Cure of Helicobacter pylori infection does not improve symptoms in non-ulcer dyspepsia patients—a double-blind placebo-controlled study


Sato Department of Gastroenterology, Juntendo University, School of Medicine, 2–1-1 Hongo, Bunkyo-ku, Tokyo, Japan 113–8421 E-mail:



It remains controversial whether the cure of H. pylori infection improves NUD symptoms.


To conduct a double-blind placebo-controlled single centre study with concealed allocation to investigate this question.

Patients and methods:

Ninety NUD patients with H. pylori infection were randomly assigned to either the treatment group (50 patients) or placebo group (40 patients). The treatment group received omeprazole, amoxycillin, clarithromycin and the placebo group received omeprazole and placebos for 7 days. Symptoms were assessed every week for up to 12 weeks after completion of medication by a symptom questionnaire. Alteration of histological parameters for gastritis was also evaluated.


The infection was cured in 41 out of 48 patients in the treatment group and none in the placebo group. There was no significant difference in the mean symptom scores at any assessment point up to 12 weeks between the treatment and placebo groups. Regarding histological parameters, activity and inflammation, not atrophy or intestinal metaplasia, were significantly improved in the treatment group.


Although histological parameters were significantly improved in the treatment group, there was no significant improvement in symptoms of NUD in the treatment group compared to placebo.


Since cure of Helicobacter pylori infection has been shown to result in a dramatic reduction in the recurrence rate of peptic ulcers, curative therapy for this bacteria in patients with peptic ulcer diseases has been performed in many countries.1[2][3]–4

Non-ulcer dyspepsia (NUD) is a multifactorial complex of diseases which is defined on the basis of symptomatology alone.5 The therapeutic approach to NUD is still limited, and includes gastric acid suppressing agents, prokinetics, psychotropic medication and antinociceptive agents.6 Accordingly, therapeutic strategies are not well established, which makes decisions regarding treatment difficult and empirical. About a half of NUD patients have H. pylori infection and are therefore likely to have histological gastritis.7 Cure of H. pylori infection has been shown to improve histological gastritis in these patients;8 if cure of histological gastritis would improve NUD symptoms, we would have another therapeutic option for treatment of NUD. Several studies have referred to a possible link between H. pylori infection and NUD symptoms.9, 10 However, whether cure of the infection actually improves symptoms in NUD patients remains highly controversial.11, 12 Accordingly, we conducted a double-blind placebo-controlled study to approach this practical and important clinical question.


Study design

This study was designed as a double-blind placebo-controlled single centre study with concealed allocation. The study protocol and the procedure for the 13C urea breath test was approved by the Ethical Committee of Juntendo University Hospital. Written informed consent was obtained from all subjects before enrolment into the study. Placebos were made in the Department of Pharmacy at our hospital and no financial support was provided by any pharmaceutical company.

Subjects were NUD patients with H. pylori infection who visited our hospital. Definition of NUD followed the criteria of the AGA working party, although patients with only symptoms of gastro-oesophageal reflux were not included in this study.5 Patients were required to have had dyspeptic symptoms at least for 4 weeks. Before entry into this study, all patients underwent endoscopy to confirm that they did not have any localized lesions. Patients were excluded if they had endoscopic evidence of peptic ulcer scar or gastric erosions. Absence of hepatobiliary or pancreatic diseases was confirmed by abdominal ultrasonography or CT scanning. Patients who were pregnant, had undergone gastric resection or had been treated previously for H. pylori infection were also excluded.

Patients were enrolled into the study after confirming that they met inclusion criteria (positive H. pylori infection and conformance with NUD criteria) and after obtaining written informed consent. The subjects were consective NUD patients, although there were several patients who refused to participate in this study. Ninety patients were randomly assigned to either the treatment group (50 patients) or placebo group (40 patients). To the patients in the treatment group, omeprazole 20 mg b.d., amoxycillin 500 mg t.d.s., and clarithromycin 200 mg b.d. were prescribed for 7 days and to the patients in the placebo group, omeprazole 20 mg b.d., and placebo of amoxycillin and clarithromycin were given for the same period. This combination of the proton pump inhibitor and treatment was previously reported to provide a sufficient cure rate in our patient population.13 Symptom questionnaires were filled out before treatment, just after completion of the treatment, and at the end of every week from 1 week to 12 weeks after treatment. Compliance of the treatment drug was recorded in the questionnaire sheet and by interview from 4 to 6 weeks after treatment. We also interviewed the patients at this time to ask whether they had problems in pursuing the clinical trial or if there were problems in using the questionnaires. Twelve weeks after treatment, 13C urea breath test and endoscopy were performed to investigate H. pylori status and alteration of the degree of gastritis. The randomization code was disclosed only after completion of the 13C urea breath test and endoscopy by the last enrolled patient. Changes in symptom scores and degree of histological parameters for gastritis were studied.

All patients were asked to refrain from taking any medications that would affect gastrointestinal symptoms at least a week prior to the interview regarding symptoms, before the start of treatment and during the study period. Use of any medicines, including antibiotics or NSAIDs, with the exception of daily use of drugs such as antihypertensive agents was prohibited. Anti-acid (histamine H2 receptor antagonist) or anticholinergic agents were allowed to be used on demand when symptoms were intolerable.

Evaluation of H. pylori status and histological gastritis

Before entry into the study and 3 months after treatment, endoscopy was performed in all patients. Two biopsies, one from the greater curvature of the antrum and the other from that of the gastric body, were obtained to evaluate H. pylori status and degree of gastritis. Biopsied specimens were stained by both haematoxylin and eosin (H&E) and Giemsa staining. In addition, the 13C urea breath test was performed before and 3 months after treatment. Patients with at least one positive result of histological examination or breath test were regarded as H. pylori-positive and those with negative results in both of these tests were regarded as H. pylori-negative. Two patients refused to undergo endoscopy 3 months after treatment, although both took the breath test; their H. pylori status was evaluated only by the breath test value and they were included only in symptom analysis.

We adapted the modified 13C urea breath test procedure as previously reported.14 Briefly, the procedure was carried out under fasting conditions, using 100 mg of 13C urea ingested, in a sitting position and with tooth brushing before dosing, breath sampling before and 20 min after administration of 13C urea, and with a 5.0‰ cut-off value. This procedure has been reported to provide 96.7% sensitivity and 97.6% specificity.15

Histological findings were determined in biopsied materials using H&E and Giemsa stained sections examined by one pathologist (S.H.) who had no knowledge of any study information. The degree of infiltration of polymorphoneutrophils was examined as a marker of activity of gastritis, and that of monocytes as a marker of inflammation of gastritis. Degree of atrophy and intestinal metaplasia were examined in the H&E stained sections and the degree of bacterial colonization was examined in Giemsa stained sections. Histological findings were separately studied on biopsied specimens taken from the antrum and gastric body and were graded according to severity by the Sydney system, that is: 0, normal; 1, mild; 2, moderate; and 3, marked.16

Symptom evaluation

Patients’ symptoms were evaluated by symptom questionnaires which consisted of the following 12 items: (1) heartburn, (2) retrosternal discomfort, (3) epigastric discomfort, (4) nausea or vomiting, (5) bloating, (6) early satiety, (7) appetite loss, (8) belching, (9) morning discomfort, (10) epigastric pain, (11) pain or discomfort which was relieved after a meal, and (12) others. All items were rated in seven steps regarding intensity, frequency, duration and impact on daily living following the Gastrointestinal Symptom Rating Scale (GSRS), that is: grade 0, absence; 1, mild; 2, moderate; 3, severe; and half steps between each rating.17 The sum of the scores of these items was adopted as the symptom score for each patient. Each symptom was scored at each assessment point (pre- and immediately post-treatment, and weekly 1–12 weeks after treatment) by the patient. The symptom scores were validated over a week; one of the authors (H.M.) explained to each patient how to score before the start of treatment.

Data analysis

Data analysis was carried out in the intention-to-treat (ITT) manner. Five patients were excluded from ITT analysis. Two of these were lost to follow-up and symptom questionnaires could not be recovered, two received additional curative therapy in another hospital and one was an H. pylori-negative entry.

For an analysis of improvement in symptoms, the mean of the total symptom scores at each assessment point were compared between the treatment group and placebo group, and between the infection-cured group and placebo group. Mean symptom scores for each item at each assessment point were also compared between those groups. In addition, improvement of symptoms according to three different categories of NUD criteria, reflux-like symptoms, dysmotility-like symptoms and ulcer-like symptoms, were evaluated and compared. Here, heartburn and retrosternal discomfort were regarded as reflux-like symptoms; epigastric discomfort, nausea or vomiting, bloating, early satiety, appetite loss, belching, and morning discomfort were regarded as dysmotility-like symptoms; and epigastric pain and pain or discomfort that was relieved after eating were regarded as ulcer-like symptoms.

In addition to the mean symptom score, the ratio of patients in whom symptoms were improved was reported by comparing pre-treatment and final (at 12 weeks) symptom scores. The proportion of patients with no or minimal symptoms (symptom score, 0 or 1) were also compared among the treatment, infection-cured and placebo groups.

To quantify histological alteration, graded histological scores before and after treatment were compared between the treatment and placebo groups and the infection-cured and placebo groups.

Data were expressed by mean ± S.E.M. For statistical analysis, the Student’s t-test, one-way analysis of variance, Fisher’s exact test and Wilcoxon’s rank sum test were used and P < 0.05 indicated a significant difference.


The 90 enrolled patients included 52 males and 48 females with a mean age of 51.2 ± 1.2 years (mean ± S.E.M.; range, 20–73 years). Five patients, two from the treatment group and three from the placebo group, were excluded from analysis. As a result, there were 48 patients in the treatment group and 37 patients in the placebo group. There were no significant differences upon entry in gender ratio, mean age, mean symptom scores and 13C urea breath values at 20 min between the treatment and placebo groups. Patient dispositions and profiles are shown in Table 1.

Table 1.  . Disposition and profile of the patients Thumbnail image of

Treatment efficacy

Infection was cured in 41 of the 48 patients in the treatment group (per protocol based cure rate was 85.4%). The infection persisted in all patients in the placebo group. Adverse effects were recorded in 27 of 85 patients, of which 17 (35.4%) were in the treatment group and 10 (27.0%) in the placebo group. No significant difference was noted between the groups. Compliance to the drug regimen was excellent, with 79 of 85 patients showing 100% compliance; the remaining six complied by taking more than 90% of the drugs. Adverse effects and compliance are shown in Table 2.

Table 2.  . Adverse effects and compliance Thumbnail image of

Improvement of symptoms

Chronological changes of mean symptom scores up to 12 weeks after completion of treatment in the treatment and placebo groups are shown in Figure 1. No significant difference was shown at any point. Such changes in the infection-cured and placebo groups are shown in Figure 2, and there were also no significant differences at any point. Improvement in mean scores of symptoms representing three different subtypes of NUD were similar in treatment and placebo groups and in infection-cured and placebo groups with no significant difference at any of the assessment points ( Figure 3).

Figure 1.

. Improvement of mean symptom scores in the treatment group and placebo group. There were no significant differences in symptom scores in any point.

Figure 2.

. Improvement of mean symptom scores in infection-cured group and placebo (infection not cured) group. There were no significant difference in symptom scores in any point.

Figure 3.

. Alteration of mean symptom scores according to symptoms representing three subgroups of non-ulcer dyspepsia. There were no significant differences between treatment and placebo groups.

Mean symptom scores for each item at each assessment point were compared between the treatment and placebo groups and the infection-cured and placebo groups. There were no significant differences at any of the 196 assessment points except for one. The mean symptom score for nausea or vomiting at pre-treatment was 0.54 ± 0.12 in the placebo group, which is significantly higher than that in the treatment group, 0.96 ± 0.13, and infection-cured group, 0.98 ± 0.15; P=0.03.

The ratio of patients whose final symptom score was improved was 85.4% (41 out of 48, 95% CI: 72–94%) in the treatment group, 82.9% (34 out of 41, 95% CI: 63–93%) in the infection-cured group and 89.2% (33 out of 37, 95% CI: 75–97%) in the placebo group, with no significant difference being noted (P=0.75). The proportion of patients with no or minimal symptoms (symptom score 0 or 1) at 12 week in the treatment, infection-cured and placebo groups was 31.3% (15 out of 48, 95% CI: 19–46), 34.5% (14 out of 41, 95% CI: 20–51%) and 24.3% (9 out of 37, 95% CI: 12–41%), respectively. There were no statistically significant differences among these groups.

Alteration of histological parameters

Biopsy specimens obtained before and 3 months after treatment were histologically assessed. Mean value of histological parameters in biopsied samples are shown in Table 3. Histological parameters for activity, inflammation and H. pylori colonization in the post-treatment biopsy specimen taken from both the gastric body and antrum were significantly improved compared to those at pre-treatment in the treatment and infection-cured groups, although parameter for atrophy in the antrum was increased after treatment. On the other hand, there were no significant differences between scores of pre- and post-treatment in the placebo group. No statistically significant changes were seen at any point in the parametric scores for atrophy and intestinal metaplasia, except in the antrum of the infection-cured group.

Table 3.  . Alteration of histological parametric scores in biopsied materials from gastric body and antrum Thumbnail image of


In published studies regarding the effects of curative therapy for H. pylori infection and symptoms in NUD patients, results have been conflicting.11, 12 Since NUD patients have a high placebo response rate and the disease is defined only by symptoms, a double-blind placebo-controlled study is necessary to investigate this question.12, 18 There have been several double-blind studies regarding this issue; however, most early studies included bismuth among the therapeutic drugs.12 Bismuth yields a black stool, which might interfere with adequate blinding of the study. Furthermore, the low eradication rate of the regimens complicated the study design, since a prolonged treatment period itself may affect the relief of symptoms.19, 20 Taking these aspects into consideration, until recently there have been only a few studies which mainly focused on improvement of NUD symptoms using a simple protocol with powerful eradication regimens for H. pylori infection.21[22][23]–24 We performed a double-blind placebo-controlled study with concealed allocation using the omeprazole–amoxycillin–clarithromycin (OAC) regimen, a new triple therapy for cure of H. pylori infection, which has been shown to provide a sufficient cure rate with few adverse effects in our patient population.25[26]–27 The lack of a significant difference in the rate of adverse effects between treatment and placebo groups in our study ensured adequate blinding. Furthermore, this regimen is not only of short duration but has a high cure rate, which contributes to the simplicity of the study design.

Another merit of this study is the weekly assessment of symptoms by which chronological changes of symptom scores could be clearly demonstrated; in most studies, such assessment points usually numbered only 2 or 3; before and at the end of the study, and sometimes with an additional point in the middle of the course.12 We assume that symptom changes will be enhanced through chronological observation by weekly self- assessment of symptoms. The symptom score used consisted of 12 items (maximal score, 36 points) and was devised by modifying the already validated Gastrointestinal Symptom Rating Scale, which is for irritable bowel syndrome and peptic ulcer disease.17 Our items consisted of 11 symptoms which are often used as expression of upper GI dyspeptic events in our patient population. The final item was listed as ‘others’, which was used to designate complaints that did not fit into other named categories but were suitable for non-specific types of NUD. In summary, the total symptom score should represent how much discomfort the patient experienced.

As to improvement of gastritis, our results demonstrated marked improvement of histological parameters in the treatment group but not in the placebo group. As reported in previous studies, activity of gastritis was dramatically decreased and inflammation of gastritis was also significantly improved in 3 months, although improvement in the latter was not as great as improvement in the former.28, 29 On the other hand, parameters for atrophy and intestinal metaplasia were not changed during this observation period except for atrophy in the antrum. This observation suggested that activity and inflammation of gastritis do not correlate with severity of NUD symptoms.

Our present study revealed that cure of H. pylori infection did not improve symptoms in NUD patients in our patient population in either the treatment group or the infection-cured group in comparison with the placebo group. We also investigated whether some specific NUD symptoms responded to eradication treatment but failed to identify any specific symptoms that improved after curative therapy, as earlier studies had also suggested.30[31]–32 Our subjects were those with NUD symptoms who met AGA inclusion and exclusion criteria and included those with both rather mild and severe complaints (range of symptom scores 3–22). However, in focusing on patients with moderate to severe symptoms (symptom scores > 10), there was also no significant difference between treatment and placebo groups (data not shown), suggesting that our data could be practically generalized. In order to ascertain outcome measurement, long-term observation, for at least 6 months or a year, may be necessary to assess improvement in symptoms by cure of the infection because of the high response rate of NUD patients to placebo.11, 18 Since the observation period of this study was 12 weeks, we cannot discuss the long-term effect of cure of H. pylori on NUD symptoms. Yet, even in determining its short- or intermediate-term effect, it is still controversial whether cure of H. pylori affects NUD symptoms. Goh et al. reported improvement of the mean symptom score in 4 weeks, and Kang et al. and Vaira et al. noted improvement 8 weeks after treatment.33[34]–35 In a recent study, Gilvarry et al. reported that in subjects in which H. pylori was eradicated symptoms significantly improved not only at 6 months and 1 years but also at 8 weeks.21 In contrast, Morgan et al. reported no improvement of symptoms in 6 weeks and Glupczynski et al. in 8 weeks.36, 37

Very recently, three large-scale long-term double-blind placebo-controlled studies using proton pump inhibitor-based triple therapy have been published, the results of which were conflicting. In the article of McColl et al. that reported improvement in symptoms after eradication of infection, their outcome measurement for the effect of eradication therapy was the proportion of patients with no or minimal symptoms at the end of the study and they failed to show greater improvement in mean symptom scores between the infection-cured and placebo groups.22 In contrast, Blum et al. and Talley et al. reported that no statistical differences were noted between the treatment and placebo groups in either the proportion of patients with no or minimal symptoms or in the mean symptom score.23, 24 Neither was any difference found in our two groups concerning chronological change in mean symptom score and ratio of patients in whom symptoms were improved. In addition, there was no significant difference between two groups regarding ratio of patients with no or minimal symptoms at the final point, although statistical significance might appear if the sample size was larger. Accordingly, our observation clearly showed that cure of H. pylori infection does not contribute to improvement of symptoms in non-ulcer dyspepsia patients at least 3 months after treatment in our patient population.


The authors express great thanks to senior fellows of Department of Gastroenterology, Juntendo University, for cooperation for this work. We also thank to Miss Hitomi Maeda for excellent secretarial assistance.