Intended for healthcare professionals


Randomised double blind controlled study of recurrence of gastric ulcer after treatment for eradication of Helicobacter pylori infection

BMJ 1997; 314 doi: (Published 22 February 1997) Cite this as: BMJ 1997;314:565
  1. A T R Axon, professor of gastroenterologya,
  2. C A Ó'Moráin, consultant gastroenterologistb,
  3. K D Bardhan, consultant physicianc,
  4. J P Crowe, consultant gastroenterologistd,
  5. M F Dixon, reader in gastrointestinal pathologya,
  6. A D Beattie, consultant physiciane,
  7. R P H Thompson, consultant physicianf,
  8. P M Smith, consultant physiciang,
  9. F D Hollanders, consultant physician and gastroenterologisth,
  10. J H Baron, consultant physician and gastroenterologisti,
  11. D A F Lynch, consultant physiciank,
  12. D S Tompkins, directorj,
  13. H Birrell, clinical research scientistl,
  14. K R W Gillon, head of projectsl
  1. a Leeds General Infirmary, Leeds LS1 3EX
  2. b Meath and Adelaide Hospitals, Dublin 8, Republic of Ireland
  3. c Rotherham District General Hospital, Rotherham S60 2UD
  4. d Mater Misericordiae Hospital, Dublin 7, Republic of Ireland
  5. e Southern General Hospital, Glasgow G51 4TF
  6. f St Thomas' Hospital, London SE1 7EH
  7. g Llandough Hospital, South Glamorgan CF64 2XX
  8. h Oldchurch Hospital, Romford, Essex RM7 0BE
  9. i St Mary's Hospital, London W2 1NY
  10. j Leeds Public Health Laboratory, Leeds LS15 7PR
  11. k Blackburn, Hyndburn and Ribble Valley Health Care NHS Trust, Blackburn, Lancashire BB2 3LR
  12. l Astra Clinical Research Unit, Edinburgh EH7 4HG
  1. Correspondence to: Professor Axon
  • Accepted 6 December 1996


Objective: To determine whether eradication of Helicobacter pylori infection reduces recurrence of benign gastric ulceration.

Design: Randomised, double blind, controlled study. Patients were randomised in a 1:2 ratio to either omeprazole 40 mg once daily for eight weeks or the same treatment plus amoxycillin 750 mg twice daily for weeks 7 and 8. A 12 month untreated follow up ensued.

Setting: Teaching and district general hospitals between 1991 and 1994.

Subjects: 107 patients with benign gastric ulcer associated with H pylori.

Main outcome measures: Endoscopically confirmed relapse with gastric ulcer (analysed with life table methods), H pylori eradication, and healing of gastric ulcers (Mantel-Haenszel test).

Results: 172 patients were enrolled. Malignancy was diagnosed in 19; 24 were not infected with H pylori; four withdrew because of adverse events; and 18 failed to attend for start of treatment, leaving 107 patients eligible for analysis (35 omeprazole alone; 72 omeprazole plus amoxycillin). In the omeprazole/amoxycillin group 93% (67/72; 95% confidence interval 84% to 98%) of gastric ulcers healed and 83% (29/35; 66% to 94%) in the omeprazole group (P=0.103). Eradication of H pylori was 58% (42/72; 46% to 70%) and 6% (2/35; 1% to 19%) (P<0.001) and relapse after treatment was 22% (16/72) and 49% (17/35) (life table analysis, P<0.001), in the two groups, respectively. The recurrence rates were 7% (3/44) after successful H pylori eradication and 48% (30/63) in those who continued to be infected (P<0.001).

Conclusions: Eradication of H pylori reduces relapse with gastric ulcer over one year. Eradication rates achieved with this regimen, however, are too low for it to be recommended for routine use.

Key messages

  • The occurrence of gastric ulcer is positively associated with H pylori infection

  • Treatment to eradicate H pylori (omeprazole 40 mg once daily and amoxycillin 750 mg twice daily) reduces recurrence of gastric ulcers compared with omeprazole treatment alone

  • Regardless of the treatment given, rates of recurrence of gastric ulcers were lower in patients with successful H pylori eradication compared with those who continued to be infected

  • The rate of eradication of H pylori was 58% in the omeprazole/amoxycillin treatment group

  • Rates of healing of gastric ulcers were not significantly higher in those patients who received eradication treatment


There is a strong association between the risk of developing peptic ulcers and infection of the gastric and duodenal mucosa by Helicobacter pylori.1 2 Over 90% of duodenal ulcers are associated with H pylori infection, but only about 70% of gastric ulcers exhibit this association.3 The remainder are caused mainly by treatment with non-steroidal anti-inflammatory drugs. Although most gastric ulcers can be healed by antisecretory treatment, many will relapse when treatment is stopped. Therefore, continuous antisecretory treatment is taken by those with relapsing ulcers. Eradication of H pylori infection in patients with duodenal ulcer greatly reduces the rate of relapse.3 At the time we started this study (1991), however, there were no data on the effect of eradication treatment on gastric ulcer relapse. Therefore, we set out to determine whether recurrence could be reduced after such treatment.

Amoxycillin treatment in combination with omeprazole eradicates H pylori in a substantial proportion of patients with minimal side effects and without the development of antibiotic resistance, and so this dual treatment was chosen as the eradication treatment.

Patients and methods

The study was approved by the ethics committees of all participating centres. Consecutive patients with gastric ulcers confirmed endoscopically were invited to enter the study. Patients with a benign gastric ulcer at least 5 mm in diameter but without other active upper gastrointestinal disease or relevant cardiovascular, renal, or liver disease were included. Intake of bismuth compounds or antibiotics within the previous month or regular intake of non-steroidal anti-inflammatory drugs within the past two months were exclusion criteria.

The study was a double blind, double dummy design of eight weeks' healing with omeprazole and H pylori eradication treatment followed by one year of untreated follow up. H pylori status was assessed at the first and subsequent endoscopy visits by histology and culture. Patients who were infected with H pylori were randomised so that two thirds received omeprazole 40 mg once daily for eight weeks with amoxycillin 750 mg twice daily for weeks 7 and 8 and one third received omeprazole 40 mg once daily for eight weeks with dummy for weeks 7 and 8 (see fig 1). Randomisation was performed at each centre by allocating consecutive patients to the next available treatment pack from a computer generated randomisation list. Randomisation was performed on entry to the study rather than immediately before eradication treatment to ensure an even randomisation as some but not all patients would have had healed ulcers after six weeks of antisecretory treatment. Relapse was defined as an endoscopically verified gastric ulcer, and this resulted in withdrawal from the study.

Fig 1
Fig 1

Study design and numbers of patients recruited

H pylori was considered to have been eradicated if both histology and culture yielded negative results one month after the end of the initial treatment. H pylori status was also assessed at six and 12 months of follow up or on ulcer relapse. If either histology or culture yielded positive results at any visit the patient was considered to be positive for H pylori.

Biopsies for assessment of H pylori and ulcer malignancy

Biopsy samples were taken from the ulcer edge to exclude malignancy. Two antral samples were taken and frozen at -70°C or in liquid nitrogen for microbiological assessment at the microbiology department, Bradford Royal Infirmary. They were ground and a homogenate made which was inoculated on Columbia agar with 7% saponin-lysed horse blood, 1% Vitox and selective antibiotics (SR147, Oxoid), and on non-selective blood agar. Plates were incubated microaerobically at 37°C in a variable atmosphere incubator for up to seven days.

One sample was taken from each of the anterior wall and roof of the duodenum, posterior and anterior walls of the antrum 2 cm proximal to the pylorus, and posterior and anterior walls of the mid-part of the corpus. They were placed in 10% neutral buffered formalin and analysed at the pathology department, Leeds General Infirmary. The samples were sectioned at three levels and assessed by the modified Giemsa technique for H pylori.

Statistical methods

To calculate the number of patients required it was assumed that H pylori eradication would be 65% in the omeprazole/amoxycillin treatment group and 5% in the omeprazole treatment group and, additionally, that the risk of relapse by 12 months would be 70% in patients who were infected with H pylori and 20% those who were not. This gave risks of 0.675 and 0.375 for the omeprazole alone and omeprazole/amoxycillin group, respectively. With a sample size of 45 and 90 (135 total) and a two sided significance test at the 5% significance level, the test power of the null hypothesis would be 80% in the survival analysis. A total of 172 patients were to be recruited to allow for those lost to follow up during the assessment period. The test power for comparison of risks of relapse in patients with or without H pylori would be about 99%.

Data were analysed on an all patients treated basis because the primary objective of the study was to establish whether the eradication of H pylori infection reduced the recurrence of benign gastric ulceration rather than to compare the effectiveness of two treatments in ulcer healing and H pylori eradication. Nevertheless, for the sake of completeness an intention to treat analysis was also done for gastric ulcer healing and H pylori eradication.

Ulcer healing and H pylori eradication were analysed by a Mantel-Haenszel test stratified by centre. Patients were excluded from these analyses if they did not have a gastric ulcer, if they had a malignant gastric ulcer, if the intake of the trial drug could not be confirmed, or if they were not infected with H pylori at baseline. All patients with data from the visit one month after stopping treatment and those in whom H pylori status could not be confirmed were also included in the analysis of H pylori eradication rates. The latter patients were included in the analysis as not eradicated.

The main end points of the study-verified ulcer relapse or remission at the one year visit-were analysed by survival analysis with time in remission compared by log rank test. Patients were excluded from the survival analysis if they were not infected with H pylori at the baseline endoscopy, when the outcome of healing treatment could not be confirmed, or if they withdrew at the end of treatment with a healed ulcer. When ulcers were unhealed at the end of treatment patients were included in the survival analysis as relapses at day zero.


A total of 172 patients entered the study. Figure 1 shows the numbers of patients enrolled and randomised, including reasons for exclusion. Of those who entered the study 107 were included in the all patients treated analysis, 72 having received omeprazole and amoxycillin and 35 omeprazole and dummy. The only notable difference in the baseline characteristics was a predominance of men in the omeprazole/amoxycillin group (table 1).

Table 1

Baseline characteristics of patients according to treatment

View this table:

Gastric ulcer healing and H pylori eradication

Table 2 and Table 3 show the results of the intention to treat analyses of gastric ulcer healing and H pylori eradication. These data do not include randomised individuals who were found to have malignancy or who did not have H pylori infection at baseline as there was never any intention to treat them with antibiotics.

Table 2

Healing rates for gastric ulcer according to treatment group

View this table:
Table 3

Eradication rates for Helicobacter pylori according to treatment group

View this table:

Gastric ulcer healing-all patients treated analysis

Data were not available for five patients (three omeprazole/amoxycillin and two omeprazole) at the end of healing treatment, and they are included as treatment failures. Overall, healing of gastric ulcer was 88%, and there was no significant difference (P=0.103) between the groups (table 2).

H pylori eradication-intention to treat analysis

H pylori status could not be confirmed at the visit one month after treatment in 24 patients (11 were lost to follow up, six were unhealed, three had symptoms of ulcer and were withdrawn, two had complications of their ulcer disease and were withdrawn, one had an adverse event, and one violated the protocol). These patients are included in the analysis as not eradicated. H pylori eradication was confirmed in 42 (58%) patients who received omeprazole/amoxycillin and in 2 (6%) who received only omeprazole (table 3).

Gastric ulcer recurrence

Twenty two per cent (16/72) of patients who received omeprazole/amoxycillin relapsed during the one year follow up period compared with 49% (17/35) of patients who received omeprazole alone. Patients who received eradication treatment were more likely to be in remission than those who received omeprazole alone. The life curves for the probability of relapse were significantly different (fig 2; P<0.001).

Fig 2
Fig 2

Life table remission curves according to treatment group

Recurrence of gastric ulcers was compared in patients in whom eradication of H pylori was successful with those who remained infected with H pylori, regardless of the treatment given. Ulcer relapse was 7% (3/44) versus 48% (30/63) in the two groups, respectively. Once again, there was a highly significant difference when the life curves were compared (fig 3; P<0.001). None of the three patients in whom H pylori was eradicated and who later relapsed were reinfected at the time of their relapse.

Fig 3
Fig 3

Life table remission curves according to H pylori status

The treatments were well tolerated, there being only one serious adverse event-a myocardial infarction in a patient taking omeprazole alone.


In this study the recruitment criteria excluded patients taking non-steroidal anti-inflammatory drugs. Nineteen of the 172 (11%) patients recruited were eventually found to have malignancy, but of the 153 remaining, 129 (84%) were infected with H pylori. These patients were randomised to eradication treatment or simple acid suppression. The eradication treatment chosen was expected to be successful in around two thirds, and for that reason, to complete the study with about equal numbers of patients with and without H pylori infection, randomisation was undertaken in a 1:2 ratio.

When we consider those patients in whom we had definite evidence of either eradication or non-eradication the numbers were 44 and 63 respectively. The recurrence of ulcers over the next 12 months in these groups was 7% versus 48%. The Mantel-Haenszel assessment (see figure 3) indicates that there was a substantially better prognosis in those in whom H pylori eradication had been achieved compared with those in whom the organism persisted. These data suggest that the eradication of H pylori from patients with gastric ulcer is associated with a lower rate of relapse over the next 12 months. This implies that the eradication of H pylori changes the natural history of gastric ulcer and that it is, in part, responsible for the disease in most patients with this condition.

The major statistical comparison was between those patients treated with an eradication regimen and one designed simply to reduce acid secretion. Statistically this is appropriate because it compares patient groups that have been independently randomised as opposed to groups of patients that have been selected on the basis of their eradication status. With this analysis there was a significant advantage for the eradication treatment (see table 3 and fig 2). These data therefore confirm that the use of a regimen that eradicates H pylori is more effective in the prevention of ulcer relapse than one designed merely to heal the ulcer by acid suppression.

Other studies

There have been two recent reports of the effect of eradication treatment on gastric ulcer recurrence.4 5 In 1995 Seppala et al obtained similar results to our own over a 12 month period with 7% of ulcer relapses in those who were not infected with H pylori and 47% in those who were.4 In their study the eradication of H pylori improved healing of gastric ulcers as well as relapse. Their eradication regimen utilised colloidal bismuth subcitrate, however, which has a cytoprotective effect on gastric mucosa, and there is evidence that bismuth may accumulate and be slowly eliminated. The difference in relapse rates in that study, therefore, might have been the result of factors other than H pylori eradication.

In the second study, from Hong Kong, results were similar to ours for eradicated versus non-eradicated patients but were better than those we report within treatment groups. This may reflect the high eradication results achieved in their study.5


The data presented in this paper show that the eradication of H pylori from patients with gastric ulcer substantially improves their long term prognosis. The H pylori eradication regimen in use at the start of this study was not as effective as the current regimens. The combination of an acid pump inhibitor combined with two antibiotics (from clarithromycin, amoxycillin, or metronidazole) provides around 90% eradication after only a seven day course. The use of these eradication regimens in combination with ulcer healing treatment should lead not only to better rates of ulcer healing and lower rates of relapse but to a reduction in the long term complications associated with gastric ulceration. Nevertheless, of the 172 patients with apparently benign ulcers recruited to this study 19 were found on biopsy to have gastric cancers. It remains essential, therefore, that in all patients with gastric ulcer, whether or not they are treated for H pylori eradication, a further endoscopy is done to ensure ulcer healing and to take biopsies from the area of the ulcer scar so that potentially curable cancers are not missed.


We thank Sharon Cocker for preparing the manuscript.

Funding: The study was supported by a grant from the Astra Clinical Research Unit, Edinburgh.

Conflict of interest: J H Baron previously undertook consultancy work for Astra.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.