Eradication of Helicobacter pylori in functional dyspepsia: randomised double blind placebo controlled trial with 12 months' follow up

Protocol

Study population— Dyspepsia was defined as pain or discomfort centred in the upper abdomen.1 We enrolled adult patients with dyspepsia for at least 3 months, normal endoscopic findings, and a positive result for H pylori on a screening test (Helisal, Cortecs Diagnostics, UK). Patients with oesophagitis (any mucosal break), Barrett's oesophagus, gastric or duodenal ulceration, duodenal erosions, malignancy, more than five gastric erosions, or alarm symptoms were excluded. H₂ receptor antagonists, prostaglandins, or prokinetics during the 7 days before enrolment, or proton-pump inhibitors, antibiotics, or bismuth during the 30 days before enrolment, were not permitted. Patients with documented peptic ulcer disease or gastro-oesophageal reflux disease were excluded.

Run-in period— After endoscopy, patients were required to fill out a diary card with scores for their symptoms during a 7 day run-in period. Only patients who had at least 3 days of at least moderate dyspepsia symptoms were randomised. No study drug was dispensed during the run-in.

Treatment period— Patients underwent a breath test using urea labelled with carbon-13 at the randomisation visit. They were randomised to receive either omeprazole 20 mg twice daily, amoxicillin 1000 mg twice daily, and clarithromycin 500 mg twice daily or placebo for 1 week. If patients had taken at least 12 out of 14 doses of drug or placebo they were considered to be compliant; no patients were withdrawn from the study because of poor compliance.

Follow up period— The patients were followed up 1, 3, 6, 9, and 12 months after cessation of treatment. Diary cards (filled out the week before each visit) were collected at each visit, quality of life forms were filled out by the patients at the 6 and 12 month visit, and a urea breath test and upper endoscopy were performed at the 3 and 12 month visits. A weak antacid (with a neutralising capacity of around 13 mmol of hydrochloric acid per tablet) was dispensed at each visit and its consumption was recorded. During follow up, patients could receive treatment for dyspeptic symptoms from their doctor but all drugs used were recorded.

Primary outcome measures

Patients recorded the severity of their dyspepsia symptoms on diary cards using a validated Likert scale comprising 7 grades: none, minimal, mild, moderate, moderately severe, severe, very severe.14

At each endoscopic evaluation, two antral and two corpus biopsy specimens were obtained. Specimens were stained with haematoxylin and eosin and with the Steiner silver method.

The biopsies were histologically graded.15 All specimens were reviewed by an experienced gastrointestinal pathologist blinded to the treatment group. Urea breath testing was performed using a standard validated European protocol.16

At pre-entry, two test results for H pylori had to be positive; by screening test (Helisal rapid blood test or a rapid urease test) and by either urea breath testing or histological assessment.

After treatment, H pylori status was assessed at 3 and 12 months. If any of the gold standard assessments (urea breath test or histology) were positive, patients were considered to be positive for H pylori. If only one test result was available, the outcome of that test determined the H pylori status.

Secondary outcome measures

The gastrointestinal symptom rating scale was used to score dyspepsia symptoms. This validated instrument measures symptoms including abdominal pain. 17 18 The psychological general well being index was used to score the patients' quality of life. This validated instrument measures subjective well being.18^–20

Patients were subdivided into symptom subgroups on the basis of their responses to the gastrointestinal symptom rating scale. Ulcer-like dyspepsia was defined as at least moderate stomach pain and hunger pain in the week before follow up. Dysmotility-like dyspepsia was defined by two or more of at least moderate bloating, nausea, stomach rumblings, or belching in the week before follow up. The subgroups were not mutually exclusive.

Statistical analyses

Patients were excluded from the intention to treat analysis who were negative for H pylori at pre-entry or who were without any assessment of treatment efficacy after randomisation (fig 1). The treatment groups were compared for symptom relief with a Mantel-Haenszel test stratified by country and for healing of gastritis with a Mantel-Haenszel test stratified by baseline gastritis.

Fig 1

Flow of participants through trial

• Download figure
• Open in new tab
• Download powerpoint

Patients who reported on the diary card no more than minimal dyspepsia symptoms during any of the 7 days before the 12 month visit were considered a priori to be a treatment success.

Chronic gastritis was considered healed when both antrum and corpus specimens had an inflammation score of zero.15

The treatment groups were compared for change in the total score of the gastrointestinal symptom rating scale and psychological general well being index from the first visit to the last visit in the study, using the baseline value as a covariate in an analysis of covariance model.

With 275 patients, the power of the study was 94% provided the true proportions of responders was 20% and 40% in the two groups (assuming an α level of 0.05 based on a two sided χ² test). The placebo response was based on data for symptom turnover.21

Randomisation was in blocks of four in proportions of 1:1 according to a computer generated list.

Identical placebos were used. Investigators and patients were blinded to all data, including H pylori assessments after randomisation, until the study was fully completed.

Analysis

Symptom relief

By an intention to treat analysis, 32 patients (24%) in the treatment group and 31 patients (22%) in the placebo group were successfully treated (minimal or no dyspepsia) at 12 months (table 2). There was no significant difference in treatment success among those who were negative for H pylori (35, 29%) and those who remained positive for H pylori (28, 21%). At the 12 month follow up, no dyspepsia symptoms were reported by 20 patients (15%) in the treatment group and 16 patients (11%) in the placebo group. A similar proportion of patients in each treatment arm had no or minimal dyspepsia symptoms at each follow up (fig 2). The mean symptom score was not significantly different at each time point (fig 3). There was no inhomogeneity among countries (Breslow-Day test, P>0.20).

Fig 2

Proportion of patients with no more than minimal dyspepsia symptoms on basis of diary card data (per protocol analysis)

• Download figure
• Open in new tab
• Download powerpoint

Fig 3

Mean symptom score and SEM on diary cards at follow up visits

• Download figure
• Open in new tab
• Download powerpoint

Mean antacid consumption over 12 months did not differ significantly between treatment (0.53 tablets per day) and placebo groups (0.65). Five patients (one in the treatment group and four in the placebo group) had treatment success according to the diary cards but were considered treatment failures in the analysis because they took a gastrointestinal drug other than antacid within 2 weeks of the 12 month visit.

In the ulcer-like dyspepsia group, treatment success was reported by 17/68 patients (25%) in the treatment group and 12/58 patients (21%) in the placebo group. The corresponding results for dysmotility-like dyspepsia were 14/78 patients (18%) in the treatment group and 13/73 patients (18%) in the placebo group.

The change from baseline to last visit between the treatment groups was not significant for either the psychological general well being index or the gastrointestinal symptom rating scale (fig 4).

Fig 4

Intention to treat analysis showing mean (SEM) scores for quality of life and severity of symptoms in patients with functional dyspepsia. Possible psychological well being index scores (top) ranged from 132 (best) to 22 (worst); possible gastrointestinal symptom rating scale (bottom) mean scores ranged from 1 (no symptoms) to 7 (worst symptoms)

• Download figure
• Open in new tab
• Download powerpoint

Gastritis scores and symptom relief— There was no association between the severity of symptoms at baseline and gastritis scores on initial biopsies. Patients at follow up were subdivided regardless of treatment into those with a chronic gastritis score of 0 or 1 (none or mild gastritis) and those with a score of 2 or 3 (moderate or severe gastritis) in a secondary analysis. At the 12 month follow up, 41/127 patients (32%) with no or mild gastritis were treatment successes (no or minimal dyspepsia) compared with 21/123 patients (17%) with moderate or severe gastritis (P=0.008). This association was not explained by age. Of the 41 patients with none or mild gastritis at follow up, only nine had received placebo (of whom only one had complete resolution of gastritis and eight had mild gastritis).

Trial design issues

We aimed to overcome previous methodological limitations. 8 9 In particular, the outcome measures were valid and responsive to change. 14 17^–20 Scrupulous attention was paid to blinding of patients and investigators. Prospective assessment of symptoms reduced the issue of recall bias.9

Predictors of symptom relief

A persistent inflammatory response could promote the development of dyspepsia.11 We observed an association between healing of chronic gastritis and symptom relief but this secondary analysis requires confirmation.

A few studies have observed that treatment response was limited to those patients with ulcer-like dyspepsia, but no link between dyspepsia subgroups and H pylori eradication was evident in the present study. 13 24 Although our results may be generalisable to secondary care, we cannot exclude the possibility that such patients have more resistant symptoms than those in primary care where trials are needed.

Management implications— A popular management strategy in otherwise healthy young patients with uninvestigated dyspepsia is to non-invasively test for H pylori and to treat all infected cases.25 Although controversial, such a strategy may have a number of potential benefits.26 On the basis of our results, however, only a minority who are treated would be likely to gain long term symptomatic relief, because most infected patients with dyspepsia have functional dyspepsia rather than peptic ulcer disease.25