A randomized, double-blind trial of the efficacy and safety of 10 or 20 mg rabeprazole compared with 20 mg omeprazole in the maintenance of gastro-oesophageal reflux disease over 5 years


  • **

    Other members listed at the end of this article.

Correspondence to: Professor B. Thjodleifsson, Professor and Chief of Gastroenterology, University Hospital, Reykjavik, Iceland. E-mail: bjarnit@landspitali.is


Background : Gastro-oesophageal reflux disease has a chronic course, and often requires long-term treatment. Proton pump inhibitors are the treatment of choice for both acute and maintenance treatment, but little is known from randomized controlled trials of their effects beyond 1 year.

Aim : To compare the efficacy and safety of two doses of rabeprazole with 20 mg omeprazole in the maintenance treatment of erosive gastro-oesophageal reflux disease over 5 years.

Methods : Two hundred and forty-three patients who had previously responded to acute treatment for erosive gastro-oesophageal reflux disease were prospectively randomized to receive 5 years of treatment with rabeprazole (10 or 20 mg daily) or omeprazole (20 mg daily). The primary outcome measure was endoscopically confirmed relapse of erosive gastro-oesophageal reflux disease.

Results : One hundred and twenty-three patients (51%) completed all 5 years of the study, with similar completion rates in the three groups. Relapses occurred in nine of 78 (11.5%), eight of 82 (9.8%) and 11 of 83 (13.3%) patients in the rabeprazole 20 mg, rabeprazole 10 mg and omeprazole 20 mg groups, respectively. Gastric biopsy showed no evidence of any harmful effects. All treatments were well tolerated.

Conclusions : Rabeprazole 10 mg, rabeprazole 20 mg and omeprazole 20 mg all had similar efficacy in the maintenance treatment of gastro-oesophageal reflux disease. All three were safe and well tolerated during 5 years of treatment.


Gastro-oesophageal reflux disease (GERD) is an important public health problem in Western countries. Symptoms of GERD, such as acid regurgitation and heartburn, have a prevalence of almost 30% amongst adults in the UK,1 and patients usually have symptoms for many years.2 The availability of treatments that are effective and safe in long-term use is therefore highly desirable. Proton pump inhibitors have emerged as the treatment of choice,3 as they are more effective than H2-receptor antagonists (e.g. ranitidine) for short-term symptom relief,4 endoscopically demonstrated healing5 and long-term maintenance treatment.6 They are also more cost-effective than surgery.7

Rabeprazole is one of the newer proton pump inhibitors, and has a faster onset of antisecretory activity than omeprazole.8 It is as effective as omeprazole in acute9 and maintenance10 treatment of erosive GERD, and is also effective in the healing of gastric and duodenal ulcers.11 Its tolerability profile has been well studied in clinical trials, and has been found to be favourable.12 In particular, no dose adjustments are necessary in special populations, such as the elderly or patients with renal or mild to moderate hepatic disease.

Despite the importance of proton pump inhibitors in the maintenance treatment of GERD, little information is available from randomized trials about their long-term use. The majority of trials investigating maintenance treatment have lasted no more than 1 year,5 and studies lasting longer than this have generally used non-randomized, uncontrolled designs, making their interpretation difficult.13–15 This prospective study was performed to compare the efficacy and safety of two doses of rabeprazole (10 and 20 mg once daily) with the standard dose of omeprazole (20 mg once daily) in the maintenance treatment of erosive GERD over a 5-year period. The results of the first year of this study have been published previously;10 here, we report the results of the entire study.


Study design

This was a multi-centre, parallel-group, randomized, double-blind, comparative study, which took place in 21 centres in nine European countries. The study was approved by the ethics committee appropriate to each centre before any patients were enrolled at that centre, and was conducted in accordance with the Declaration of Helsinki. All patients gave written informed consent before entering the study. The study began as a 1-year maintenance study (protocol number NRRQ), which was subsequently extended with four 1-year extension studies (protocol numbers 310, 310-1, 310-2 and 310-3); the total study duration was therefore 5 years. Patients were followed up at weeks 4, 13, 26, 39 and 52 in the maintenance study, with endoscopy at baseline and at weeks 13, 26 and 52. In each of the extension studies, patients were followed up every 3 months, with endoscopy annually. Throughout the study, endoscopy could be performed at any time if patients had symptoms that suggested the relapse of GERD.


Patients were eligible for the maintenance study if they had a previous diagnosis of erosive GERD and had endoscopically confirmed healing within 90 days before the start of the study; 79% of the patients had previously been treated in a clinical trial comparing 20 mg rabeprazole with 20 mg omeprazole in the acute healing of erosive GERD.9 At inclusion into the acute healing study, the initial Hetzel–Dent oesophagitis grades were 2, 3 and 4 in 43%, 52% and 4% of the patients included in that study, respectively. At the end of the acute healing study, 93% of patients were healed and therefore potentially eligible for inclusion in the maintenance study. The distribution of oesophagitis grades in patients (21%) not originating from the healing study was 38%, 55% and 6% for Hetzel–Dent oesophagitis grades 2, 3 and 4, respectively. Male or female patients aged at least 18 years were included. The main exclusion criteria were: oesophageal strictures; primary motility disorders; previous gastrointestinal surgery; varices; pyloric stenosis; renal or hepatic insufficiency; cancer; active ulcer or bleeding; Zollinger–Ellison syndrome; pregnancy; and abnormal leucocyte counts, haemoglobin or blood urea nitrogen. Patients were eligible for the extension studies if they had completed the previous year of the study without endoscopically confirmed relapse of GERD.


Patients were randomly allocated to one of three treatments on entry to the study on the basis of a computer-generated randomization scheme. The treatments were rabeprazole 10 mg, rabeprazole 20 mg and omeprazole 20 mg. All treatments were taken once daily in the morning. Treatment was double-blind; to maintain blinding, all patients were given placebos that were identical in appearance to the two treatments that they did not receive. During the study, patients were not allowed to take H2-receptor antagonists, sucralfate, other proton pump inhibitors, prostaglandins, anticholinergics or motility agents. Antacids could be taken as rescue medication, and their use was recorded on diary cards.

In addition to the study medication, all patients were given lifestyle advice (avoid fatty foods, chocolate, peppermint and spearmint; abstain from eating 2–3 h before bedtime; decrease or discontinue smoking; limit consumption of caffeine and alcohol; raise the head of the bed by 15–20 cm).

Outcome measures

The primary outcome measure, prospectively specified, was endoscopically confirmed relapse of GERD. Relapse was defined as a score of at least two on the modified Hetzel–Dent grading scale in a patient who had not previously experienced a relapse.16 Secondary efficacy variables were time to endoscopic GERD relapse, daytime and night-time heartburn pain (assessed by the patient as none, mild, moderate, severe or terrible), overall physical well-being (assessed by the patient as very good, good, fair, poor or very poor), antacid use and impact on daily living (number of hours of normal activities missed as a result of GERD). Safety and tolerability were assessed by adverse events, routine laboratory safety tests, vital signs, physical examination, serum gastrin and gastric biopsy.


Biopsy samples were taken from the corpus and antrum after 13, 26 and 52 weeks, and annually thereafter. Formalin-fixed, paraffin-embedded samples were stained with haematoxylin–eosin, modified Giemsa (for Helicobacter pylori detection) and silver Grimelius (for endocrine cells of the oxyntic mucosa). Gastritis and H. pylori infection were graded according to the updated Sydney system;17 enterochromaffin-like cell changes were assessed in accordance with the criteria of Solcia et al.18 The presence of intestinal metaplasia was assessed as complete or incomplete, and a percentage score was given to quantify the amount of the mucosal area involved by intestinal-type glands.

Statistical methods

The sample size for the study was calculated to give 80% power to rule out an absolute difference in relapse rates between rabeprazole and omeprazole of 18%, based on two-sided significance tests at the 5% level, assuming a relapse rate of 20% for both rabeprazole and omeprazole after 6 months. The calculations assumed that the number of patients enrolled in the study would decrease as the study progressed.

The primary analysis was a comparison of relapse rates in the intention-to-treat population, with the calculation of the 95% confidence intervals for the difference in proportion of relapses between each rabeprazole dose and omeprazole. If patients withdrew from the study prematurely, their last valid observation on relapse was carried forward (last observation carried forward). Secondary analyses were based on endoscopies actually performed, and a survival analysis of the time to relapse. Results for heartburn at the end of the study are presented only for those patients who completed the study. No statistical hypothesis testing was carried out for the safety variables.


Patients studied

Of the 243 patients who entered the study, 123 (51%) completed all 5 years of treatment. The flow of patients through the study is shown in Figure 1. Baseline demographic variables were well matched amongst the three treatment groups (Table 1). In each of the participating countries, patients were reasonably evenly distributed amongst the three treatment groups.

Figure 1.

Flow of patients. *‘Did not continue’ refers to patients who completed the maintenance study and up to three of the subsequent extension studies, but did not continue into the following year's study.

Table 1.  Baseline demographics
 Rabeprazole 20 mg (n = 78)Rabeprazole 10 mg (n = 82)Omeprazole 20 mg (n = 83)
Mean age (range) (years)52 (20–75)52 (25–76)54 (20–83)
Number of females (%)24 (31)26 (32)31 (37)
Number of current smokers (%)15 (19)18 (22)20 (24)
Mean units of alcohol  per week (range)2.5 (0–15)3.6 (0–36)3.6 (0–30)

Primary efficacy variable

Relapse rates were low and similar in all three treatment groups (Table 2). Neither of the rabeprazole groups differed significantly from the omeprazole group. The Kaplan–Meier survival curve for the time to relapse is shown in Figure 2, which confirms the similarity of the three treatments. The survival function also did not differ significantly between either rabeprazole group and omeprazole (20 mg rabeprazole vs. 20 mg omeprazole: χ2 = 0.45, P = 0.50; 10 mg rabeprazole vs. 20 mg omeprazole: χ2 = 0.90, P = 0.34; log rank test).

Table 2.  Relapse rates during the study
 Number of relapsesRelapse rate (%)Difference from omeprazole (%) (95% CI)
Rabeprazole 20 mg 9/7811.5− 1.7 (− 11.8–8.5)
Rabeprazole 10 mg 8/829.8− 3.5 (− 13.2–6.2)
Omeprazole 20 mg11/8313.3 
Figure 2.

Kaplan–Meier graph of the relapse of gastro-oesophageal reflux disease.

Secondary efficacy variables

The analysis of the secondary efficacy variables confirmed the results of the primary variable. Control of heartburn symptoms was good and similar in the three treatment groups. At most visits, more than 90% of patients in all treatment groups reported no or only mild heartburn pain both during the day and at night. The severity of heartburn at baseline and at the end of 5 years is shown in Table 3. Most patients reported their overall physical well-being as ‘good’ or ‘very good’ throughout the study in all treatment groups, and most reported that their disease did not have any impact on their activities of daily living. The proportion of patients reporting any recent antacid use was below 10% at most visits in all treatment groups.

Table 3.  Number (%) of patients with different grades of daytime and night-time heartburn severity at the beginning and end of the study
 Rabeprazole 20 mg (n = 78)Rabeprazole 10 mg (n = 82)Omeprazole 20 mg (n = 83)
 None63 (81)38 (88)53 (65)36 (82)59 (71)33 (92)
 Mild9 (12)3 (7)16 (20)5 (11)11 (13)3 (8)
 Moderate5 (6)2 (5)10 (12)3 (7)10 (12)0 (0)
 Severe0 (0)0 (0)3 (4)0 (0)2 (2)0 (0)
 Terrible0 (0)0 (0)0 (0)0 (0)1 (1)0 (0)
 None67 (86)38 (88)66 (80)35 (80)67 (81)32 (89)
 Mild6 (8)3 (7)11 (13)6 (14)10 (12)3 (8)
 Moderate4 (5)2 (5)5 (6)2 (5)6 (7)1 (3)
 Severe0 (0)0 (0)0 (0)1 (2)0 (0)0 (0)
 Terrible0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)

Serum gastrin

As would be expected in patients taking long-term proton pump inhibitor treatment, increases in serum gastrin levels were observed during the study, although the increases were modest in most patients. The mean gastrin concentrations tended to be higher in the omeprazole group, particularly at later time points (Figure 3). These mean values were heavily influenced by small numbers of patients in the omeprazole group with high serum gastrin concentrations: no patient in the rabeprazole groups had a serum gastrin concentration of more than 800 ng/L recorded at any time during the study, whereas three patients in the omeprazole group had serum gastrin concentrations of > 1000 ng/L for at least one time point during the final 2 years of the study. The highest serum gastrin concentration recorded in the omeprazole group was 2034 ng/L.

Figure 3.

Mean serum gastrin during the study. The broken line shows the upper limit of normal.

Gastric biopsy results

The results presented here are a brief summary of the biopsy results, which we intend to publish in full elsewhere.

The proportion of H. pylori-positive samples declined during the study (Table 4). This was true of all treatment groups in samples taken from the antrum, but only the rabeprazole 10 mg group showed a substantial decrease in H. pylori positivity in corpus biopsy samples. Part of this decrease was due to preferential dropouts amongst patients who were H. pylori positive at baseline, but there were nonetheless several patients in all groups whose H. pylori status changed from positive to negative during the study. The difference between the 10 mg rabeprazole group and the other groups must be interpreted with caution, as the investigation of H. pylori status was not a primary objective of the study.

Table 4.  Number of Helicobacter pylori -positive biopsy samples at baseline and after 5 years
Baseline5 yearsBaseline5 years
Rabeprazole 20 mg34/74 (46%)17/43 (40%)17/67 (25%)5/38 (13%)
Rabeprazole 10 mg34/78 (44%) 9/43 (21%)21/69 (30%)2/38 (5%)
Omeprazole 20 mg28/78 (36%)14/37 (38%)24/73 (33%)4/32 (13%)

In general, biopsy results showed improvement during the study, with considerably less mucosal inflammation and activity of inflammation at the end of the study than at the start. Atrophy of the antral mucosa also decreased during the study, although there were modest increases in the percentage of patients with atrophy of the body mucosa (from 14%, 12% and 8% at baseline to 21%, 19% and 24% at the end-point in the rabeprazole 20 mg, rabeprazole 10 mg and omeprazole 20 mg groups, respectively). The prevalence of intestinal metaplasia was low, with median metaplasia scores of zero in all groups. Gastric biopsy results stratified by H. pylori status generally showed more favourable results in H. pylori-negative patients; both the frequency and severity of inflammation and atrophy were higher in H. pylori-positive patients. There were no obvious differences in measures of inflammation and atrophy between the treatments.

The frequency of argyrophil enterochromaffin-like cell hyperplasia tended to decrease during the study in the rabeprazole groups, but there was an increase in the percentage of patients with argyrophil enterochromaffin-like cell hyperplasia in the omeprazole group (Figure 4). Hyperplasia was generally mild, and did not progress to dysplasia or neoplasia in any patient at any time during the study. Only two patients had adenomatoid hyperplasia at any time during the study, both at week 13 of the first year of the study in the rabeprazole 10 mg group.

Figure 4.

Argyrophil enterochromaffin-like cell hyperplasia during the study.

Safety and tolerability

Adverse events recorded during the study showed all three treatments to be well tolerated. Of the 1086 adverse events recorded during the 5 years of the study, only 177 were considered to be related to treatment. The adverse events were approximately equally distributed amongst the treatment groups (Table 5). The majority of treatment-related adverse events were mild in intensity, and only 23 (13%) were considered to be severe. There was no obvious pattern to the adverse events: the event reported by the greatest number of patients was diarrhoea, which occurred in only five (6.4%), three (3.7%) and four (4.8%) patients in the rabeprazole 20 mg, rabeprazole 10 mg and omeprazole 20 mg groups, respectively. There were two deaths during the study; neither was considered to be related to the study medication. Laboratory safety tests, vital signs and physical examination gave no evidence of any harmful effects in any of the treatment groups.

Table 5.  Number (%) of patients with adverse events (AEs)
 Rabeprazole 20 mg (n = 78)Rabeprazole 10 mg (n = 82)Omeprazole 20 mg (n = 83)
All AEs62 (80)67 (82)69 (83)
Treatment-related AEs19 (24)20 (24)26 (31)
Serious AEs21 (27)17 (21)23 (28)
Treatment-related serious AEs 2 (3) 1 (1) 2 (2)
Withdrawals due to AEs 9 (12) 6 (7)11 (13)


The main finding of this study was that rabeprazole 20 mg, rabeprazole 10 mg and omeprazole 20 mg were all of similar efficacy in preventing the relapse of erosive GERD during 5 years of treatment. It is important to note that the 10 mg dose of rabeprazole provided equivalent efficacy to the higher dose. This demonstration of the efficacy of low-dose rabeprazole in GERD maintenance treatment is in apparent contrast with the findings of other investigators for low-dose omeprazole. Bardhan et al. found that a dose of 10 mg omeprazole for GERD maintenance led to relapse in about 40% of patients during 18 months of treatment; these patients required titration to a dose of 20 mg daily to maintain healing.19 Ladas et al. found alternate-day dosing with 20 mg omeprazole to be effective in GERD maintenance, but only in patients who had been pre-selected on the basis of 24-h intragastric pH monitoring.20 Clearly, however, these results are from separate studies, and are not therefore directly comparable with the results of the present study.

Any study claiming to show equivalence of treatments is potentially vulnerable to the criticism that the results may simply represent a statistical type II error: in other words, the reason for the failure to detect a difference between treatments is simply due to an inadequate power to detect a difference, even though a difference exists. This is particularly true for studies with a high dropout rate, where the power of the study may be reduced by the decreasing numbers of patients. Although almost one-half of the patients who started the present study failed to complete it, it must be borne in mind that it is unreasonable to expect a high completion rate in a study that lasts for 5 years. The power calculation for the present study also took into account the probability that the number of patients would decrease as the study progressed.

Although a power calculation can be useful in estimating the precision of the study that is likely to be obtained, what is important is the precision of the results that are actually obtained. In the present study, the confidence intervals for the differences amongst the treatment groups were well within the pre-specified limits of clinical equivalence, and so the results of this study show clearly that the treatments were indeed of similar efficacy. Of course, it is impossible to exclude the possibility that the efficacy of one treatment is truly slightly greater than another, but we believe that any such differences between treatments, if they exist at all, are not likely to be large enough to be of any clinical significance.

The high dropout rate in our study introduces a potential for bias in the results. If patients who dropped out of the study were more likely to have a relapse of GERD than those who remained in the study, our results would overestimate the efficacy of the treatments. However, it is noteworthy that only 10 of the 120 patients who dropped out of the study did so due to a lack of efficacy, and so it is unlikely that this bias is large. Moreover, there is no reason to believe that any such bias would have varied amongst the treatment groups, and so our comparison of the treatment groups should be relatively robust. Our findings of similar proportions of relapses in the groups at the end-point are confirmed by the findings of similar survival functions, which take into account the relapses at each time point as a proportion of the patients remaining in the study at that time point.

The population included in the present study was representative of the overall population of patients requiring maintenance treatment for erosive GERD, and so our results should be able to be generalized to most patients in that situation. However, it should be noted that patients who did not respond to acute treatment for GERD were not eligible for inclusion in the study, and so our results may not be relevant to patients with the most severe forms of GERD who do not respond well to acute treatment. Fortunately, however, such cases form only a small subset of GERD patients (7% in the acute study that preceded the present study9). Our study did not include patients with a previous diagnosis of non-erosive or endoscopy-negative GERD, who make up more than 50% of patients with reflux disease.21 However, endoscopy-negative reflux disease has a similar pathophysiology to erosive GERD, but probably requires a lower degree of acid inhibition for treatment.22, 23 We therefore believe it likely that our results can be generalized to patients with endoscopy-negative reflux disease, even though we did not study them directly.

A noteworthy finding from this study was that serum gastrin levels were only modestly raised in most patients, and that any hypergastrinaemia was therefore unlikely to be of clinical significance. However, some patients in the omeprazole group had particularly high serum gastrin concentrations. One possible explanation for this could be genetic polymorphism of the cytochrome P450 (CYP) 2C19 enzyme, which is responsible for the metabolism of omeprazole. Sagar et al. found that patients who were CYP 2C19 poor metabolizers, and who therefore metabolized omeprazole poorly, had significantly higher serum gastrin concentrations after long-term treatment with omeprazole than did patients with normal CYP 2C19 enzymes.24 In contrast, the metabolism of rabeprazole is not significantly affected by CYP enzyme polymorphism, and so there is less variability in its effects.25

Whether markedly raised serum gastrin concentrations are associated with any appreciable risk of harm has been debated. Patients with Zollinger–Ellision syndrome or pernicious anaemia, who typically have serum gastrin concentrations of around 1000 ng/L,26 have an elevated risk of carcinoid tumours of enterochromaffin-like cells, and there is therefore a theoretical risk that the incidence of such tumours could be increased in patients with hypergastrinaemia secondary to proton pump inhibitor treatment.27 However, no such increased risk has been found, and so it is likely that any stimulation of enterochromaffin-like cells by gastrin during proton pump inhibitor treatment leads only to benign changes.28

The present study provides valuable insight into the question of the long-term effects of proton pump inhibitor treatment on enterochromaffin-like cells, as it included enterochromaffin-like cell evaluation in gastric body biopsies. The lack of any enterochromaffin-like cell dysplasia or neoplasia in any patient lends support to the idea that long-term proton pump inhibitor treatment does not have harmful effects on enterochromaffin-like cells in humans. Indeed, argyrophil enterochromaffin-like hyperplasia tended to decrease during the study in the rabeprazole groups.

Despite the lack of eradication treatment, some patients became H. pylori negative during the study. Clearance of H. pylori may be due to the inhibitory effects of both rabeprazole and omeprazole against the organism, or to patients taking concurrent antibiotic treatment during the study. However, concomitant medications were recorded throughout the study, and few of the patients who showed clearance of H. pylori were recorded as having taken antibiotics. Omeprazole has bacteriostatic activity against H. pylori, and also inhibits urease, an enzyme used by H. pylori to protect itself from the harsh acidity of the intragastric environment.29 Rabeprazole not only inhibits H. pylori growth and urease, but also reduces the motility of the organism.30

When making a choice between several similar drugs to treat a particular condition, the practitioner will generally have to consider the relative merits of the drugs in terms of efficacy, safety, tolerability and cost. The present study shows that rabeprazole and omeprazole have similar efficacy, safety and tolerability in the maintenance treatment of GERD. This is in accord with other authors' assessments of the choice between proton pump inhibitors, which have generally concluded that all are of similar efficacy, and that cost is an important factor in choosing between the drugs.31–33

Members of the european rabeprazole study group

P. Hoang, Brussels, Belgium; J. H. Pedersen, Solrød Strand, Denmark; B. J. Plucnar, Solrød Strand, Denmark; B. Thjodleifsson, Reykjavik, Iceland; J. P. Crowe, Dublin, Ireland; P. W. N. Keeling, Dublin, Ireland; C. P. M. Dekkers, Breda, The Netherlands; J. A. Beker, Leidschendam, The Netherlands; A. Gabryelewicz, Bialystok, Poland; E. Butruk, Warsaw, Poland; T. Popiela, Krakow, Poland; K. Marlicz, Szczecin, Poland; L. Szczepánski, Lublin, Poland; J. M. Herrerias Gutierrez, Seville, Spain; I. Wright, Göteburg, Sweden; A. Söderlind, Visby, Sweden; D.-A. Hallbäck, Karlskoga, Sweden; H. V. Tanghøj, Eskilstuna, Sweden; D. G. Maxton, Shrewsbury, UK; C. P. Swain, London, UK; K. D. Bardhan, Rotherham, UK.


The authors would like to thank Dr Adam Jacobs of Dianthus Medical Limited for his expert assistance in preparing the manuscript.

This study was funded by Eisai Limited, London, UK.