Article Text


The significance of cagA + Helicobacter pylori in reflux oesophagitis
  1. V J Warburton-Timmsa,
  2. A Charlettd,
  3. R M Valorib,
  4. J S Uffc,
  5. N A Shepherdc,
  6. H Barrb,
  7. C A M McNultya
  1. aPublic Health Laboratory, Gloucestershire Royal Hospital, Gloucester, UK, bGloucester Gastroenterology Group, Gloucestershire Royal Hospital, Gloucester, UK, cDepartment of Histopathology, Gloucestershire Royal Hospital, Gloucester, UK, dPHLS Statistics Unit, 61 Colindale Avenue, London, UK
  1. Dr C A M McNulty, Public Health Laboratory, Gloucestershire Royal Hospital, Great Western Road, Gloucester GL1 3NN, UK. jwhiting{at}


BACKGROUND Helicobacter pylori is a gastroduodenal pathogen associated with ulceration, dyspepsia, and adenocarcinoma. Recent preliminary studies have suggested that H pylori may be protective for oesophageal adenocarcinoma. In addition, strains ofH pylori identified by the presence of the cytotoxin associated gene A (cagA) are shown to have a significant inverse association with oesophageal adenocarcinoma. Given that cagA + H pylori may protect against oesophageal carcinoma, these strains may be protective for oesophagitis, a precursor of oesophageal carcinoma.

AIMS The aim of this study was to investigate the association betweencagA + H pylori and endoscopically proved oesophagitis.

PATIENTS The study group included 1486 patients attending for routine upper gastrointestinal tract endoscopy.

METHODS At endoscopy the oesophagus was assessed for evidence of reflux disease and graded according to standard protocols. Culture and histology of gastric biopsy specimens determined H pylori status. The prevalence of cagA was identified by an antibody specific ELISA (Viva Diagnostika, Germany).

RESULTS H pylori was present in 663/1485 (45%) patients and in 120/312 (38%) patients with oesophagitis. Anti-CagA antibody was found in 499/640 (78%) H pylori positive patients. Similarly, anti-CagA antibody was found in 422/521 (81%) patients with a normal oesophagus and in 42/60 (70%) with mild, 24/35 (69%) with moderate, and 11/24 (46%) with severe oesophagitis. The risk of severe oesophagitis was significantly decreased for patients infected withcagA + H pylori after correction for confounding variables (odds ratio 0.57, 95% confidence interval 0.41–0.80; p=0.001).

CONCLUSIONS These results suggest that infection bycagA + H pylori may be protective for oesophageal disease.

Statistics from

Helicobacter pylori is an important and prevalent gastroduodenal pathogen associated with ulceration, dyspepsia, and adenocarcinoma.1 2 Circumstances that promote the development of H pyloriassociated diseases include host and environmental influences and features of the infecting strain.3

Significant gastroduodenal disease is particularly associated with infection by evidently virulent H pyloristrains that possess the cytotoxin associated gene (cag) pathogenicity island (PAI).4 The cag PAI is a segment of DNA with a distinct nucleotide composition compared with the rest of the chromosome, containing genes thought to have an association with pathogenesis. With H pylori the guanosine and cytosine content is 38–45% for other chromosomal sequences and 35% for the cag PAI.5 The cag PAI contains, among others, the cytotoxin associated gene A (cagA).6

cagA is present in about 70% of UK strains.7 Patients infected bycagA + H pylori have a greater risk of developing duodenal ulcer disease and gastric adenocarcinoma.8 H pylori expressing cagA are also known to contribute to significant gastric inflammation and cytokine production.9 Anti-CagA antibodies predict infection by acagA + strain. The CagA protein is possibly a surface expression protein or part of the export machinery necessary for stimulating the enhanced local inflammatory response seen with cagA +infections.5

The role of H pylori in gastro-oesophageal reflux disease (GORD) has only recently received attention largely because the prevalence of H pylori in patients suffering from GORD is similar to the normal population.10 However, eradication ofH pylori from those suffering from duodenal ulcer may promote reflux oesophagitis with a consequent increased risk of oesophageal/gastric cardia adenocarcinoma.11 H pylori with thecagA gene are significantly less prevalent in subjects with oesophageal adenocarcinoma.12 Diminution of peptic ulcer disease and adenocarcinoma of the distal stomach have paralleled the decreasing prevalence of H pylori infections in the developed world.13 At the same time, there has been an increase in GORD, Barrett's oesophagus, and adenocarcinoma of the distal oesophagus and proximal stomach of epidemic proportions,14 suggesting thatH pylori may protect against these oesophageal diseases, including those that predispose to carcinoma, such as Barrett's oesophagus. Indeed, in a nested case control study, Barrett's oesophagus and oesophageal adenocarcinoma were less common in H pylori infected patients.15

Our study was specifically aimed at assessing the association betweencagA + H pylori and endoscopically proved oesophagitis in a large cohort of unselected patients being investigated for upper gastrointestinal symptoms. The information was collected before the widespread use of proton pump inhibitors (PPIs).



Gloucestershire Royal Hospital has had an open access endoscopy service for 20 years. In 1986 a study of the role ofH pylori in upper gastrointestinal tract disease in an unselected cohort of 1486 upper gastrointestinal endoscopy patients was undertaken and therefore forms an ideal cohort to examine the role of H pylori in GORD. Endoscopies were performed by consultant gastroenterologists, surgeons, trainee gastroenterologists, GP gastroenterology hospital practitioners, and clinical assistants. A detailed endoscopy report form was completed by all endoscopists immediately after the examination. This had been in routine use in the department for some years and all endoscopists were trained and familiar with the grading. The endoscopist recorded the oesophageal appearance on a proforma as normal, mild oesophagitis (definite mild erythematous inflammation on the ridges of the oesophagus), severe oesophagitis (confluent inflammation with superficial ulceration), or moderate oesophagitis (moderate or confluent inflammation without ulceration). This grading was based on the system described by Blackstone.16 Hiatus hernia, varices, achalasia, monilia, or a motility problem were graded as normal. Oesophageal ulcer, stricture, and long segment Barrett's oesophagus (2 cm of concentric columnar epithelium above the oesophageal sphincter) were deemed as evidence of severe oesophagitis. The presence of short segment Barrett's, an entity not well recognised in 1986, was not recorded. Gastric biopsy specimens were taken from all patients within 5 cm of the pylorus and the presence ofH pylori infection was assessed by histology, microaerobic culture, and the biopsy urease test.


Two gastric biopsy specimens were placed in 10% formol saline and processed for histopathological evaluation. Sections were stained by haematoxylin and eosin, periodic-acid Schiff, and the half Gram method.17 Biopsy specimens were assessed by a histopathologist without knowledge of the patient's clinical details. Sections were graded for neutrophil (polymorphonuclear) infiltration, mononuclear cells, mucin depletion, H pylori, intestinal metaplasia, and lymphoid aggregates on a 0–3 scale with 3 being severe. The severity of chronic active gastritis was graded as mild, moderate, or severe according to standard protocols, or the biopsies were deemed normal.


Two biopsy specimens for culture were sent in normal saline to the laboratory. The biopsies were smeared across chocolate agar medium andCampylobacter selective medium.H pylori type strain NCTC11637 was incubated with the isolates to control for changes in atmospheric conditions. Plates were then incubated in a microaerobic atmosphere for 5–7 days. Plates were examined at five and seven days. H pylori was confirmed by Gram film and the rapid urease test.H pylori infection was considered present when H pylori was confirmed by either culture or histology.


Venous blood (10 ml) was stored as serum at −70°C in duplicate. Anti-CagA antibody titres were determined without prior knowledge of H pylori or oesophagitis status by use of the p120 CagA ELISA kit (Viva Diagnostika, Germany). This is a semiquantitative kit with a calibration calculation for each plate. Plates were read at 450 nm on a Titertek Multiscan (mcc/340 MK11) ELISA plate reader. Results were judged positive or negative according to the manufacturer's instructions and by cut off values validated by a select group of samples assessed blindly by western blot (Helicoblot 2.0; Biorad, UK).


Patient consent for the study and data on age, sex, past medical history, drug history, tobacco use, alcohol consumption (none, mild <15 units/week, moderate 15–30 units/week, severe >30 units/week), and family history of upper gastrointestinal disease (peptic ulcer, malignancy, hiatus hernia, dyspepsia in spouse, parent, sibling, or child) were obtained by standard questionnaire administered by nurses before the endoscopy. Full details of the questionnaire can be obtained from the authors. Ethics approval was obtained from the Gloucestershire Royal Hospital ethics committee and patients gave informed signed consent for biopsy specimens to be taken.


Initially the association between oesophagitis andH pylori and cagAstatus was assessed using χ2 tests of associations and trend. The association between oesophagitis and other factors was also assessed. Any factor having some evidence of association (p<0.2) was used in a multivariable logistic regression analysis.

The dependent variable in this analysis was whether or not the patient had endoscopically confirmed oesophagitis. The candidate predictor variables came from several areas. Demographic factors were: age group (<40, 40–49, 50–59, 60–69, ⩾70), sex (male or female), smoking (none, ex, or current), alcohol consumption (none, mild, moderate, or heavy), and past medical history of, or current, hiatus hernia. Family history factors were: other gastrointestinal disease and parent with a gastrointestinal disease. Drug history factors were: antacids more than twice a week, non-steroidal anti-inflammatory drug (NSAID), aspirin, and H2 blocker as either treatment or maintenance. Histological factors were: intestinal metaplasia (absent, present), mucosal type, and mononuclear and polymorphonuclear infiltration. The final factor was the status of the endoscopist. The variable giving H pylori status was classified as H pylori negative,H pylori positive, andcagA , orH pylori positive andcagA +.

Logistic regression analysis results in some loss of information as the ordinal classification of oesophagitis was dichotomised into present or absent. The likelihood ratio test was used to assess the significance of the predictor variables.


Oesophageal grading was available for 1485/1486 (787 male, 53%) non-duplicate patients. In the multivariable analysis, no specific grade of endoscopist was more or less likely to report oesophagitis.H pylori infection was detected in 663/1484 (45%) patients and in 120/312 (38%) patients with oesophagitis (table1).

Table 1

Prevalence of Helicobacter pylori infection in subjects with and without oesophagitis

Anti-CagA antibody was detected in 499/640 (78%) H pylori positive patients. The cagAstatus of the 663 H pylori positive patients and its association with oesophagitis is presented in table 2.cagA + H pylori were found in 422/521 (81%) patients with a normal oesophagus, in 42/60 (70%) with mild oesophagitis, in 24/35 (69%) with moderate oesophagitis, and in 11/24 (46%) with severe oesophagitis. The inverse association with acagA strain remained, even after correction for other confounding factors in a multivariable logistic regression analysis (table 3). For those patients who were H pylori positive with acagA + strain the estimated odds of oesophagitis were nearly half those in the H pylori negative group (odds ratio 0.57, 95% confidence interval (CI) 0.41–0.80; p=0.0001). Those patients that wereH pylori positive but with acagA strain were at a slightly increased odds of oesophagitis (odds ratio 1.05, 95% CI 0.67–1.66); however, this was not statistically significant.

Table 2

Proportion of patients cagA positive by category of oesophagitis

Table 3

Multivariable logistic regression analysis

The results of the multivariable logistic regression model are presented in table 3. Considering age as a risk factor, the odds of oesophagitis was not significantly higher in the 40–49 year category compared with the <40 year old category. For the other three age groups, the odds of oesophagitis compared with those <40 years increased by over 50% (odds ratios 1.55, 1.71, and 1.72 in the 50–59 years, 60–69 years, and 70 years or over age categories, respectively). The pattern of the odds ratios indicates that there was not a steadily increasing frequency of oesophagitis with age but rather a large step up in the frequency around the age of 50 years.

The risk of oesophagitis was significantly lower in females, with an odds ratio of approximately two thirds that for males. There was no association between smoking status and oesophagitis. The estimated odds of oesophagitis increased with increasing alcohol consumption. A test of linear trend (p=0.11) indicated weak evidence. The precision with which the odds ratios were estimated could indicate that the increase observed was due entirely to chance.

Those patients taking NSAIDs were associated with an increased odds of oesophagitis (odds ratio 1.54, 95% CI 0.76–3.15). However, this increase failed to reach statistical significance possibly because of the small numbers; only 46 patients had taken NSAIDs. Those patients taking antacids more than twice a week were associated with an increased odds of oesophagitis (estimated odds ratios of 1.58, 95% CI 1.19–2.11). There was no association with aspirin or H2blockers (p=0.89 and 0.92, respectively).

Several histology variables exhibited a weak association with oesophagitis grade. However, polymorphonuclear infiltration and intestinal metaplasia were the only variables significantly associated with oesophagitis. After allowing for other variables, there remained a weak association between intestinal metaplasia and oesophagitis; those patients with intestinal metaplasia had a reduced odds of oesophagitis (odds ratio 0.83, 95% CI 0.57–1.21).

Those patients with either a past history of hiatus hernia or who currently have this condition were associated with a significantly increased risk of oesophagitis (estimated odds ratio 4.86, 95% CI 3.59–6.57; p<0.0001).


The aim of this study was to investigate the relationship betweencagA + H pylori and endoscopically proved oesophagitis. Our study was performed prospectively on 1486 unselected consecutive patients attending open access endoscopy for investigation of dyspepsia. The study was carried out before the widespread use of PPIs and therefore our results were not influenced by the effects of hypochlorhydria. Other preliminary studies have found similar results but were multicentre investigations, lacked serology in many cases,18 or were performed on fewer patients.19 Several studies have examined the relationship with H pylori but have not determined CagA status.20 21

We found that patients infected bycagA + H pylori had half the likelihood of developing oesophagitis compared with those infected bycagA H pylori and those not infected. Over 80% ofH pylori isolated from patients with an endoscopically normal oesophagus werecagA +, this proportion decreasing to 70%, 69%, and 46% in mild, moderate, and severe oesophagitis, respectively. When other factors associated with the development of oesophagitis were taken into account in a multivariable logistic regression analysis, this association betweencagA + H pylori was still evident. The risk of oesophagitis in thecagA H pylori positive patients was no different to that in theH pylori negative patients.

A possible shortcoming of our study is that grading oesophagitis by endoscopic appearance does not necessarily reflect the extent of gastro-oesophageal reflux. Ambulatory 24 hour pH monitoring is the gold standard test of reflux but it is cumbersome and would be difficult to justify in a study such as this.22 Given the few patients suffering from severe oesophagitis and the highly significant difference in the prevalence of cagA, we feel that our results are accurate. One criticism may be that we identified cagA positivity by serology. Although the cagA gene is a marker of the presence of the PAI, it may not be a completely reliable marker of virulence. Some cagA + strains may have more virulent genes than others, depending on how much of the cag PAI is present. This problem could be overcome with a more reliable marker of a fully intact cag PAI which has not yet been reported. For this study, detection of anti-CagA antibodies was the most suitable method for differentiating H pylori strains and it is considered to be the most accurate method for assessing expression of the cag PAI.23

Age and sex were strongly associated with patients having oesophagitis. In addition, a weak trend for increasing alcohol consumption appeared to correlate with an increased risk of oesophagitis. The protective effect associated with cagA + H pylori infection was significant even after adjusting for age, sex, smoking, alcohol intake, drug history, and histological inflammation. Many studies have shown that age increases the risk of oesophagitis24 and a uniform lack of age adjusted control populations in most studies means that this is usually overlooked.25 Our finding of a close association between oesophagitis and hiatus hernia is in keeping with other studies of these two conditions.26-28 These studies have shown that hiatus hernia is more frequent in patients with reflux oesophagitis.26 Acid exposure of the oesophagus is higher, and lower oesophageal sphincter (LOS) pressure is lower in the presence of hiatus hernia.27 28 It seems likely that the relationship of hiatus hernia and reflux oesophagitis is causal, mediated by a reduction in protective mechanisms for oesophagitis in hiatus hernia.

It seems a contradiction thatcagA + H pylori are positively and negatively associated with two conditions, peptic ulcer and reflux oesophagitis, that both respond to control of acid secretion.29-31 This is probably explained by the severity and distribution of the gastritis associated with H pylori infection in a particular individual. Antral predominant gastritis is probably the most important predictor of duodenal ulcer disease in the presence ofH pylori infection.25 Antral predominant gastritis leads to reduced antral D cell density and somatostatin concentrations. This in turn leads to high serum gastrin and, in the presence of a normal corpus, to increased gastric acid, leading to a higher duodenal acid load and duodenal ulceration.32 33 As acid is produced in the gastric glands of the corpus, the amount of acid secreted in anyH pylori infected stomach is dependent largely on the severity of corpus gastritis. More severe corpus gastritis is associated with lower acid output, which returns to normal when H pylori is eradicated.34 35

Compared with cagA negative strains,cagA positive strains are associated with enhanced development of atrophic gastritis36 wherever the organism is present in the stomach. Atrophy of the corpus leads to the destruction of gastric glands and, in turn, hypochlorhydria. As this advances, the acid load presented to the duodenum and oesophagus diminishes, so protecting against GORD. This effect is lost whenH pylori is treated, leading to an increased risk of developing oesophageal reflux.37-39 The possibility that corpus gastritis plays a crucial protective role is supported by a study showing that patients with reflux symptoms without oesophagitis more often had active corpus gastritis than those with GORD with erosive oesophagitis.40 Our data showed that patients with antral intestinal metaplasia were less likely to have severe oesophagitis. Corpus biopsies were not taken for histological analysis in 1986 and therefore we are unable to predict the pattern of gastritis. H pylori is found in the cardia in almost all patients in whom infection is detected in the gastric antrum and body.41 42 Thus patients who develop gastrointestinal metaplasia and atrophy may have an increased risk of gastric carcinoma but a decreased risk of GORD.43-45Patients without corpus atrophic gastritis (so-called antral predominant gastritis) will have normal acid production and are therefore at increased risk of GORD and possibly oesophageal adenocarcinoma.

Gastro-oesophageal reflux arises from a weak LOS or one that relaxes inappropriately.46 47 LOS pressure can be affected by smoking48 and it was interesting to find that the ex-smokers in our study had a higher, although not significantly increased, risk of oesophagitis than current smokers. This may suggest that the effect of smoking on the LOS is not reversible. It is more likely however that the continued risk may be due to persistent airways disease and chronic cough that might promote gastro-oesophageal reflux.

It has been suggested that LOS pressure may be increased by the raised serum gastrin concentrations present in H pylori infections.49 Although initial animal studies demonstrated this association,50 a recent study showed no association between increasing gastrin concentration and LOS pressure.51

One mechanism that affects the pattern of gastritis and the likelihood of GORD is administration of PPIs.52 Our data were collected before the widespread use of PPIs and general practitioners were asked to stop H2 antagonists before endoscopy. Those patients taking antacids more than twice a week had an increased risk of oesophagitis but it is difficult to interpret this association as causal as it could simply reflect use to relieve symptoms.

H pylori infection per se was not associated with a protective effect against oesophagitis. Some studies have found an inverse relation between H pylori and oesophagitis once H pylori has been eradicated11 43 but most studies have found, like ours, that H pylori is not associated with oesophageal disease, positively or negatively.51-55

The debate on treatment of patients infected with H pylori continues. The WHO has classified H pylori as a grade 1 pathogen56 and therefore it would be unwise to stop treating this infection or base management on CagA status alone.57 However, our study has raised the possibility that eradication of CagA strains of H pylori may lead to exacerbation of oesophagitis. If this proves to be the case, eradication may lead to a significant increase in health care costs as treatment of reflux symptoms consumes nearly 1% of the NHS budget.58 Perhaps more significantly, it may lead to an increase in oesophageal adenocarcinoma.59Further well controlled studies are required to resolve the intriguing relationship between H pylori and oesophageal disorders and to inform future eradication policies.


Many thanks to all the endoscopy staff involved in this study. The study was funded by the Proctor and Gamble Company, USA. This paper was presented in part at the XIth International Workshop on Gastroduodenal Pathology and Helicobacter pylori, Budapest, Hungary, 2–5 September 1998.

View Abstract


  • Abbreviations used in this paper:
    cytotoxin associated gene
    cagA, cytotoxin associated gene A
    PAI, pathogenicity island
    gastro-oesophageal reflux disease
    non-steroidal anti-inflammatory drug
    proton pump inhibitor
    lower oesophageal sphincter

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.