

Reference Uninfected Infected Cog- Infected Cog +(n=35) (n.SO) (n-30) (n=36)

H. pylori status

Fig. 1. Absorbance ratios in serum IgG CagA ELISA for 181 patients. Patients weredivided into four groups: uninfected H. pylori reference group (n â€”35); uninfected H.pylori study group (n = 80); H. pv/ori-infected with cagA~ strain isolated (n = 30); andH. /H'/ori'-infected with cagA+ strain isolated (n = 36). Absorbance ratios were deter

mined in relation to standardized serum controls in an ELISA with the use of orv220 (5Hg/ml) as antigen. Hari-mnal hars, means for each group. The two highest absorbance

ratio values for the persons from whom a cagA strain was isolated nearly overlap atapproximately 0.60.

2112



H. pylori cagA AND GASTRIC CANCER RISK

ance, 0.17 Â±0.03) values were much lower. CagA antibody levelswere stably low in persons uninfected with H. pylori, as expected(Table 2). With the use of the threshold determined from our previoustrials, the initial serum specimens indicated that 12 and 15 personswere infected with cagA+ or cagA~ strains, respectively. In general,

mean anti-CagA levels changed little among these patients (Table 2);

however, a single patient showed a specific absorbance ratio rise from0.06 to 0.98 during the 15-year interval between serum specimens,indicating the acquisition of a cagA+ strain after prior infection witha cagA~ strain. In total, these data indicate that the cagA products

induce a persistent and specific immune response as long as H. pyloriinfection persists.

Association of cagA Positivity with Gastric Cancer. The characteristics of the 103 H. pv/ori-infected men who developed gastriccancer during the 21-year observation period and their age-matched

controls are shown in Table 3. The two groups of men were similar

Table 1 Diagnostic value of CagA ELISAusing orv220

GroupNo. of

subjects

No. (%) exceeding threshold"

1 SD 2SD 3 SD 4SD

UninfectedInfected, cagA* strain

8036

12(15) 9(11.3) 6(7.5) 3(3.8)36(100) 35(97.2) 34(94.4) 31(86.1)

" Threshold defined as mean absorbance ratio values plus stated intervals of SD for

reference group of 35 persons known not to be infected by H. pvlori.

Table 3 Characteristics of H. pylon-infected patients with gastric cancer and controlsubjects at the time the serum specimen was obtained

CharacteristicMean

age at examination(yr)Bornin United States(%)Married

(%)Alcoholuse(%)Mean

body massindexMeandiastolic blood pressure (mmHg)Meanserum cholesterol(mmol/liter)Meanserum glucose (mmol/liter)Patients

(n =103)58.883936523.581.65.79.3Controls(n =103)58.783957324.082.15.69.2Pvalue0.180.830.720.270.330.740.740.92

Table 4 Odds ratios for the association between infection with a cagA* H. pylori

strain and gastric cancer

typeAllDistalIntestinalDiffuseMatched-pair

status"

(patients/controls)+/+69684818+/-2120182-/+111182-/-2211Total1031017523Oddsratio1.91.82.31.095%interval*(0.9-4.0)(0.9-3.8)(1.0-5.2)(0.1-7.1)

" +/+, both patient and matched control show serological response to CagA; +/-,patient but not matched control show serological response to CagA; -/+, control but notpatient show serological response to CagA; -/-, neither patient nor control show serological response to CagA.

h Two-tailed analysis.

(1.1)Table 5 Odds ranos for gastric cancer according to cagA test results

and antibodv levels

0.2 03 0.4 0.5 0.Â« 0.7 0.6 0.9 1Fabo-PcBlttve Fraction (1- Specificity}

Fig. 2. Receiver operator characteristic analysis of the utility of orv220 for detectingserum IgG to CagA. The false-positive fraction (1-specificity) versus the true-positivefraction (sensitivity) was plotted for each of three criteria, when: (A) only observationsclassified definitely positive were considered positive; (ÃŸ)observations classified definitely or probably positive were considered positive; and (C) all except those classifieddefinitely negative were considered positive. The anchor points (0, 0) and (1, 1)depict theextreme conditions.

Table 2 Stability of serum antibodies to H. pylori CagA orv220 antigens in 36 subjectsAbsorbance ratios"

Timing ofserumspecimenInitial

Follow-up''H.

pyloriuninfected(n

=9)0.027Â±0.0002

0.036 Â±0.0003H.

pyloriinfectedOriginal

value Original valuenegativepositive^(n

= 15) (n = 12)0.036 Â±0.02 0.587 Â±0.260.096 Â±0.24 0.600 Â±0.03

CagA testresultsNegativePositive0.152-0.3990.400-0.59920.600P

value for trendNo.

ofpatients13373318No.

ofcontrols22292129Oddsratio1.02.32.61.00.9595%confidenceinterval1.0-5.61.1-6.40.4-2.5

" Mean Â±SD of absorbance ratios for serum specimens examined in CagA ELISAusing orv220, as described in "Materials and Methods."

Absorbance ratios in initial serum <0.152, as defined previously.r Absorbance ratios in initial serum >0.152, as defined previously.

Follow-up specimen obtained an average of 7.59 Â±1.0 years after initial specimen.

2113

with respect to demographic characteristics and laboratory values. Wethen asked whether initial CagA seropositivity was a risk factor fordevelopment of gastric cancer over the 21-year observation period.

For this population, presence of serum antibodies to CagA was associated with increased risk of cancer development (OR = 1.9; Table 4),but the association was not statistically significant (P = 0.08). An

analysis confined to the 101 cases of cancer of the distal stomach andtheir controls showed highly similar values, as expected (OR = 1.8;P = 0.11). When the cases with distal cancers were stratified by the

histological type of tumor, the strongest association was with theintestinal type (OR = 2.3; P = 0.056) but not the diffuse type(OR = 1.0; P = 1.0); 3 men had an indeterminant histological pattern.

Positivity in the cagA ELISA was not associated with the intervalbetween when serum was obtained and the diagnosis of cancer(OR = 2.5 and 95% CI = 0.5-12.9 for interval <10 years; OR = 1.7and 95% CI = 0.7-3.8 for interval >10 years). In the population ofH. pylori-infected persons, the presence of CagA antibodies wasassociated with an attributable risk of 28% (95% CI = 0-57) for

gastric cancer, but the height of the CagA antibody response was nota risk factor for development of distal gastric cancer (Table 5). Toassess whether men infected with cagA+ strains were more likely to

have high antibody levels to conserved H. pylori antigens, we compared results from the previous IgG ELISA to detect responses tothose antigens (5) and from the present anti-CagA assay. The age-

adjusted mean IgG values (2.26 Â±0.06) for 170 subjects (cases and



H. pylori cagA AND GASTRIC CANCER RISK

controls) infected with cagA+ strains were nearly identical with thosefor the 36 men infected with cagA" strains (2.31 Â±0.13; P = 0.73).

Independent analyses restricted to either cases or controls aloneshowed no significant differences in IgG values to the conservedantigens by cagA status (data not shown).

Next we stratified the analysis by the age of the patients at whichgastric cancer was diagnosed. CagA seropositivity was associatedwith a 3-fold (95% CI = 1.0-9.3; P = 0.057) increase in gastric

cancer risk for 52 men diagnosed under the age of 72 years; incontrast, for 51 men who were a 72 years at the time of diagnosis, theassociation was not significant (OR = 1.3; 95% CI = 0.5-3.5). CagA

seropositivity was associated with risk of presenting with an advancedstage tumor (3 or 4) at time of cancer diagnosis in 81 men (OR = 2.6;95% CI = 1.1-6.2; P = 0.03) but not an earlier stage tumor (1 or 2)in 21 men (OR = 1.0). The risk of developing gastric cancer associ

ated with CagA seropositivity was not increased when subjects werestratified according to birth order or sibship size.

DISCUSSION

A study of a total of 115 persons known not to be infected with H.pylori showed that there was little background serological reactivitywith the recombinant CagA fusion protein encoded by orv220 andtherefore confirms the specificity of this diagnostic test. Amongpersons known to be infected with cagA+ strains, there was a broad

range in serological reactivity but without substantial overlap with thereactivity of uninfected persons. The stability of the results over yearsin the absence of antimicrobial therapy further indicates the utility ofthe assay.

Simultaneous gastric infection with two H. pylori strains has beenreported with frequencies of 10-13% (14, 15), and simultaneous

infection with 3 different strains also has been observed (33). In ourinitial validation study, only a single H. pylori isolate from eachpatient had been examined so there was no opportunity to detectmultiple infection. However, the fact that 3 of 30 (10%) persons fromwhom the isolate was a cagA~ strain had high level serological

responses to CagA suggests that these persons were coinfected with acagA+ strain. Since only the conjunction of simultaneous infectionwith a cagA" index strain and a second cagA+ strain [and notsimultaneous cagA+/cagA+, cagA~/cagA~, and cagA+ (index)/cagA"] could show dichotomous results between colony testing and

serological assay, our data suggest that the frequency of multipleinfection may be substantially higher than what was reported previously, which was based on a small number of biopsies (14, 15). Thepresent work further supports the notion that noninvasive serologicalassays for H. pylori infections are desirable because they are globalassays that in essence sample the entire stomach. In contrast, biopsy-

based techniques only sample a tiny fraction (<Cl%) of the gastricmucosa. Because CagA is an immunodominant antigen (17), serological assays potentially have the power to detect infections with cagA +organisms even if numbers are low in relation to cagA" isolates.

Having an accurate assay, we thus were able to test the hypothesisthat infection with a cagA+ strain was a risk factor for the develop

ment of gastric cancer. The study design of the Japan-Hawaii cohort(34) and our previous investigations of the association (OR = 6.0;95% CI = 2.1-17.3) of H. pylori infection with the development ofgastric cancer (5) facilitated the present study. Infection with a cagA +

strain nearly doubled the risk of developing gastric cancer over theensuing 21 years, compared with infection with a cagA~ strain, and

the effect was more marked for persons who developed intestinal typeneoplasms.

In these studies, the increase in risk associated with a cagA+ straincompared with a cagA~ strain ranged from 1.8 to 2.3, and some of the

differences between strata in cagA status were not statistically significant. Undetected differences in risk between the two groups, lack ofperfect accuracy of the assay, and variation in host response to thisinfection could help account for the lack of significance. Furthermore,we used the conservative two-tailed analysis of significance. It may beargued that the one-tailed analysis is more appropriate since the prior

literature suggests an association of enhanced inflammation and gastric cancer risk (5, 8, 11); thus, we did not expect an inverse association between cagA positivity and cancer risk. Use of one-tailed

analysis indicates that the associations with all cancers and intestinaltype cancers reach statistical significance (P = 0.04 and 0.028,

respectively). Similarly, Crabtree et al. (35) showed that CagA seropositivity among H. /ry/ori-infected patients, as determined by immu-

noblotting, was significantly greater in gastric cancer patients at thetime of diagnosis (91%) than in infected controls with non-ulcerdyspepsia (72%; P = 0.028).

The associations with the subset of more aggressive tumors(younger age and higher stage when diagnosed) and the consistencyof the data with our a priori hypothesis suggest that the effect is real.This positive effect is biologically plausible for several reasons: (a)infection with cagA+ strains has been associated with enhanced

epithelial cell injury (18, 19), and injury to surface gastric epithelialcells may promote or possibly initiate oncogenesis; (b) infection withcagA+ strains is associated with higher degrees of gastric inflamma

tion (18, 19) and with enhanced expression of proinflammatory cy-tokines such as IL-1 and IL-8 (19, 36). These may contribute toepithelial injury; (c) most cagA+ strains also express vacuolating

cytotoxin activity (16,17). Infection with cytotoxin-producing strains,

as assessed by presence of serum neutralizing antibodies, may beassociated with the presence of gastric cancer (37); and (d) intestinaltype gastric cancer (associated with CagA seropositivity in this study)is more highly associated with increased inflammation and advancedatrophie gastritis than is the diffuse type (38, 39). We hypothesize thatthe enhanced intensity of inflammation induced by the cagA+ strain

results in accelerated mucosal damage with loss of epithelial structures and subsequent atrophy and eventually metaplasia, consistentwith the model proposed by Correa (11). Considering the smallnumber of cases of diffuse cancer studied, the lack of significancedoes not preclude an association of cagA positivity and these tumors.

For the original group of 109 patients with gastric cancer and theirmatched controls, those who were infected with a cagA" strain of

H. pylori still had an increased risk of developing cancer (data notshown); thus, cagA status did not account entirely for the associationof H. pylori with gastric cancer. In our previous study (5) and in amore recent Japanese study (8), high titered antibody responses toconserved H. pylori antigens were associated with gastric cancer risk.One interpretation of this phenomenon is that high antibody levels aremarkers for the degree of inflammation, and active inflammation isconsidered a precursor of oncogenic events (1). The lack of correlation of anti-CagA antibody levels and gastric cancer risk (Table 5)

may reflect that this antibody response does not mirror inflammatoryphenomena. In any event, the significance of these findings must betempered by the observation that CagA seropositivity was present in78% of the H. py/on-infected controls who did not develop cancer[and 72% of the non-ulcer dyspepsia patients in the study by Crabtreeet al. (35)]. Thus, infection with a cagA+ strain is neither necessary

nor sufficient for oncogenesis but is one of several factors that may beinvolved in this process.

Finally, the function of cagA to H. pylori is unknown, as is itsparticular role in induction of gastric inflammation. Presence of cagAin a strain may only be a marker for adjacent genes or for a particularphenotype that itself is relevant to inflammation or to the oncogenicprocess. Nevertheless, the results of this study provide for the first

2114



H. pylori i-agA AND GASTRIC CANCER RISK

time an indication that particular H. pylori strains may be associatedwith differential risk of gastric cancer. Further work in this directionis warranted.

ACKNOWLEDGMENTS

We thank Irene Feurer for statistical consultation and Dr. Julie Parsonnet forproviding clinical samples.

REFERENCES1. Blaser, M. i., and Parsonnet, J. Parasitism by the "slow" bacterium Helicobacler

pvlori leads to altered gastric homeostasis and neoplasia. J. Clin. Invest., 94: 4-8,

1994.2. Correa, P. Is gastric carcinoma an infectious disease? N. Engl. J. Med., 325:

1170-1171, 1991.

3. Talley, N. J., Zinsmeister, A. R., Dimagno, E. P., Weaver, A., Carpenter, H. A.,PÃ©rÃ©z-PÃ©rÃ©z,G. I., and Blaser, M. J. Gastric adenocarcinoma and Helicobacler pyloriinfection. J. Nail. Cancer Inst., 83: 1734-1739, 1991.

4. Parsonnet, J., Friedman, G. D., Vandersteen, D. P., Chang, Y., Vogelman, J. H.,Orentreich. N., and Sibley, R. K. Helicobacler pvlori infection and the risk of gastriccarcinoma. N. Engl. J. Med., 325: 1127-1131, 1991.

5. Nomura, A., Stemmermann, G. N., Chyou P., Kalo, I., PÃ©rez-PÃ©rez,G. I., and Blaser,M. J. Helicobacter pvlori infection and gastric carcinoma in a population of Japanese-Americans in Hawaii. N. Engl. J. Med., 525; 1132-1136, 1991.

6. Forman, D., Silas, F., Newell, D. G., Stacey, A. R., Boreham, J., Peto, R., Campbell,T. C., Li, J, and Chen, J. Geographic association of Helicobacter pylori antibodyprevalence and gastric cancer mortality in rural China. Int. J. Cancer, 46: 608-611,

1990.7. EUROGAST Study Group. An international association between Helicobacler pylori

infection and gastric cancer. Lancet, 341: 1359-1362, 1993.8. Asaka, M., Kimura. T., Kalo, M., Kudo, M., Miki, K.. Ogoshi, K., Kalo, T., Tatsuta,

M., and Graham, D. Y. Possible role of Helicobacter pylori infection in early gastriccancer development. Cancer (Phila.), 73: 2691-2694, 1994.

9. Sipponen. P., and Hyvarinen, H. Role of Helicobacler pylori in the pathogenesis ofgastritis, peptic ulcer and gastric cancer. Scand. J. Gastroenterol., 196 (Suppl): S3â€”S6,1993.

10. Forman D., Newell, D. G., Fullerton F., Yarnell, J. W., Stacey, A. R., Wald, N., and Sitas,F. Association between infection with Helicobacter pylori and risk of gastric cancer:evidence from a prospective investigation. Br. Med. J., 302: 1302-1305, 1991.

11. Correa, P. Human gastric carcinogenesis: a multistep and multifactorial processâ€”First American Cancer Society Award lecture on cancer epidemiology and prevention. Cancer Res., 52; 6735-6740, 1992.

12. Taylor, D. N., and Blaser, M. J. The epidemiology of Helicobacter pvlori infections.Epidemiol. Rev., 13: 42-59, 1991.

13. Akopyanz, N., Bukanov, N. O., Westblom, T. U., Kresovich, S., and Berg, D. E. DNAdiversity among clinical isolates of Helicobacler-pvlori detected by PCR-basedRAPD fingerprinting. Nucleic Acids Res., 20: 5137-5142, 1992.

14. Fujimoto, S., Marshall, B., and Blaser, M. J. PCR-based restriction fragmentlength polymorphism typing of Helicobacter pvlori. J. Clin. Microbio!., 32:331-334, 1994.

15. Prewett, E. J., Bickley, J., Owen, R. J., and Pounder, R. E. DNA patterns ofHelicobacter pvlori isolated from gastric antrum, body, and duodenum. Gastroenter-ology, 702; 829-833, 1992.

16. Tummuru, M. K. R., Cover, T. L., and Blaser, M. J. Cloning and expression of a highmolecular weight major antigen of Helicobacter pylori: evidence of linkage tocytotoxin production. Infect. Immunol., 61: 1799-1809, 1993.

17. Covacci, A., Censini, S., Bugnoli, M., Petracca, R., Burroni, D., Macchia, G.,Massone, A., Papini, E., Xiang, Z., Figura, N., et al. Molecular characterization of the128-kDa immunodominant antigen of Helicobacter pvlori associated with cytotoxic-ity and duodenal ulcer. Proc. Nati. Acad. Sci. USA, 90: 5791-5795, 1993.

18. Crabtree, J. E., Taylor, J. D., Wyalt. J. 1., Heally, R. V., Shallcross, T. M., Tompkins.D. S., and Rathbone, B. J. Mucosal IgA recognition of Helicobacler pylori 120 kDaprotein, peptic ulcÃ©ration,and gastric pathology. Lancet, 338: 332-335, 1991.

19. Peek, R. M., Miller, G. G., Tham, K. T., PÃ©rÃ©z-PÃ©rÃ©z,G. I., Cover, T. L., Dunn, D.,

and Blaser, M. J. Detection of cagA expression in vivo and demonstration ofpreferential cytokinc expression by cagA ' H. pylori strains in gastric mucosa. Am. J.

Gastroenterol., 89: 1344, 1994.20. Cover, T. L., Dooley, C. P., and Blaser, M. J. Characterization of and human serologie

response to proteins in Helicobacter pylori broth culture supcrnatants with vacuoliz-ing cytotoxin activity. Infect. Immunol., 58: 603-610, 1990.

21. Cover, T. L., Glupczynski, Y., Lage, A. P., Burette, A., Tummuru, M. K. R.,Perez-Perez, G. I., and Blaser, M. J. Serologie detection of infection with cagA4

Helicobacter pylori strains. J. Clin. Microbio!., in press, 1995.22. Dooley, C. P., Fitzgibbons, P. L., Cohen, H., Appleman, M. D., PÃ©rez-PÃ©rez,G. I.,

and Blaser, M. J. Prevalence of Helicobacter pylori infection and histologie gastritisin asymptomatic persons. N. Engl. J. Med., 321: 1562-1566, 1989.

23. Cover, T. L., Cao, P., Murthy, U. K., Sipple. M. S., and Blaser, M. J. Serumneutralizing antibody response to the vacuolating cytotoxin of Helicobacter pvlori.!. Clin. Invest., 90: 913-918, 1992.

24. Peek, R. M., Miller, G. G., Tham, K. T., PÃ©rez-PÃ©rez,G. L, Cover, T. L.. Atherton,

J. C., DÃ¼nn,D. Gâ€žand Blaser, M. J. Detection of Helicobacter pylori gene expressionin human gastric mucosa. J. Clin. Microbiol., 33: 28-32, 1995.

25. PÃ©rez-PÃ©rez,G. I., Dworkin, B. M., Chodos, J. E., and Blaser, M. J. Campylobacterpylori antibodies in humans. Ann. Int. Med.. 709; 11-17, 1988.

26. Studier, F. W., and Moffatt, B. A. Use of bacteriophage T7 RNA polymerascto direct selective high-level expression of cloned genes. J. Mol. Biol., 189:113-130, 1986.

27. Baicr, G., Baier-Bitlerlich, G., Couture, C., Telford, D., Giampa. L., and Altman, A.

An efficient expression, purification and immunodetection system for recombinantgene products. Biotechniques, 17: 94-99, 1994.

28. Parsonnet J., Blaser, M. J., PÃ©rez-PÃ©rez,G. I., Hargrett-Bean. N., and Tauxe, R. V.

Symptoms and risk factors of Helicobacter pylori infection in a cohort of epidemiologists. Gastroenterology, 102: 41-46, 1992.

29. Breslow, N. E., and Day, N. E. Statistical methods in cancer research. Vol. 1. pp.191-246. Lyon, France: International Agency for Research on Cancer, 1980.

30. Harrell, F. E., Jr. The PHGLM procedure. In: S. P. Joyner (ed.), SUGI SupplementalLibrary User's Guide, pp. 267-294. Cary, NC: SAS Institute, 1983.

31. Walter S. Calculation of attributable risks from epidcmiological data. Int.J. Epidemiol., 7; 175-182, 1978.

32. Lusted, L. B. Decision-making studies in patient management. N. Engl. J. Med., 284:416-424, 1971.

33. Taylor, N. S., Fox, J. G., Akopyants, N. S., Berg, D. E., Thompson, N.. Shames, B.,Correa, P., Fontham, H., and Janney, F. Long-term colonization with single and

multiple strains of Helicobacler pylori assessed by DNA fingerprinting. J. Clin.Microbiol., 33: 918-923, 1995.

34. Worth, R. M., and Kagan A. Ascertainment of men of Japanese ancestry in Hawaiithrough World War II selective service registration. J. Chronic Dis., 23: 389-397,

1970.35. Crabtree, J. E., Wyatt, J. 1., SÃ³bala, G. M., Miller, G., Tompkins, D. S., Primrose.

J. N., and Morgan, A. G. Systemic and mucosal humoral responses to Helicobacterpylori in gastric cancer. Gut, 34: 1339-1343, 1993.

36. Crabtree, J. E., Wyatt, J. I., Trejdosiewicz, L. K., Peichl, P., Nichols, P. H., Ramsay,N., Primrose, J. N., and Lindley, 1. J. D. Interleukin-8 expression in Helicobacter

pvlori infected, normal, and neoplastic gastroduodenal mucosa. J. Clin. Palhol., 47:61-66, 1994.

37. Hirai, M., Azuma, T., Ito, S., Kalo, T., Kohli, Y., and Fujiki, N. High prevalence ofneutralizing activity to Helicobacter pylori cytotoxin in serum of gastric-carcinomapatients. Int. J. Cancer, 56; 56-60, 1994.

38. Sipponen, P., Riihela, M., Hyvarinen, H., and Seppala, K. Chronic non-atrophic"superficial" gastritis raises the risk of gastric cancer. Scand. J. Gaslroenterol., 29:

336-340, 1994.39. Sipponen, P. Gastric cancer: a long-term consequence of Helicobacler pylori infec

tion? Scand. J. Gastroenterol., 29; 24-27, 1994.

2115



1995;55:2111-2115. Cancer Res Martin J. Blaser, Guillermo I. Perez-Perez, Harry Kleanthous, et al. Adenocarcinoma of the StomachAssociated with an Increased Risk of Developing

IscagA Strains Possessing Helicobacter pyloriInfection with

Updated version

http://cancerres.aacrjournals.org/content/55/10/2111

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/55/10/2111To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



infection with helicobacter pylori strains - cancer research

Documents