Pathogenesis of Gastric Metaplasia of the Human Duodenum: Role of Helicobacter pylori, Gastric Acid, and Ulceration

Transcription

1 GASTROENTEROLOGY 1996;110: Pathogenesis of Gastric Metaplasia of the Human Duodenum: Role of Helicobacter pylori, Gastric Acid, and Ulceration SAM KHULUSI,* SUNIL BADVE, PRAFUL PATEL,* REBECCA LLOYD,* JOSÉ M. MARRERO,* CAROLINE FINLAYSON, MICHAEL A. MENDALL,* and TIM C. NORTHFIELD* Departments of *Medicine and Histopathology, St. George s Hospital Medical School, London, England Background & Aims: The pathogenesis of gastric meta- pression. The effect of prolonged and profound acid supplasia (GM) in the duodenum is unclear. The aim of pression on the extent of GM has not been determined. this investigation was to study the effect on the extent GM is present in the duodenal bulb of up to two of GM of ulcer healing, Helicobacter pylori eradication, thirds of dyspeptic subjects, even in the absence of ulcerand acid suppression singly and in combination. The ation. 7 Its prevalence is not influenced by Helicobacter relationship between GM and gastroduodenal inflampylori status, 7,8 but its extent is greater in those who mation and H. pylori infection density was also studied. are H. pylori positive, 3 suggesting that H. pylori may Methods: Duodenal and gastric antral biopsy specibe responsible for extending preexisting GM. H. pylori mens were obtained from H. pylori positive patients with duodenal ulcer and from H. pylori positive nonulthe duodenum, which results in chronic active inflam- colonizes both the gastric antrum and areas of GM in cer subjects. Biopsy procedures from patients with duomation denal ulcer were repeated after 6 months of treatment. at both sites. H. pylori antritis and duodenitis 3,9 Nonulcer subjects were treated symptomatically and are important factors in the pathogenesis of duodenal did not undergo re-endoscopy. Results: Ulcer healing ulceration. Eradication of H. pylori results in both the alone produced no change in GM or in gastroduodenal healing of duodenal ulcers and the resolution of active inflammation. H. pylori eradication produced a 42% re- inflammation. 10,11 Recently, two small studies 12,13 have duction in GM and improvement in inflammation. Acid given conflicting results on whether eradication of H. suppression produced a 43% reduction in GM but withpylori reduces the extent of GM. out a significant change in inflammation. A combination of H. pylori eradication and acid suppression produced This study aimed to investigate the pathogenesis of an additive effect with a 66% reduction in GM. A posiacid suppression, of H. pylori eradication, and of duodenal GM by determining the effect of prolonged and profound tive relationship was detected between the extent of GM and antral H. pylori density, duodenitis score, and ulcer healing, singly and in combination. It also investiantral gastritis score. Conclusions: This study shows gated the relationship between GM and gastroduodenal that the extent of duodenal GM is unrelated to the inflammation and H. pylori infection density in duodenal presence or absence of ulceration but is partly due to ulcer and nonulcer subjects. H. pylori and partly due to acid. Subjects and Methods astric metaplasia (GM) of the duodenum is the re- of groups of duodenal epithelial cells by Subjects Gplacement those with a gastric mucosal phenotype. It is a feature Ninety-seven consecutive subjects with H. pylori posi- of the duodenal bulb generally believed to occur as a tive duodenal ulceration undergoing endoscopy for dyspepsia nonspecific response to acid/peptic damage and is were studied; 62 were men (64%) with a median age of 48 strongly associated with duodenal ulceration. 1 3 In expositive nonulcer dyspepsia were also studied; 18 were men years (range, years). Thirty subjects with H. pylori perimental animals, it appears within a few days of muco- (60%) with a median age of 48 years (range, years). sal injury and resolves several months after ulcer healing, 4 Subjects who were taking nonsteroidal anti-inflammatory showing the reversible nature of GM in animal models. drugs or who had taken antibiotics or received acid suppression In humans, the extent of GM is related to gastric acid treatment in the previous month and patients with previous output independent of duodenal ulceration, 3 suggesting gastric surgery were excluded. The study was approved by the that acid itself might lead to the formation of GM. However, studies that have investigated the effect of acid Abbreviation used in this paper: GM, gastric metaplasia. reduction on the extent of GM have produced conflicting 1996 by the American Gastroenterological Association results, 5,6 possibly because of varying degrees of acid sup /96/$3.00

2 February 1996 GASTRIC METAPLASIA, H. PYLORI, ACID, AND ULCERS 453 St. George s Hospital Ethics Committee, and informed consent was obtained from all patients. Clinical Methods At initial endoscopy, three gastric antral biopsy specimens were obtained 2 cm from the pylorus, one for a rapid urease test and two for histology. Three duodenal bulb biopsy specimens per patient were obtained from the anterior, inferior, and superior walls, at least 1 cm away from any ulcer margin and from the pylorus. Biopsy specimens were fixed in formol saline and embedded in paraffin wax within 12 hours. Nonulcer subjects were given symptomatic treatment and did not undergo re-endoscopy. All patients with duodenal ulcer were randomized to 1 of 3 treatment regimens: 500 mg amoxicillin three times daily, 400 mg metronidazole three times daily, and 240 mg colloidal bismuth subcitrate two times daily for 2 weeks plus 40 mg omeprazole daily for 6 months (n Å 32); 500 mg amoxicillin three times daily, 400 mg metronidazole three times daily, and 240 mg colloidal bismuth subcitrate two times daily for 2 weeks plus omeprazole placebo for 6 months (n Å 32); or 2 weeks of antibiotic placebo plus 40 mg omeprazole for 6 months (n Å 33). Treatment began after the first endoscopy, and compliance with treatment was determined monthly from the numbers of returned tablets. Thirteen patients (2, 6, and 5 from the respective treatment groups above) did not fully comply with the protocol, and their results were excluded. All the remaining patients with duodenal ulcer underwent re-endoscopy at 6 months with treatment having been discontinued 1 week be- fore endoscopy; antral and duodenal biopsies were repeated from similar topographical sites. Eradication of H. pylori was diagnosed by a negative biopsy urease test and by negative histology. Urea breath tests were performed on all ulcer sub- jects 1 month after the second endoscopy. Histological Methods Sections from duodenal biopsy specimens were stained with diastase periodic acid Schiff/alcian blue and were exam- ined for GM and duodenitis (Figure 1). Antral sections were stained with H&E, and those negative for H. pylori were reassessed using a modified Giemsa stain. Histological material was examined by two histopathologists working together and unaware of whether the biopsy specimens were obtained from baseline or follow-up endoscopies. Grading was based on their mutually agreed values. The extent of gastric-type epithelium in the duodenal biopsy specimens was subjectively scored on a scale of 0 10 (representing 0% 100% of the total epithelial surface present 14 ). The severity and activity of inflammation in both antral and duodenal biopsy specimens was graded on a scale of 0 3. Grade 0 corresponded to no inflammatory cell infiltrate, grade 1 to mildly increased cellularity in the lamina propria including lymphocytes and plasma cells but without significant numbers of neutrophils, grade 2 to moderate numbers of lymphocytes and infiltration of the mucosa with neutrophils, and grade 3 to severe mucosal infiltration with large numbers of both lymphocytes and neutrophils. 15 The density Figure 1. Duodenal gastric metaplasia stained with diastase periodic acid Schiff/alcian blue. Positive staining of metaplastic epithelium (GM) is in contrast to the discrete goblet-cell (GC) staining in the native duodenal epithelium. (A) Small patches of gastric metaplasia are seen in nonulcer subjects. (B) More extensive metaplastic epithelium is detected in duodenal ulceration, often associated with an inflammatory cell (IC) infiltrate. of H. pylori was also assessed in antral biopsy specimens on a 0 3 scale corresponding to absent, scant, moderate, and heavy colonization. The mean score of the biopsy specimens obtained from the same area in the same subject was used. The scoring of GM was validated on 30 randomly selected sections. In these sections, the extent of GM was also measured by a morphometric technique using an interactive image analy- sis system (OsteoMeasure; Osteometrics, Atlanta, GA). There was a close correlation between the semiquantitative scores and quantitative measurements (r Å 0.89; P õ 0.001) (Figure 2). In the same sections, both histopathologists agreed on 24 of 30 (80%) GM scores. In replicate histological scoring of GM by the same histopathologists, 16 of 24 sections were graded as the same and 8 of 24 varied by no more than 1 grade from the previous mutually agreed scores. Statistical Methods The extent of GM and inflammation scores for the antrum and the duodenum were expressed as means. Differ- ences between follow-up values for subjects with duodenal

3 454 KHULUSI ET AL. GASTROENTEROLOGY Vol. 110, No. 2 Figure 2. Relationship between the extent of gastric metaplasia (%) determined by semiquantitative scores and quantitative measurement by morphometric technique. Figure 5 shows the extent at baseline of GM, duodenal, and antral inflammation scores and the density of antral H. pylori on histology in subjects with duodenal ulcer and nonulcer subjects. These histological parame- Results Effects of H. pylori Eradication, Acid Suppression, and Ulcer Healing Amoxicillin, metronidazole, and bismuth achieved an 85% H. pylori eradication rate when combined with omeprazole but a rate of only 63% without omeprazole. ulcer were compared using analysis of variance. The relation- ship between GM and inflammation and H. pylori density were assessed by Spearman rank correlation. The difference between ulcer and nonulcer subjects was assessed by the Mann Whitney U test. An eradication rate of 4% was obtained with 6 months of treatment after omeprazole alone. Table 1 shows the four outcomes of treatment (in addition to ulcer healing) that were further assessed in this study: H. pylori eradication, H. pylori eradication together with acid suppression, acid suppression, and failed H. pylori eradication without acid suppression. Incomplete compliance with treatment or the presence of an ulcer at follow-up resulted in the exclusion of 8 and 13 subjects, respectively. Table 2 shows the characteristics of the duodenal ulcer and nonulcer subjects with respect to age, sex, smoking, and family history of peptic ulceration. Figure 3 shows a significant reduction in metaplastic epithelium with ulcer healing plus H. pylori eradication (42%; P õ 0.002), ulcer healing plus acid suppression (43%; P õ 0.002), and ulcer healing with eradication plus acid suppression (66%; P õ 0.002) but no change with ulcer healing in the absence of eradication and acid suppression. Figure 4A and B show a reduction in duodenal and gastric antral inflammation scores after both eradication and eradication plus acid suppression but no change with noneradication or acid suppression alone. Interestingly, acid suppression produced a small but significant decline in antral H. pylori density (mean density score, 2 vs. 1.8; P õ 0.02). Gastric Metaplasia, Inflammation, and H. pylori Infection Density in Subjects With Duodenal Ulcer and Nonulcer Subjects Table 1. Treatment Outcomes Included Duodenal ulcer healing plus Excluded Noneradication, Eradication, Eradication no acid no acid and acid Acid suppression, Persistent suppression suppression suppression noneradication ulcer Noncompliance Amoxicillin, metronidazole, bismuth, and omeprazole (n Å 32) Amoxicillin, metronidazole, bismuth, and placebo (n Å 32) Omeprazole and placebo (n Å 33) NOTE. The three treatment regimens used in patients with duodenal ulcer resulted in four outcomes that were further assessed in the study: ulcer healing together with H. pylori eradication, eradication plus acid suppression, acid suppression alone, and failed H. pylori eradication.

4 February 1996 GASTRIC METAPLASIA, H. PYLORI, ACID, AND ULCERS 455 Table 2. Patient Characteristics Duodenal ulcer Ulcer healing plus Nonulcer Noneradication, Eradication, Eradication and Acid suppression, no acid suppression no acid suppression acid suppression noneradication n Å 30 (n Å 9) (n Å 15) (n Å 24) (n Å 28) Median age (yr) (range) 48 (19 71) 44 (24 70) 47 (29 73) 50 (20 70) 46 (22 72) Male/female 18/12 6/3 11/4 16/8 17/11 Smokers Family history of duodenal ulcer ters were similar in subjects with active ulcers and in Omeprazole treatment also caused a significant reduction subjects in whom the ulcer had healed, but H. pylori had in the extent of GM but had no effect on inflammation not been eradicated. However, regardless of whether an in the antrum or in the duodenum. The effect of omepraactive ulcer was present, all parameters were significantly zole was probably the result of prolonged acid suppresgreater in patients with duodenal ulcer than in nonulcer sion, although omeprazole is also known to have an inhibsubjects (P õ 0.001). itory effect on H. pylori, 16 and its effects may have been Figure 6 shows a positive relationship between the partly caused by an inhibition of the organism. Ulcer extent of GM and antral H. pylori density (r Å 0.58; P õ healing itself, without eradication of H. pylori or omepra ), duodenitis score (r Å 0.84; P õ 0.001), and zole treatment, produced no change in the extent of GM, antral gastritis score (r Å 0.54; P õ 0.001). indicating that ulceration is not itself a stimulus for a generalized increase in metaplastic epithelium in the Discussion duodenal bulb. This study shows that the extent of GM in the The use of biopsy specimens in the study of the extent duodenum is related to both duodenal and antral in- of GM in the duodenum carries the potential for a samflammation and to the density of H. pylori in the gastric pling error. Biopsy specimens from an inflamed part of antrum. Eradication of H. pylori resulted in a reduction the duodenal bulb may indicate a greater extent of metain inflammation at both sites and in the extent of GM, plasia compared with adjacent, less inflamed areas. 7 To suggesting that H. pylori is at least partly responsible reduce this error, biopsies were targeted to similar topofor the maintenance and probably the extension of GM. graphical sites in all subjects and the endoscopist was unaware of the treatment status of the patients. Wyatt et al. 3 showed that, although GM has a patchy distribution in the duodenal bulb, a single anterior duodenal biopsy could detect 63% of the GM identified by multiple biopsies. In this study, we used three duodenal bulb biopsy specimens, including one from the anterior wall, to provide adequate sampling of the extent of GM. In our current study, greater resolution of GM might have occurred with a longer period of follow-up, but the results of animal experiments suggest that 6 months of follow-up should be adequate to show maximum resolution. 4,17 Some of the resolution of GM with H. pylori eradication may have been artifactual. Severe duodenitis caused by H. pylori infection often produces a degree of villous atrophy and a reduction in the epithelial surface. Resolution of the villous atrophy after H. pylori eradication may itself have reduced the percentage of surface that Figure 3. Extent of gastric metaplasia (mean % / SEM) following H. was GM. Antibiotic regimens for H. pylori eradication are pylori eradication (E), eradication plus acid suppression (E&A), acid suppression alone (A), and failed H. pylori eradication (NE). *P õ known to produce incomplete compliance. In this study, 0.002; ANOVA., Pretreatment;, posttreatment. subjects were excluded if poor compliance was evident

5 456 KHULUSI ET AL. GASTROENTEROLOGY Vol. 110, No. 2 Figure 4. (A) Duodenitis and (B) gastritis (mean scores / SEM) following H. pylori eradication (E), eradication plus acid suppression (EA), acid suppression alone (A), and failed H. pylori eradication (NE). **P õ 0.001; ANOVA., Pretreatment;, posttreatment. from patient disclosure or from the number of returned after treatment was not reported. Noach et al., 12 in a tablets. The presence of an ulcer at follow-up in those small study of 20 subjects with duodenal ulcer, showed receiving omeprazole may also have been caused by poor a trend toward a reduction in GM after H. pylori eradication, compliance with treatment. Despite exclusions from this but the reduction did not achieve statistical significompliance trial, incomplete compliance may have reduced some of cance, probably because of insufficient statistical power. the effects observed. The extent of GM is greater in acid hypersecretory This study confirms the findings from our earlier, states such as duodenal ulceration 19,20 and gassmaller study, 13 which showed a close association be- trinoma, 21,22 and several studies have investigated the tween duodenitis and GM in the presence of H. pylori role of acid in the pathogenesis of GM. Wyatt et al. 14 antritis. It also supports the findings of Moshal et al., 18 showed that low acid output was associated with a rewho showed that duodenal ulcer healing after colloidal duced extent of GM and in the same study noted that bismuth treatment was associated with the reversal of duodenal biopsy specimens from subjects with atrophic GM, whereas ulcer healing with H 2 blockers produced gastritis showed no areas of metaplasia. Wyatt et al. 5 no such change. This could be explained by an inhibitory have also shown that patients had a lesser extent of GM effect of bismuth on H. pylori, although H. pylori status 2 years after vagotomy compared with a control group about to undergo vagotomy. In contrast, Jönsson et al. 6 found no reversal of GM up to 3 years after parietal cell vagotomy. The same study also failed to show any effect after the same period of treatment with cimetidine. Similarly, Noach et al. 12 found no significant difference in the mean extent of GM between patients taking H 2 - receptor antagonists for more than 1 year and subjects not taking these drugs. It is possible that the degree of acid suppression produced by the H 2 antagonist, at the doses given in these studies, was insufficient to affect GM, because the degree of suppression is known to be greater with omeprazole. Fullman et al. 23 showed that GM was found more frequently around the site of a healed ulcer than in an area of the duodenal cap away from the ulcer. GM probably provides protection for the duodenal cap from acid/peptic damage. This theory is supported by the relationship between the extent of GM and the level of gastric acid Figure 5. Gastric metaplasia (GM), duodenitis, gastritis, and H. exposure, 3,24 which suggests a possible dose-response efpylori density (mean score / SEM) in subjects with duodenal ulcer fect, 3,25 and the distribution of GM in the duodenum before treatment ( ) after ulcer healing alone (failed H. pylori and its rarity beyond the duodenal bulb. 3 The protective eradication therapy) ( ), and in subjects with nonulcer dyspepproperties of GM are probably related to the production sia ( ). **P õ 0.001, H. pylori positive duodenal ulcer subjects vs. nonulcer subjects. of mucus, which both provides a physical barrier and

6 February 1996 GASTRIC METAPLASIA, H. PYLORI, ACID, AND ULCERS 457 entraps bicarbonate to provide a buffering capacity against acid. 26 H. pylori infection in patients with duodenal ulcer has been shown to increase gastric acid production, leading to greater duodenal acidity. 27,28 It is possible that H. pylori produces an extension of GM through increased acid secretion. El-Omar et al. 29 have shown that both basal and gastrin-releasing peptide stimulated acid secretion rates decrease by more than 50% within 1 month after completing H. pylori eradication treatment. The effect of eradication on GM in this study, therefore, may have been partly caused by a reduction in acid secretion. GM resembles gastric foveolar epithelium in many respects, including the specificity of H. pylori colonization. Infection of the duodenum by H. pylori only occurs at the site of GM 10,14 and only in the presence of an infected gastric antral mucosa. This suggests that antral infection spreads to the duodenum in the presence of preexisting GM, resulting in duodenitis. Inflamed GM secretes mucus with an altered composition 30 and, rather than providing a stable barrier against acid, becomes more susceptible to peptic damage and ulceration. The current study has shown a relationship between the extent of GM and the density of H. pylori colonization in the antrum. The greater load of infecting organisms in the antrum of patients with duodenal ulcer compared with nonulcer subjects increases the likelihood of H. pylori colonization of GM, which may be an important stimulus for further extension of metaplastic epithelium. 13 In conclusion, this study has shown that the extent of GM in the duodenal bulb is associated with both duodenitis and gastritis and with the density of H. pylori colonization in the gastric antrum. Eradication of H. pylori resulted in a reduction of both antral and duodenal inflammation and a decline in the extent of GM, suggesting that inflammation is an important mechanism in the genesis and maintenance of GM, although part of the effect of H. pylori eradication may have been caused by reduced acid secretion. Omeprazole treatment also resulted in a reduction in GM but without significant change in antral or duodenal inflammation, indicating that acid has an effect on the formation of GM, which is independent of inflammation. Furthermore, the combination of H. pylori eradication and acid suppression produced an additive effect with a greater reduction in GM than either treatment alone. References Figure 6. Relationship between the extent of gastric metaplasia and (A) H. pylori infection density in the gastric antrum, (B) duodenitis, and (C) antral gastritis., Subjects with duodenal ulcer;, nonulcer subjects. 1. Caselli M, Trevisani L, Aleotti A, Bovolenta MR, Stabellini G. Gastric metaplasia in duodenal bulb and Campylobacter-like organisms in development of duodenal ulcer. Dig Dis Sci 1989;34: Carrick J, Lee A, Hazell S, Ralston M, Daskalopoulos G. Campylo-

7 458 KHULUSI ET AL. GASTROENTEROLOGY Vol. 110, No. 2 bacter pylori, duodenal ulcer, and gastric metaplasia: possible 17. Florey H, Harding H. The healing of artificial defects of the duodenal role of functional heterotopic tissue in ulcerogenesis. Gut mucosa. J Pathol Bacteriol 1935;40: ;30: Moshal MG, Gregory MA, Pillay C, Spitaels JM. Does the duodenal 3. Wyatt JI, Rathbone BJ, Sobala GM, Shallcross T, Heatley RV, cell ever return to normal? A comparison between treatment with Axon AT, Dixon MF. Gastric epithelium in the duodenum: its asso- cimetidine and DeNol. Scand J Gastroenterol (Suppl) 1978;13: ciation with Helicobacter pylori and inflammation. J Clin Pathol ;43: Marshall BJ, Barrett LJ, Prakash C, McCallum RW, Guerrant RL. 4. Tatsuta M, Iishi H, Yamamura H, Yamamoto R, Taniguchi H. Urea protects Helicobacter (Campylobacter) pylori from the bacte- Enhancement by tetragastrin of experimental induction of gastric ricidal effect of acid. Gastroenterology 1990;99: epithelium in the duodenum. Gut 1989;30: Graham DY, Opekun A, Lew GM, Evans DJ, Klein PD, Evans DG. 5. Wyatt JI, Rathbone BJ. Gastric metaplasia in the duodenum and Ablation of exaggerated meal-stimulated gastrin release in duo- Campylobacter pylori. Gastroenterol Clin Biol 1989;13:78B denal ulcer patients after clearance of Helicobacter (Campylo- 82B. bacter) pylori infection. Am J Gastroenterol 1990;85: Jönsson KA, Strom M, Bodemar G, Norrby K. Histologic changes 21. James A. Gastric epithelium in the duodenum of a patient with in the gastroduodenal mucosa after long-term medical treatment gastric hyper acidity. Proc 2nd Wld Congr Gastroenterology with cimetidine or parietal cell vagotomy in patients with juxtapy- 1963;2: loric ulcer disease. Scand J Gastroenterol 1988;23: Parrish J, Rawlins D. Intestinal mucosa in the Zollinger Ellison 7. Frierson HJ, Caldwell SH, Marshall BJ. Duodenal bulb biopsy syndrome. Gut 1965;6: findings for patients with non-ulcer dyspepsia with or without 23. Fullman H, Van Deventer G, Schneidman D, Walsh J, Elashoff J, Campylobacter pylori gastritis. Mod Pathol 1990;3: Weinstein W. Healed duodenal ulcers are histologically ill (abstr). Gastroenterology 1985;88: Kreuning J, Wal AM, Kuiper G, Lindeman J. Chronic non-specific 24. Graham DY. Campylobacter pylori and peptic ulcer disease. Gasduodenitis. A multiple biopsy study of the duodenal bulb in health troenterology 1989;96: and disease. Scand J Gastroenterol Suppl 1989;167: Patrick W, Denham D, Forrest A. Mucous change in the human 9. Johan G, Offerhaus A, Molyvas EN, Hoedemaeker PJ. Helicoduodenum: a light and electron microscopic study and correlation bacter pylor infection of gastric mucin cell metaplasia: the duodewith disease and gastric acid secretion. Gut 1974;15: num revisited. J Pathol 1990;162: Isenberg J, Selling J, Hogan D, Koss M. Impaired proximal duode- 10. Marshall BJ, Goodwin CS, Warren JR, Murray R, Blincow ED, nal mucosal bicarbonate secretion in patients with duodenal ul- Blackbourn SJ, et al. Prospective double-blind trial of duodenal cer. N Engl J Med 1987;316: ulcer relapse after eradication of Campylobacter pylori. Lancet 27. Moss S, Calam J. Acid secretion and sensitivity to gastrin in 1988;2: duodenal ulcer patients: effect of eradicating the organism. Gut 11. Rauws EA, Tytgat GN. Cure of duodenal ulcer associated with 1993;34: eradication of Helicobacter pylori. Lancet 1990;335: Levi S, Beardshall K, Haddad G, Playford R, Ghosh P, Calam J Campylobacter pylori and duodenal ulcers: the gastrin link. Lan- 12. Noach LA, Rolf TM, Bosma NB, Schwartz MP, Oosting J, Rauws cet 1989;1: EAJ, Tytgat GNJ. Gastric metaplasia and Helicobacter pylor infec- 29. El-Omar E, Penman I, Dorrian CA, Ardill JES, McColl KEL. Eradication. Gut 1993;34: tion of Helicobacter pylori infection lowers gastrin mediated acid 13. Khulusi S, Mendall M, Badve S, Patel P, Finlayson C, Northfield secretion by two thirds in patients with duodenal ulcer. Gut T. The effect of Helicobacter pylori eradication on gastric metapla- 1993;34: sia of the duodenum. Gut (in press). 30. Bode G, Malfertheiner P, Kullmann C, Ditschuneit H. Gastritis and 14. Wyatt JI, Rathbone BJ, Dixon MF, Heatley RV. Campylobacter duodenal ulcer: lectin-binding characteristics of Campylobacter pylori and acid induced gastric metaplasia in the pathogenesis pylori infected and non-infected surface mucous cells. In: Me- of duodenitis. J Clin Pathol 1987;40: graud F, Lamouliatte H, eds. Gastroduodenal pathology and 15. Fitzgibbons PL, Dooley CP, Cohen H, Appleman MD. Prevalence Campylobacter pylori. New York: Elsevier, 1989: of gastric metaplasia, inflammation, and Campylobacter pylori in the duodenum of members of a normal population. Am J Clin Received February 1, Accepted October 17, Pathol 1988;90: Address requests for reprints to: T. C. Northfield, M.D., Department 16. Iwahi T, Satoh H, Nakano M, Iwasaki T, Yamazaki T, Kubo K, et of Medicine, St George s Hospital Medical School, Cranmer Terrace, al. Lansoprazole, a novel benzimidazole proton pump inhibitor London SW17 ORE, England. Fax: (44) and its related compounds have selective activity against Helico- Supported by the Astra Foundation Research Fellowship for Gastrobacter pylori. Antimicrob Agents Chemother 1991;35: enterology (to S.K.).