Double-blind, placebo-controlled comparison of rabeprazole 20 mg vs. omeprazole 20 mg in the treatment of erosive or ulcerative gastro-oesophageal reflux disease


Humphries Eisai Ltd, Hammersmith International Centre, 3 Shortlands, 2nd Floor, London W68 EE, UK.



Rabeprazole sodium is the most recent member of a class of substituted benzimidazole molecules known as proton pump inhibitors. Other proton pump inhibitors have been shown to be effective in healing oesophagitis.


In this randomised, double-blind, multicentre study, conducted at 27 European sites, the efficacy and safety of rabeprazole and omeprazole were compared in patients with erosive or ulcerative gastro-oesophageal reflux disease (GERD).100 patients received rabeprazole 20 mg, and 102 patients omeprazole 20 mg once daily for 4 or 8 weeks, with healing monitored by endoscopy.


Overall GERD healing rates observed and evaluated at weeks 4 and 8 were equivalent. Four-week healing rates for rabeprazole and omeprazole were 81%–81% and 92%–94% for 8-week healing. Rabeprazole-treated patients had similar relief of the frequency and intensity of heartburn to those treated with omeprazole. Both drugs were well tolerated over the 8-week treatment period. Mean changes in fasting serum gastrin were comparable. No significant differences in laboratory parameters were seen. Biopsies for argyrophil ECL cell histology at the end-point revealed a similar distributions of hyperplasia grades to those at baseline in both groups. Biopsies of body and antral mucosa for other parameters were similar between treatments for Helicobacter pylori colonization, presence or degree of inflammation, atrophy or intestinal metaplasia at the end-point.


In this study, GERD healing rates following rabeprazole 20 mg once daily were equivalent to those obtained with omeprazole 20 mg once daily. Both treatments resulted in a comparable relief of the frequency and intensity of heartburn associated with this disease, and both were well tolerated.


Rabeprazole sodium is the newest member of a class of substituted benzimidazole molecules known as proton pump inhibitors. As with other members of this group of compounds, such as omeprazole, lansoprazole and pantoprazole, rabeprazole is a potent inhibitor of H+,K+-ATPase, the proton pump responsible for the terminal step in gastric acid secretion.1 Rabeprazole has been demonstrated to be a potent antisecretory agent, suppressing both basal and stimulated gastric acid secretion in a dose-dependent manner.1[2]–3 Its antisecretory effect appears to approach its maximal response with respect to stimulated acid output and decreased 24-h intragastric acidity with a dose of 20 mg.1, 2 After multiple daily dosing, rabeprazole 20 mg results in a 86% decrease in basal acid output, a 95% decrease in stimulated output and a 80–87% decrease in 24-h intragastric acidity.1[2]–3

Rabeprazole 20 mg has shown to be clinically effective in the treatment of acid-related diseases such as gastric and duodenal ulcer and erosive gastro-oesophageal reflux disease (GERD). ‘Gastro-oesophageal reflux disease’ refers to symptoms and oesophageal mucosal injury caused by the reflux of gastric contents. It is characterized by prolonged and repeated exposure of the oesophageal mucosa to acidic gastric contents, often resulting in ulcerative and erosive damage to the oesophagus. The presence of reflux can be confirmed with 24-h oesophageal pH monitoring, and the normalization of 24 h reflux measurements correlates with symptomatic relief and endoscopic healing.4, 5

In a placebo-controlled, dose-finding study of rabeprazole in patients with moderate to severe GERD, daily rabeprazole doses of 10 mg, 20 mg and 40 mg resulted in statistically significantly higher healing rates compared with placebo. After 8 weeks of treatment, healing rates ranged from 84% to 93% in the three rabeprazole groups, compared with 12% in the placebo group. There was no additional therapeutic benefit from the 40 mg dose and, in view of the higher plasma gastrin levels observed in the rabeprazole 40 mg group, the 20 mg dose was selected for further study in GERD.6

In clinical studies to date, rabeprazole has been well tolerated, and in specific drug-interaction studies, no significant interaction was seen with diazepam, theophylline, phenytoin or warfarin, a feature that distinguishes rabeprazole from various other members of the proton pump inhibitor class.7 Modest and predictable interactions were seen with drugs whose absorption is dependent upon intragastric pH, such as ketoconazole and digoxin.7

The proton pump inhibitor omeprazole has been shown to have superior efficacy compared with both placebo and H2-receptor antagonists in the treatment of GERD, including severe and refractory cases.8 In this study, we compared rabeprazole 20 mg given once daily with the European recommended dose of omeprazole (20 mg once daily) in the treatment healing of erosive oesophagitis and relief of symptoms in patients with erosive GERD.



This study was conducted as a randomized, balanced, double-blind, multicentre, parallel-group comparison that was carried out in accordance with the recommendations of the Declaration of Helsinki [as amended in Tokyo (1975), Venice (1983) and Hong Kong (1989)], the EEC Guidelines for Good Clinical Practice for Trials of Medicinal Products in the European Community (1990) and in compliance with Section 33(6) of the UK Data Protection Act (1984).

This study was conducted by 27 members of the European Rabeprazole Study Group (Table 1 ) in 10 European countries. Men and women of 18 years of age or older were included in the study if they had a history of GERD at least 3 months prior to enrolling in the study and, at admission, endoscopic evidence of erosive or ulcerative oesophagitis (Grade ≥ 2, when measure by modified Hetzel–Dent grading scale; Table 2 ).9 Women were included in the study if they were not of childbearing potential or were using an approved method of contraception. In one country (Sweden), the concurrent use of oral contraceptives was not permitted because of a perceived risk by the regulatory authority of that country of a potential interaction with proton pump inhibitors.

Table 1.  . Members of the rabeprazole study group Thumbnail image of
Table 2.  . Modified Hetzel–Dent grading scale Thumbnail image of

Major exclusion criteria included primary oesophageal stricture or oesophagitis secondary to systemic events such as scleroderma or irritant ingestion (however, patients with Barrett’s oesophagus could be included, provided that oesophagitis grading criteria were met), historical evidence of oesophageal motility disorders, a history of definitive acid-lowering surgery or previous oesophageal or gastric surgery (except for simple closure of perforations), oesophageal and/or gastric varices, pyloric stenosis, treatment with full therapeutic doses of an H2-receptor antagonist, prostaglandin or sucralfate for more than 5 consecutive days within 2 weeks prior to enrolling in the study, treatment with therapeutic doses of a proton pump inhibitor for 3 consecutive days within 2 weeks prior to enrolment and any treatment with therapeutic doses of a proton pump inhibitor within 5 days to enrolment, and concurrent treatment with high doses of corticosteroids, nonsteroidal anti-inflammatory drugs, anticoagulants, anticholinergics, antidepressants, motility agents, antineoplastics or antiepileptics.

In addition, patients were not included in the study if they presented active peptic ulceration, endoscopic evidence of active gastrointestinal bleeding or had known Zollinger–Ellison syndrome.


Patients were randomly assigned to receive either: (i) 20 mg rabeprazole and matching placebo omeprazole once daily in the morning (n = 100) or (ii) 20 mg of omeprazole and matching placebo rabeprazole once daily in the morning (n = 102) for up to 8 weeks. Patients were permitted to use paracetamol for pain relief and an antacid containing a balanced formulation of magnesium and aluminium hydroxides as needed for relief of their gastro-oesophageal symptoms. The use of this rescue medication was documented in a diary. In addition, the use of low-dose aspirin (≤ 165 mg/day) as prophylactic cardiovascular treatment was permitted.

Study medication was begun within 3 days of the baseline endoscopy and endoscopic examinations were repeated at weeks 4 and 8. Patients healed at week 4 exited the study at that point. The primary efficacy criterion was healing defined according to the modified Hetzel–Dent grading scale (Table 2 2 ) as the absence of oesophageal erosions or ulcerations (Grade 0 or 1) as determined by endoscopic evaluation. Patients were advised to follow several conservative measures related to diet to improve their response to treatment for oesophagitis (e.g. avoid fatty foods, chocolate, mint, limit consumption of caffeine, alcohol and tobacco).

Patient evaluation

In addition to the primary efficacy end-point of endoscopic monitoring of healing at weeks 4 and 8 of the study, secondary end-points of the frequency and severity of daytime and night-time heartburn rated according to a five-point scale (frequency: 0 = none, 4 = continual; severity: 0 = none, 4 = terrible), the number of antacid doses taken per day during the study and the patients’ ratings of overall well-being (0 = very good, 4 = very poor) were recorded at baseline and at the endoscopy visits. The patients recorded all of these items in diary cards on a daily basis. The investigator used the diary card information to complete the study case report forms.

Safety monitoring at each visit included the collection of adverse events data, complete blood count, standard blood chemistry panel, urinalysis, vital signs and electrocardiograms (ECG). The effect of treatment on thyroid function and fasting serum gastrin levels (radioimmunoassay, SmithKline Beecham Clinical Laboratories, range 0–149 pg/mL) were also determined. In addition, multiple biopsies of body (4) and antral (2) mucosa were taken at baseline and at the end of treatment to assess the effects of treatment on argyrophil enterochromaffin-like (ECL) cells, H. pylori colonization and the presence and degree of inflammation, atrophic gastritis and intestinal metaplasia. Biopsies were processed for routine light microscopy and stained with haematoxylin and eosin (H&E), and the silver impregnation technique (Grimelius).

Statistical analysis

The study was designed to include ≈ 200 qualified patients randomly assigned to two treatment groups. This sample size was calculated to provide at least an 80% power to rule out a difference of at least 15% between rabeprazole and omeprazole, assuming 8-week healing response rates of 84% for both drugs. The primary efficacy variable, endoscopic healing at weeks 4 and 8, was analysed as an equivalence variable with an equivalence range of 15%. The GERD healing rates were based on an unpublished meta-analysis of clinical studies comparing omeprazole and ranitidine. The sample size was computed using the approximation described by Donner et al.10 and the approximation of Casagrande et al.,11 assuming a two-sided hypothesis test performed at a 5% significance level.

Endoscopic healing rates of erosive or ulcerative oesophagitis (defined as a score of 0 or 1 on the modified Hetzel–Dent grading scale listed in Table 2 2 ) were evaluated using two different approaches. The first approach examined the intention-to-treat (ITT) or last-observation-carried forward population of patients, and was considered to be the primary method for determining efficacy. This technique incorporated data for all randomized patients who had at least one post-dose measurement of any efficacy variable. Endoscopy results were carried forward to the next scheduled time point if the data for that time point were missing. This method tends to under-estimate the true healing response rate.

The second approach used was based on completed visits or endoscopies performed (the ENDO method). With this method, if endoscopy results were not available for a time point, the missing value was not filled in with the result of the previous endoscopic evaluation, unless the previous result indicated healing. This method tends to over-estimate healing rates, as does a true per-protocol analysis.

The significance of differences between the two treatment groups in overall healing rates, as well as the between-group differences in response rates for the secondary efficacy variables of frequency and severity of pain and overall well-being rating, were assessed using the stratified Cochran–Mantel–Haenszel (CMH) procedure with the investigator’s site as stratum. Only patients with GERD symptoms at baseline were included in the secondary analysis of improvement for that symptom. The differences between the both groups with respect to antacid use was assessed using an analysis of covariance ( ANCOVA) model with effects for the respective baseline value, investigator site and treatment.

For the safety analysis, only those signs and symptoms that emerged during the study, or those that were present at baseline but increased in severity during the study, were summarized for the both groups [treatment-emergent signs and symptoms (TESS) analysis]. Differences between both groups in the incidences of TESS reported by at least two patients in one of the two treatment groups were assessed using Pearson’s χ2-statistic. With respect to the laboratory evaluation of safety, differences between the two treatment groups in the incidences of laboratory values outside, below and above the normal reference ranges were compared using Pearson’s χ2 statistic. The differences between the two treatment groups in fasting serum gastrin values were assessed using an ANCOVA model. Distributions of histological evaluations of body mucosa and antral mucosa at baseline and end-point were compared between both treatment groups using the stratified CMH statistic with the investigator’s site as stratum.


Study population

Of the 202 patients enrolled in the study, 100 received rabeprazole and 102 received omeprazole. 95% of the patients completed the study. Ten patients (5%) discontinued from the study, five in the rabeprazole group and five in the omeprazole group. There were four patients who were protocol violators, two in the rabeprazole group and two in the omeprazole group. The other discontinuations were for adverse events (n = 1), lack of efficacy (n = 3) and lost to follow-up (n = 2).

The two groups were comparable at baseline for all demographic and baseline characteristics, except for gender and alcohol consumption (Table 3 ). There was a significant relationship between gender and treatment group (= 0.004) such that more males than females were found in the omeprazole group (73 vs. 29), while the proportions of males and females in the rabeprazole group did not differ. The proportion of patients who consumed alcohol was significantly higher (= 0.013) in the omeprazole group (69%) than the rabeprazole group (51%).

Table 3.  . Summary of demographic and baseline characteristics Thumbnail image of

At baseline, the majority of patients had either Grade 2 (87/202, 43%) or Grade 3 (106/202, 52%) erosive or ulcerative GERD. Both groups were comparable with respect to baseline endoscopy modified Hetzel–Dent oesophagitis grade.


In the ITT population, the GERD healing rates observed at weeks 4 and 8 were equivalent in the two treatment groups (Figure 1 ). At Week 4, the healing rate was 81% in the rabeprazole group compared with 81% in the omeprazole group (95% CI: ± 11%). At Week 8, the healing rate was 92% in the rabeprazole group and 94% in the omeprazole group (95% CI:−9%, +5%). The findings were similar in the ENDO analysis, with healing rates of 82% and 83% for rabeprazole and omeprazole, respectively, at Week 4, and 95% in the rabeprazole group and 96% in the omeprazole group at Week 8. These healing rates were also within the equivalence interval of 15%. The differences between groups were not significant.

Figure 1.

. Erosive and ulcerative GERD healing after 4 and 8 weeks’ administration of rabeprazole 20 mg or omeprazole 20 mg once daily (ITT analysis). There was no statistically significant difference at either time point.

Rabeprazole was compared with omeprazole for the relief of the frequency and intensity of heartburn based on a series of 12 evaluations: percentage improvement (i.e. decrease of severity grade by ≥ 1 point) and percentage resolution (i.e. severity grade = 0) at Weeks 4 and 8 for frequency of heartburn, severity of daytime heartburn and severity of night-time heartburn. Rabeprazole was comparable to omeprazole in the relief of GERD symptoms in all of these 12 comparative evaluations (Table 4 4 ).

Table 4.  . Effect of rabeprazole 20 mg vs. omeprazole 20 mg on relief of heartburn. Summary of improvement or resolution of frequency and severity grades Thumbnail image of

The proportions of patients with improvement or normalization (well-being evaluation grade of 0 = very good) in overall well-being were also comparable for both treatment groups. At Week 8, 44% of rabeprazole treated patients and 41% of omeprazole treated patients reported normalization (N.S.). Patients in both treatment groups tended to use fewer doses of antacid during treatment. There was no statistically significant difference between groups in the mean reduction from baseline in antacid consumption at Week 8, although changes from baseline were greater in the rabeprazole group (rabeprazole −1.90 ± 0.62 dose/day vs. omeprazole −0.83 ± 0.29 dose/day; N.S.).


Both rabeprazole and omeprazole, administered as a daily dose of 20 mg for up to 8 weeks, were well tolerated by the patients in this study. Table 5 lists the significant TESS reported by at least 2% of patients in either treatment group, without regard to causality. There were no statistically significant differences between treatment groups in the incidence of any TESS reported in this study, except for flatulence, which was reported by a significantly (= 0.045) higher percentage of patients in the omeprazole group (4%) than in the rabeprazole group (0%).

Table 5.  . Number of patients with treatment-emergent signs and symptoms (TESS) reported by at least 2% of patients in either treatment group Thumbnail image of

Serious adverse events were reported for one patient in the rabeprazole group, and were considered to be unrelated to study medication. This was an asymptomatic 70-year-old male with a history of hypertension whose ECG at the completion of the study showed T-wave changes compatible with myocardial damage. Cardiac enzymes were normal and the patient received no treatment. One patient in the rabeprazole group was prematurely discontinued from the study because of a rash which was thought by the investigator to be possibly related to the study medication.

With respect to the standard laboratory tests performed in this study, minor changes, occasionally statistically significant, occurred between groups, but none were clinically meaningful. Mean changes from baseline to end-point in urine pH were 0.34 in the rabeprazole group and 0.14 in the omeprazole group (= 0.033). However, mean values remained within normal limits.

Statistically significant differences were observed between the two treatment groups in the percentages of patients who had ALT/SGPT above the normal range at the end-point (= 0.031) and those with sodium below the normal range at end-point (= 0.024).

Most of the patients with elevations in ALT/SGPT (two patients in the rabeprazole group and nine patients in the omeprazole group) had end-point values which were less than 20% above the upper limit of the normal, and the highest value was less than twice the upper limit of normal. The low serum sodium levels at end-point (seven patients in the rabeprazole group, one patient in the omeprazole group) were all only slightly under the lower limit of normal.

No statistically significant differences were observed between the treatment groups for thyroid function tests at end-point, and mean values were within normal limits for all parameters.

Serum gastrin levels increased over time in both groups. The two treatment groups were comparable in the mean change in fasting serum gastrin from baseline to end-point: +36.3 ± 11.1 pg/mL in the rabeprazole group and +23.0 ± 6.9 pg/mL in the omeprazole group (N.S.). Mean values at end-point were well within normal limits for both groups. Figure 2 2 displays baseline and end-point fasting serum gastrin data for both treatment groups as box and whisker plots. Eighteen patients in the rabeprazole group and eight patients in the omeprazole group had normal fasting serum gastrin values at baseline and values which were above the upper limit of normal at the end-point. One of these patients in the rabeprazole group had an end-point value which was more than threefold the upper limit of normal (610 pg/mL). Five patients in the rabeprazole group and two in the omeprazole group had unexplained abnormal values at baseline and higher values at end-point, but all values were less than 50% higher, and in no case was the value more than threefold the upper limit of normal.

Figure 2.

. Fasting serum gastrin levels for rabeprazole and omeprazole treated patients at baseline at end-point (normal range: 0–149 pL/mL). At each time point, the 10%, 25% 50%, 75% and 90% values are given as well as all data outside the 10th and 90th percentiles.

ECL-cell biopsy results

Biopsies for argyrophil ECL cell histology at end-point revealed similar grade distributions to those at baseline, and no significant difference was observed between the two groups. Eight patients in the rabeprazole group and 10 in the omeprazole group with normal argyrophil ECL cell histology at baseline had abnormal findings at end-point; however, no patient in either treatment group had adenomatoid hyperplasia, dysplasia, or carcinoid at baseline or end-point. Seven patients in the rabeprazole group and five in the omeprazole group had abnormal findings at baseline and normal findings at end-point. A histological evaluation of additional parameters (H. pylori colonization; presence and activity of gastritis, gastric atrophy, and intestinal metaplasia) in the corpus and antrum showed no difference between treatments. The incidence of H. pylori colonization decreased in both groups at both biopsy sites, whereas the incidence of atrophy and intestinal metaplasia either decreased or remained unchanged.


Several studies have documented that healing of GERD patients with standard or even high doses of H2-receptor antagonist drug therapy is poor, with 25–50% of patients that have been treated with these drugs for 6–12 weeks remaining endoscopically unhealed,12 even though their symptoms disappear. This failure is apparently caused by an inadequate suppression of gastric acid secretion.13 The severity of oesophagitis increases with the length of time that the intra-oesophageal pH is < 4 increases.14 Proton pump inhibitors normalize 24-h oesophageal acid exposure and are associated with the highest healing rates.14 Rabeprazole at a dose of 20 mg normalizes 24-h oesophageal acid exposure.15

A meta-analysis of randomized trials that were published between 1977 and 1992 involving 3710 patients with erosive or ulcerative GERD revealed that overall 4-week endoscopic healing rates were greater with omeprazole (78%) than with H2-receptor antagonists (42%), sucralfate (41%), miscellaneous agents (35%) or placebo (18%).8 Another meta-analysis of three comparative studies covering a total of 437 patients revealed a 4-week healing rate of 75% in patients treated with 20 mg omeprazole daily compared with 45% in patients treated with ranitidine 150 mg b.d. (< 0.001); at 8 weeks the endoscopic healing rates were 90% and 55%, respectively.16 Lansoprazole and pantoprazole are effective in healing erosive oesophagitis with 8-week healing rates ranging from 87% to 95%.17[18][19][20]–21

The present study compared the effectiveness of once daily rabeprazole 20 mg with once daily omeprazole 20 mg administered to 202 patients with GERD of Grade 2 (43%), Grade 3 (53%) or Grade 4 (4%) using a modified Hetzel–Dent endoscopic grading scale at baseline. The primary variable of endoscopic healing was analysed as an equivalence trial with a range of ±15%.

In both groups, the intention-to-treat GERD healing rates observed at the week 4 and week 8 evaluations were very similar and are consistent with the healing rates from proton pump inhibitors reported in the literature. A demonstration of equivalence is supported by the healing rates in the ENDO analysis (95% CI: ± 4%) which approximates a true per-protocol approach. In addition, the number of dropouts (n = 5) was the same in both groups. Both treatments provided a similar, rapid relief of the frequency and intensity of heartburn that is associated with GERD, although a suggestion of more consistent improvement was seen with rabeprazole.

The favourable tolerability profile of short-term (≤ 12 weeks) omeprazole therapy has been well-established.8 In this study, up to 8 weeks of treatment with rabeprazole 20 mg once daily was as equally well tolerated as omeprazole 20 mg once daily. No clinically significant treatment-related TESS was observed in either group. Only flatulence was reported by a significantly higher percentage of omeprazole-treated patients.

As expected, mean fasting gastrin levels increased in both treatment groups. There was no significant difference between groups in the magnitude of the increase, and the mean levels remained well within normal limits at the end of treatment. Values greater than twice the upper limit of normal were rare, and only one patient in the rabeprazole group, with an unexplained high normal value for fasting serum gastrin at baseline, had a value above 500 pg/mL at the end-point, a value that has been considered ‘critical’.19 Biopsies for argyrophil ECL histology were comparable after both treatments and no patient had adenomatoid hyperplasia, dysplasia or carcinoid histology at baseline or at end-point. Therefore, although short-term treatment with rabeprazole and omeprazole resulted in slight hypergastrinaemia in this study, these changes were mild and were not associated with any trends in increased ECL cell hyperplasia over the 8-week treatment period. There were no treatment-related changes in other histological parameters in this short-term study. The incidence of atrophy and intestinal metaplasia either decreased or remained unchanged.


The results from this study indicate that rabeprazole 20 mg is equivalent to omeprazole 20 mg in the healing of erosive GERD, and has similar efficacy in providing symptomatic relief. Rabeprazole offers an alternative to omeprazole in the short-term treatment of reflux oesophagitis.


The authors are indebted to the following individuals for their critical input into this study and manuscript: S. Wood and B. Sytnik (Medical Writing), J. Jaskir (statistical analysis), Dr J. Gardner (consultant), A. Schwendimann and S. Cuthbert (manuscript preparation).