Results

A total 342 patients was initially enrolled into the study. Forty patients were excluded for the following reasons: presence of a bleeding lesion that precluded biopsy (n=11), early termination of the procedure due to patient discomfort (n=10), lack of corpus biopsy specimens (n=8), postgastrectomy (n=5), impassable oesophageal stricture (n=3), and gastric cancer (n=3). A total of 302 patients was included in the final analysis. Of those, 154 case subjects presented with endoscopic evidence of reflux oesophagitis, while the remainder of 148 control subjects had no endoscopic signs of reflux oesophagitis. In 52 case subjects, reflux oesophagitis was classified as grade 1, in 19 cases as grade 2, in eight cases as grade 3, and in 75 cases as grade 4. Barrett’s oesophagus was found in 73 case subjects, of whom 33 presented with a short segment of less than 3 cm, and 40 with a long segment of oesophageal metaplasia. Of 73 patients with Barrett’s oesophagus, 38 presented without erosions or strictures. In subsequent analyses, controls were compared with the total case population of all reflux patients (n=154) or with the subset of patients harbouring erosive oesophagitis (n=116), excluding the 38 patients who presented with Barrett’s oesophagus only. Patients with reflux oesophagitis were slightly older than controls without oesophagitis. There were more men and more whites among patients with reflux oesophagitis than among controls. The overall infection rate with H pylori, detected in any of the six gastric biopsy specimens, was similar in patients and controls (table 1).

Corpus biopsy specimens were available for all 302 study patients. Antral specimens were available in 298 patients (99%) and cardiac specimens in 283 patients (94%). Table 1 shows the proportions of patients and controls with acute or chronic gastritis by gastric region, irrespective of its grade. Acute and chronic gastritis tended to be less frequent in patients with reflux oesophagitis compared with controls. In general, the difference between patients and controls became more pronounced if the case population was limited to patients with erosive oesophagitis. Chronic and acute corpus gastritis were both significantly less common in patients with oesophagitis. The difference between patients and controls was less striking with respect to gastritis of the antrum or cardia.

Table 2 contains a more detailed breakdown of corpus gastritis by type and grade among patients with and controls without reflux oesophagitis. Compared with controls, larger proportions of patients with reflux oesophagitis were free of acute or chronic corpus gastritis, the patient-control difference being more striking with respect to chronic gastritis. In addition, the activity and severity of both types of gastritis were more pronounced in controls than patients. Atrophy and intestinal metaplasia tended to be slightly more common in controls without reflux oesophagitis. Similar patterns were found in comparisons of both types of patient population with the control population. Grading of H pylori infection of the corpus mucosa by density of organisms failed to reveal any difference between patients and controls. Overall, none of 90 patients without any form of gastritis were found to harbour H pylori, while the organism was detected histologically in 131 of 212 patients with gastritis (Spearman’s correlationr=0.57, p<0.0001). It might suspected that treatment with proton pump inhibitors suppressed H pylori; however, the infection rates of 34% and 47% were not significantly different in subjects with and without such treatment, respectively.

Table 3 is a correlation matrix of different forms of gastritis in patients with reflux oesophagitis. Different forms of gastritis tended to cluster in the same individual patients, as evidenced by the highly significant Spearman correlation coefficients. Two trends could be discerned. Firstly, the correlations seemed to be stronger among gastritis of any type from the same gastric region, such as acute and chronic gastritis of the antrum. Secondly, the correlations also seemed to be stronger among gastritis of the same type from different gastric regions, such as chronic gastritis of the antrum and chronic gastritis of the corpus. However, these trends did not reach statistical significance. Correlation coefficients similar to those in table 3 were observed in control subjects without reflux oesophagitis (data not shown).

Table 4 shows the results of the multiple logistic regression analysis, using as outcome variable the presence of any reflux disease (erosive oesophagitis plus Barrett’s oesophagus) or erosive oesophagitis only. Several models were tested and the most parsimonious one is presented here. Because of the significant correlations among various types of gastritis (table 3), only one type of gastritis was analysed at a time. Consistent with the results of the univariate analyses, chronic corpus gastritis yielded the best model (the highest log likelihood ratio). According to the multivariate model of table 4, chronic corpus gastritis was associated with a 54% reduced risk for reflux oesophagitis (95% confidence interval: 21 to 73%). The reduction was even more pronounced if the analysis was restricted to erosive oesophagitis alone (66%; 95% confidence interval: 39 to 81%). Tobacco smoking and male sex were stronger risk factors in the first regression, consistent with their particularly strong influence on the occurrence of Barrett’s oesophagus.

Discussion

In this study, corpus gastritis was found less frequently in patients with reflux oesophagitis than in those without oesophagitis. Chronic corpus gastritis was associated with a 54% reduction in the risk for reflux oesophagitis. Acute corpus gastritis was also protective against oesophagitis, but to a lesser degree than the chronic form. Moreover, both types of corpus gastritis tended to be less severe in patients with reflux oesophagitis.

Contact of the oesophageal lining with acid is crucial for the development of reflux symptoms, as well as mucosal damage.1 Although most subjects with reflux oesophagitis do not hypersecrete gastric acid or pepsin,9 ,10 some patients with refractory reflux disease present evidence of increased acid and pepsin output.4 Exposure of the oesophageal mucosa to hydrochloric acid alone results in injuries that are particularly severe, if the pH value is less than 1.3.11The severity of oesophagitis correlates with the time of acid exposure.12 Moreover, healing of reflux oesophagitis is achieved in most patients if their gastric acid output is suppressed below 1 mEq/h.4 The presence of acid is also necessary for other substances of the gastric refluxate to exert their injurious influence. For instance, pepsin becomes activated only at pH levels less than 4. The combination of hydrochloric acid and pepsin produces more severe oesophagitis than acid alone.3 The conjugated bile salts of duodenogastro-oesophageal reflux become mostly undissociated and membrane permeable molecules at an acidic pH.13

H pylori is the commonest known cause of gastritis. The influence of H pyloriinfection on gastric acid secretion is rather complex and depends, in part, on the type of ensuing gastritis. H pylori itself produces acid inhibitory proteins,14while the infection induced inflammation generates interleukin 1β, which is a potent inhibitor of acid secretion.15 ,16 In some of the patients infected by H pylori, the organism colonises primarily the antrum, resulting in an antrum dominant gastritis characterised by increased gastrin and acid secretion. Such patients may respond to successful eradication ofH pylori with reduced acid secretion.17 In other groups of patients,H pylori infection spreads from the antrum towards more proximal parts of the stomach. The corpus is the main acid producing region of the stomach.5 Corpus gastritis has been linked to reduced gastric acid secretion, the extent and severity of the gastritis being inversely related to the maximum gastric acid output.6 Eradication of H pylori in corpus dominant gastritis may be associated with an increased acid secretion.18 ,19 El-Omaret al investigated 16 hypochlorhydric subjects who concomitantly harboured H pylori related corpus dominant gastritis.18Eradication of H pylori led to a resolution of the corpus gastritis and recovery of gastric acid secretion within six months after treatment. Atrophy or intestinal metaplasia did not resolve after treatment.18 Gutierrezet al showed a correlation between the improvement in the gastritis score and increased gastric acid secretion after antibiotic therapy.19 The increased acid output after H pylori eradication may become clinically significant: Labenz et alfollowed patients with duodenal ulcer for three years after antibiotic treatment of H pyloriinfection.20 The incidence of reflux oesophagitis was 26% after eradication of H pylori compared with 13% in patients with persistent infection. Corpus gastritis before treatment was more severe in patients who subsequently developed reflux oesophagitis. In a similar study by Schütz et al, 10 of 16 patients with duodenal ulcer developed reflux oesophagitis after eradication of H pylori.21

Several investigators have studied the association betweenH pylori and GORD.22-29 Of eight trials, the six controlled trials yielded somewhat contrasting results. On the one hand, four studies observed no difference in the prevalence rate of H pylori among patients with and controls without oesophagitis.24-27 On the other hand, two studies reported lower prevalence rates ofH pylori in patients with oesophagitis compared with controls without oesophagitis.28 ,29 These results seem to suggest that H pylori does not play a major role in the pathogenesis of oesophagitis. Besides determining the presence of H pylori, De Koster et al also studied the occurrence and distribution of gastritis in their biopsy specimens and found lower prevalence rates of corpus gastritis among patients with erosive oesophagitis.26 Our results suggest that the regional distribution and the severity of gastritis are the important risk factors in the pathogenesis of reflux oesophagitis rather than the mere presence of H pylori in the gastric mucosa. The rates of infection with H pylori in patients with and controls without reflux oesophagitis were similar. More patients than controls were treated with proton pump inhibitors. Therefore, any potential suppression of H pylori by proton pump inhibitors would have affected patients with oesophagitis to a greater degree.

In a recent study of 132 patients infected with H pylori, the presence of gastritis in the three regions of the stomach was strongly correlated. Gastritis of the antrum, corpus, and cardia was found in 100%, 94%, and 98% of patients, respectively. In 14 patients with reflux oesophagitis, the grades of acute or chronic gastritis in the corpus were similar to those in the antrum and cardia.30 In a second study of 42 patients infected withH pylori, Genta et al detected H pylori organisms in 97%, 95%, and 95% of antral, corpus, and cardiac biopsy specimens. A similar density of organisms was found in all three regions. There were strong correlations between the antrum and cardia regarding the type and severity of gastritis. Both acute and chronic gastritis were more prominent in the antrum and the cardia than in the corpus.31 We found the presence of both acute and chronic gastritis from different gastric sites to be significantly correlated. Acute and chronic gastritis represent variations in the mucosal response to mainly one disease process, namely, infection withH pylori. Similarly, gastritis in different regions of the stomach reflects the spread and progression ofH pylori infection. The prevailing influence of one disease process is probably responsible for the clustering of different forms of gastritis in the same patients.

In conclusion, we found that acute and chronic corpus gastritis were significantly less common and less severe in patients with reflux oesophagitis than in controls. The infection rate ofH pylori was similar in the two groups. This pattern suggests that the distribution and severity ofH pylori related gastritis, rather than the mere presence or absence of H pyloriinfection, play a role in the pathogenesis of reflux oesophagitis.