Long-term prevention of erosive or ulcerative gastro-oesophageal reflux disease relapse with rabeprazole 10 or 20 mg vs. placebo: results of a 5-year study in the United States


Dr A. Caos, Central Florida Clinical Studies, 10000 W Colonial Drive, Suite 289, Ocoee, FL 34761, USA.
E-mail: acsr52@hotmail.com


Background:  Erosive gastro-oesophageal reflux disease (GERD) is a chronic condition requiring long-term maintenance treatment. However, few trials of proton pump inhibitors in maintaining healing of erosive or ulcerative GERD are conducted for longer than 1 year.

Aim:  To compare the efficacy and safety of 10- and 20-mg rabeprazole with placebo in the 5-year maintenance of healing in patients previously diagnosed with erosive/ulcerative GERD healed in an acute efficacy trial.

Methods:  Patients (N = 497) were randomized to receive once-daily doses of 10- or 20-mg rabeprazole or placebo. The primary efficacy measure was endoscopically documented absence of oesophageal erosions or ulcerations.

Results:  After 5 years, relapse rates in both rabeprazole groups were significantly lower than with placebo (rabeprazole 20 mg, 11%; 10 mg, 23%; placebo, 63%; P < 0.001 for rabeprazole vs. placebo; P = 0.005 for rabeprazole 20 mg vs. 10 mg). Both rabeprazole doses were significantly superior to placebo in preventing relapse of heartburn frequency and improving patient quality of life. Analyses of adverse events, biopsy findings and laboratory values showed no evidence of clinically significant effects.

Conclusions:  Five-year maintenance therapy with rabeprazole is effective in preventing relapse of erosive or ulcerative GERD and is well tolerated.


Gastro-oesophageal reflux disease (GERD) is both prevalent and chronic.1 Symptoms commonly persist for years, and a body of research testifies to the deleterious effect of GERD on patient quality of life. Quality-of-life deficits range from interference with sleep to time lost from work or school.2, 3

Proton pump inhibitors (PPIs) are regarded as the most effective and successful class of drugs for GERD treatment. The profound and consistent effect of PPIs on gastric acid secretion enables these drugs to effectively alleviate symptoms and heal erosions.1, 4, 5 Within 6 months to 1 year of stopping treatment, however, 50–80% of patients experience relapse.1 Despite the consequent need for agents that are effective and safe for long-term use, few randomized, controlled trials evaluating PPIs for maintenance treatment have gone beyond 1 year.6, 7

Rabeprazole is a potent inhibitor of H+K+-ATPase, the enzyme responsible for the final step in gastric acid secretion.8 Exposure of the oesophagus to the damaging effects of gastric acid is key to the pathogenesis of erosive oesophagitis, and the severity of the disease is related to the degree and duration of oesophageal acid exposure. Furthermore, the healing rate of erosive oesophagitis has been shown to be directly related to the duration of suppression of gastric acid secretion achieved over 24 h.9 In providing rapid, sustained acid control, rabeprazole effectively relieves the symptoms of GERD and has been shown effective for the acute healing of erosive oesophagitis and 1 year maintenance of healing.1, 4, 8, 10 Furthermore, a long-term study conducted in nine European countries showed that 10- and 20-mg rabeprazole was as effective as omeprazole 20 mg in maintaining healing of oesophageal erosions for 5 years.6 The aim of this study was to compare the long-term (up to 5 years) safety and efficacy of 10- and 20-mg doses of rabeprazole with placebo in preventing relapse of patients with healed erosive or ulcerative GERD in the United States. Results of the first year of this 5-year study have been published previously.1, 11 This paper reports the complete study results.


Study design

Phase 1 of the study consisted of two parallel, identical studies, NRRK-odd and NRRK-even, which were collapsed into one extension phase. These studies were both 1-year multicentre, randomized, double-blind, placebo-controlled, parallel-arm maintenance trials enrolling patients from sites in the United States who had been previously diagnosed with erosive/ulcerative GERD and had been healed in an acute efficacy trial.1, 11 After successful completion of 1 year of treatment (defined as endoscopically verified maintenance of healing), patients continued an extension phase on their previous assignment for up to 4 years for a total study period of 5 years (260 weeks). Treatment assignments for the extension phase were unblinded to the sponsor for subsequent treatment allocation, but blinded to the treating doctors, study centre support staff and patients.

The study was conducted in accordance with the Declaration of Helsinki provisions and recommendations. Informed consent was obtained from each patient who agreed to participate. Institutional review boards associated with the study's primary investigators approved the protocol and informed consent form.


The initial 1-year maintenance studies enrolled men and women aged ≥18 years with healed erosive or ulcerative GERD confirmed by endoscopy ≤90 days before study entry. Patients who successfully completed the 1-year maintenance phase, defined as patients with maintenance of oesophageal healing confirmed by endoscopy ≤3 days before extension-study entry, were eligible for enrolment in the extension phase. Patients who participated in the extension phase continued on their assigned study drug.

Exclusion criteria included endoscopic evidence of erosive or ulcerative oesophagitis (modified Hetzel-Dent grades 2–4); active peptic ulceration or endoscopic evidence of gastrointestinal bleeding; primary oesophageal motility disorders; previous gastric or oesophageal surgery; oesophageal ring or strictures, oesophageal or gastric varices, or pyloric stenosis; concurrent treatment with anticoagulants, anticholinergics, tricyclic antidepressants, motility agents or antineoplastics; or the presence of serious systemic disorders, including renal and hepatic insufficiency. Women who were pregnant or lactating and patients with Zollinger–Ellison syndrome were also excluded. Patients with Barrett's oesophagus were eligible to participate.

Treatments and measurements

Patients were randomized, at the beginning of phase 1, via a computer-generated randomization schedule to receive rabeprazole 20 or 10 mg or placebo once daily in the morning. Placebo tablets were identical in appearance to the study drug. Upon entry into the extension phase, a medication kit number was assigned according to the randomization schedule in phase 1 (i.e. patients were not re-randomized). At each visit (every 13th week) through week 247, patients received a medication kit containing 13 weeks of study medication. Treatment assignments for the extension phase were unblinded to the sponsor for purposes of treatment allocation, as patients remained on their phase 1 assignment. Treating doctors, study centre support staff and patients remained blinded.

Patients were provided with Mylanta antacid tablets and were instructed to take them as needed for GERD symptom relief. Patients recorded the number of antacid tablets taken on daily diary cards. Short courses of non-steroidal anti-inflammatory drugs, aspirin or acetaminophen were permitted for pain relief, but histamine2-receptor antagonists (H2RAs), sucralfate, prostaglandins, anticholinergics, motility agents and other PPIs were prohibited. Patients were counselled on dietary and other lifestyle measures to improve response (e.g. eliminating chocolate and fatty foods, limiting alcohol and tobacco, and avoiding eating 2–3 h before bedtime).

Visits were scheduled 13 weeks apart, with a maximum of 22 visits for phase 1 and the extension phase combined. Endoscopy was performed at baseline and weeks 4, 13, 26, 39 (optional) and 52 (baseline for the extension phase) in the first year, and yearly intervals (weeks 104, 156, 208 and 260) in the extension phase to determine relapse. Endoscopy was also performed at other times if clinically indicated. If participation was terminated prematurely, endoscopic examinations were performed to determine whether relapse had occurred. Relapse in erosive oesophagitis was defined as a grade >1 on the modified Hetzel-Dent grading scale.12 Biopsy data were obtained throughout the study to evaluate the effects of long-term rabeprazole treatment on gastric mucosa. Efficacy and safety of treatment was assessed at all follow-up visits.

Patients rated heartburn frequency, heartburn severity (daytime and night-time) and overall physical well-being on daily diary cards, based on 5-point Likert scales:

  • 1Daytime and night-time heartburn severity: 0 =none, 1 = mild (present, daytime; disturbing, night-time), 2 = moderate (annoying), 3 = severe (causing marked discomfort), 4 = terrible (disabling).
  • 2Daytime and night-time heartburn frequency: 0 = none, 1 = few (occurring in <25% of days since the preceding visit), 2 = several (occurring ≥25% but <50% of days since the preceding visit), 3 = many (occurring ≥50% but <75% of days since the preceding visit), 4 = continual (occurring ≥75% of days since the preceding visit).
  • 3Overall physical well-being: 0 = very good, 1 =good, 2 = fair, 3 = poor, 4 = very poor.

Study end points

The study's primary outcome measure was relapse of previously healed oesophageal erosions or ulcerations in patients with erosive or ulcerative GERD. Secondary end points evaluated the long-term efficacy and safety of rabeprazole, including relapse of heartburn frequency and daytime and night-time heartburn severity, both defined as a score of ≥2; number of daily antacid doses used; relapse of rating of well-being, defined as the percentage of patients whose well-being ratings worsened to a score of ≥2; comparative quality-of-life indicators for the 10- and 20-mg rabeprazole doses; adverse events, defined as treatment-emergent signs and symptoms; and effects of rabeprazole treatment on enterochromaffin-like (ECL) cells and serum gastrin levels.

Statistical methods

A total of 497 patients entered phase 1, 261 patients completed phase 1, and 205 patients opted to continue the extension phase of the trial. The extension phase was not powered as the design of this phase gave patients the option, after phase 1, to enter it and allowed patients to discontinue from the trial. The extension phase reported here (additional 4 years) was designed to evaluate the continued efficacy and safety of rabeprazole in preventing erosive or ulcerative GERD recurrence for those patients who entered the extension phase.

The intention-to-treat (ITT) population (which included all patients who were randomized to study medication and who received at least one dose of study medication) was the basis for analyses of the primary and secondary efficacy variables. When endoscopic results were missing, the last-observation-carried-forward method imputed the missing data by using the last recorded value. Kaplan–Meier analysis estimated the median time to relapse for each treatment group. The difference between treatment groups was based on log-rank and Wilcoxon statistics. The relationship between demographic characteristics and relapse rates was assessed using the Cochran–Mantel–Haenszel test (general association) adjusted for treatment. An observed-case analysis was used on secondary efficacy variables, where only actual recorded observations were employed. Differences between treatment groups for treatment compliance, mean daily antacid consumption, and duration of study participation were assessed by analysis of variance (anova). The Cochran–Mantel–Haenszel test was used for comparison of histological results, clinical laboratory results and adverse events between treatment groups. For treatment-emergent abnormal laboratory values, a chi-square or Fisher's exact test was used to test between-group differences.



A total of 497 patients were enrolled in phase 1 and were randomly assigned to receive rabeprazole 20 mg (n = 163), rabeprazole 10 mg (n = 165) or placebo (n = 169). There were no significant differences in demographics and patient characteristics among treatment groups (Table 1). Of the 497 patients who entered phase 1, 261 completed it, and of those, 205 opted to continue the extension phase of the trial.

Table 1.  Baseline demographics and characteristics: phase 1
CharacteristicRabeprazole 20 mg (n = 163)Rabeprazole 10 mg (n = 165) Placebo (n = 169)
Male, %
Race, %
Mean age, years54.8354.3252.70
  18–65, %68.773.979.9
  66–75, %24.517.616.6
  >75, %
Mean weight, lbs184.19190.44189.57
Antacid use, %
Smoking habits, %
Caffeine consumption, %


Primary end point.  At the end of 5 years, relapse rates for the rabeprazole-treated groups were significantly lower than for placebo patients (rabeprazole 20 mg, 11%; 10 mg, 23%; placebo, 63%; P < 0.001 for active treatment vs. placebo) (Figure 1). The difference between the two rabeprazole doses for this measure was also significant (P = 0.005).

Figure 1.

Relapse rates of endoscopically confirmed gastro-oesophageal reflux disease (modified Hetzel-Dent grade ≥2) over 5 years.

Patient discontinuation throughout the 5 years of the study was much higher for those patients randomized to placebo than for patients randomized to either the 10- or 20-mg rabeprazole treatment groups (Table 2). Clinical relapse (defined as a modified Hetzel-Dent grade ≥2) was the primary reason for terminating placebo treatment before the end of the study. More than half (59%) of the placebo patients terminated treatment for this reason, compared with 15% in the 10-mg and 7% in the 20-mg rabeprazole groups. Most of these discontinuations occurred in the first year of treatment. Based on the Kaplan–Meier analysis, the probability of GERD erosions remaining healed after 5 years favoured active treatment: 87% for the rabeprazole 20-mg group, 33% for the 10-mg group and 20% for placebo (Figure 2).

Table 2.  Number of patients discontinuing the study over 5 years
 Rabeprazole 10 mgRabeprazole 20 mgPlacebo
  1. Values within parenthesis are expressed in percentage.

Number enrolled (phase 1 – year 1)165163169
Dropped out of study
 Phase 1 (year 1)61 (37.0)41 (25.2)134 (79.3)
 Year 220 (25.6)25 (25.3)9 (32.1)
 Year 35 (9.3)12 (16.7)1 (5.9)
 Year 41 (2.1)1 (1.9)0
 Year 515 (32.6)14 (27.5)5 (33.3)
Figure 2.

Kaplan–Meier plot on proportion of patients remaining in endoscopic remission: days to first relapse.

Median time to relapse was not reached in the rabeprazole 20-mg group because 50% of treated patients did not relapse during the study period. Median time to relapse in the rabeprazole 10-mg group was 1848 days, compared with 32 days in placebo-treated patients (P = 0.0001, rabeprazole 10 and 20 mg vs. placebo; P ≤ 0.007, rabeprazole 20 mg vs. rabeprazole 10 mg) (Figure 2). There was no significant relationship between relapse rates and baseline characteristics (age, race, gender, heartburn frequency and Hetzel-Dent oesophagitis grades) for any of the three treatment groups. In particular, relapse rates and age categories had no apparent relationship (P = 0.1651) within any treatment group (Table 3). Furthermore, the relapse rate in the rabeprazole 20-mg group was virtually the same for patients in the age groups 18–65, 66–75 and >75 years, suggesting that rabeprazole is consistently effective across age groups. There was no relationship between baseline age category and relapse rates for any of the treatment groups. The older age categories comprised relatively few patients, however, and these data should be interpreted with caution.

Table 3.  Relationship between GERD relapse rate (ITT population) and age category
Age, yearsRabeprazole 20 mg, n/N (%)Rabeprazole 10 mg, n/N (%)Placebo, n/N (%)
  1. GERD, gastro-oesophageal reflux disease; ITT, intention-to-treat.

  2. Differences not statistically significant for the relationship between baseline age category and relapse rates for any treatment group.

18 to ≤6511/110 (10.0)23/110 (19.5)87/135 (64.4)
>65 to ≤754/39 (10.3)9/28 (32.1)15/28 (53.6)
>752/11 (18.2)5/13 (38.5)5/6 (83.3)

Secondary end points. Heartburn frequency.  The rate of heartburn frequency relapse (defined as the number and percentage of patients with a symptom score of 0 or 1 at baseline vs. those whose heartburn relapsed [score ≥2] by study week) was significantly lower in the rabeprazole groups than in the placebo group through all study weeks (P < 0.001; Figure 3). The incidence of heartburn frequency relapse in the rabeprazole 20-mg group ranged from 21% at week 4 to 39% at study end point (week 260). For rabeprazole 10 mg, the range was 22–48%, and, for placebo, 68–78%. Heartburn frequency relapse rates also differed significantly between the rabeprazole 20- and 10-mg groups for more than one-third of the 260 study weeks (P ≤ 0.05).

Figure 3.

Heartburn frequency relapse rate over 5 years.

Heartburn severity.  Both rabeprazole groups had significantly lower relapse rates for daytime heartburn than did placebo at every study visit (rabeprazole 20 mg, P < 0.001; rabeprazole 10 mg, P ≤ 0.018). Daytime heartburn relapse rates were similar in the two rabeprazole groups (Figure 4a), although night-time heartburn rates favoured the 20-mg rabeprazole dose (Figure 4b). At all study visits, rabeprazole 20 mg was significantly superior to placebo for night-time heartburn relapse rates (P ≤ 0.005).

Figure 4.

(a, b) Daytime and night-time heartburn severity relapse rates over 5 years.

Antacid use.  The mean daily antacid consumption was reduced from baseline in patients taking rabeprazole compared with placebo (Figure 5). These effects were statistically significant at weeks 4, 13 and 104 for rabeprazole 20 mg, and at weeks 4–104 and week 143 for rabeprazole 10 mg. At week 260, the mean daily dose of antacid consumed was less in the rabeprazole 20-mg group (0.17) than for rabeprazole 10 mg (0.24) or placebo (0.24).

Figure 5.

Mean change in daily antacid consumption from baseline over 5 years.

Safety and tolerability

Adverse events.  Rabeprazole was generally well tolerated. Rates of study termination due to adverse events (5-year totals) were not significantly different across all groups (rabeprazole 20 mg, 12%; rabeprazole 10 mg, 11%; placebo, 4%). Most events occurred early in the study, gradually tapering off to 0 events in the last 2 years. No specific patterns in the distribution of adverse events leading to premature termination were apparent for any of the treatment groups throughout the study. The most frequent treatment-emergent adverse events are listed in Table 4. Over 5 years, more treatment-emergent signs and symptoms occurred in the rabeprazole treatment groups than in the placebo group.

Table 4.  Most frequently reported treatment-emergent signs and symptoms during five treatment years
Signs/symptomsRabeprazole 20 mg* (n = 163)Rabeprazole 10 mg* (n = 165) Placebo (n = 169)
  1. Values are expressed in percentage.

  2. P ≤ 0.018 vs. placebo for all.

Flu syndrome23208
Surgical procedure20194
Back pain19198
Abdominal pain17196

Overall, 42 patients (8%) in the study experienced adverse events that were judged to be related to a study drug; however, no intergroup differences were apparent. Adverse event rates in the rabeprazole groups remained statistically equivalent throughout the study. Only one serious adverse event was judged to be related to administration of study medication, and this occurred in a patient receiving placebo. The six deaths that occurred during or 3 months following the study were found to be unrelated to study medications. No differences that emerged in pattern of occurrence, subgroup association (age, race, gender) or symptom severity among the three treatment groups were regarded as having clinical significance.

Treatment compliance.  Patient compliance to treatment was significantly better with rabeprazole than with placebo. Compliance was calculated in two ways, each using the number of actual tablets consumed, but comparing this with different numbers of expected tablets consumed. Dispensed study medication compliance was calculated based on the number of tablets dispensed, and actual study medication compliance was calculated based on the number of days between visits. Mean overall dispensed study medication compliance was statistically significantly greater in the rabeprazole 20-mg (80.60%) and 10-mg (80.06%) groups than in the placebo group (74.9%; P = 0.005 and P = 0.016 vs. placebo respectively). The mean overall actual study medication compliance was not significantly different among the treatment groups (94.0–97.2%).

Duration of exposure to placebo and treatment drugs.  A statistically significant difference (P < 0.001) was observed for the rabeprazole groups vs. the placebo group with regard to the mean length of time (days) that patients remained on treatment. The mean duration was 863.9, 742.2, and 258.9 days for rabeprazole 20 mg, rabeprazole 10 mg and placebo respectively. No statistically significant difference was observed between the two active treatment groups regarding mean exposure duration. Thus, the average duration of patient participation in the study was significantly greater for patients who received active treatment compared with those who received placebo and somewhat greater for the rabeprazole 20-mg group compared with the 10-mg group.

Laboratory evaluations.  No clinically important adverse biochemical or haematological effects of study medications were observed. Subgroup analyses, including analysis by age categories, revealed no clinically meaningful differences in the incidence of treatment-emergent abnormal laboratory values.

Serum gastrin levels.  Median serum gastrin in rabeprazole-treated patients fluctuated little throughout the study. Hypergastrinaemia increased by 10–20% over baseline during the initial 13–16 weeks and remained 10–20% higher than baseline at end point. Median serum gastrin levels in the placebo group decreased in the first 13–26 weeks of the study, which can be attributed to patients who were on antisecretory treatment prior to randomization to placebo in the maintenance study, and remained stable thereafter. All three treatment groups had median serum gastrin values within the normal range (upper limit of normal range: 150 pg/mL) at the end of the study (Figure 6).

Figure 6.

Median serum gastrin levels over 5 years.

Histological evaluations.  Biopsies were evaluated for Helicobacter pylori infection, inflammation, atrophy, intestinal metaplasia and ECL cell hyperplasia. Few patients without baseline pathology developed these conditions subsequently. Helicobacter pylori infection was present at baseline in 16–18% of corpus biopsy specimens, with no significant difference in distribution between the treatment groups.

The incidence of ECL cell hyperplasia increased over the course of the study, but few cases progressed in grade. No patient progressed beyond a grade of micronodular hyperplasia, and no evidence of adenomatoid hyperplasia, dysplasia, or neoplasia was seen during the 5-year period. Table 5 summarizes the findings of ECL cell hyperplasia during the study.

Table 5.  Summary findings of enterochromaffin-like (ECL) cell hyperplasia
 Rabeprazole 20 mgRabeprazole 10 mgPlacebo
  1. n/N = number of patients with ECL cell hyperplasia at baseline who had an increase of ≥1 grade in severity at the time indicated/number of patients who had ECL cell hyperplasia at baseline assessed at the time indicated.

  2. † n/N = number of patients with ECL cell hyperplasia at the time indicated/number of patients normal at baseline.

Baseline prevalence, n/N (%)11/135 (8)12/142 (8)14/136 (10)
 Diffuse hyperplasia11/11 9/1214/14
Progression of ECL cell hyperplasia, n/N*
 Biopsies subsequently normal 3/11 4/910/14
 Biopsies subsequently inconsistent 3/11 3/9 –
 No change in severity 1/11 – –
 Increase in severity
  52 weeks 1/7 0/5 0/2
  260 weeks 2/3 1/3 0/1
Incidence of ECL cell hyperplasia, n/N
 52 weeks 1/51 0/43 0/9
 End point18/27 9/19 1/5

Patient well-being

At the 5-year end point, patients in the rabeprazole treatment groups scored significantly higher on overall well-being than did placebo patients (P < 0.001). Rates of well-being were 86% and 81% for the rabeprazole 20- and 10-mg groups, respectively, compared with 67% for placebo.

The incidence of relapse in well-being ratings (defined as a decrease in score from 0 or 1 to ≥2) was significantly lower in the rabeprazole 20-mg group compared with placebo from week 4 to 65. During that interval, relapse rates ranged from 7 to 18% for rabeprazole 20 mg, 14–29% for rabeprazole 10 mg and 26–34% for placebo. A minority of patients (9–11%) in all groups reported time lost from daily activities as a result of GERD, with no intergroup differences.


The study findings demonstrate the effectiveness of maintenance treatment with rabeprazole in preventing relapse in patients with healed erosive GERD for up to 5 years. Patients given the 20-mg dose had a high probability (87%) of remaining healed after 5 years of treatment. Relapse rates for heartburn frequency favoured rabeprazole over placebo. The 20-mg dose fared significantly better than the lower dose on this measure for one-third of the study weeks. The two doses were equally effective for control of daytime heartburn; rabeprazole 20 mg showed consistent superiority for control of night-time heartburn. Rabeprazole treatment also improved patients’ overall well-being compared with placebo. The well-being of rabeprazole-treated patients was rated as consistently superior to that of placebo patients throughout the study (P < 0.001). A dose–response pattern was observed in rates of well-being between the active-treatment arms.

The European study that assessed maintenance treatment of erosive GERD over 5 years compared the efficacy and safety of rabeprazole 10- and 20-mg doses with once-daily omeprazole 20 mg, the standard omeprazole dose.6 This randomized, double-blind study enrolled a population of 243 adults with endoscopic evidence of healing.6 As in the present study, the primary outcome measure was endoscopically confirmed GERD relapse (a score of ≥2 on the modified Hetzel-Dent scale).6 Overall healing relapse rates were low and similar for the three treatment groups (rabeprazole 20 mg, 11.5%; rabeprazole 10 mg, 9.8%; omeprazole 20 mg, 13.3%). In both the US study and the European study, both active treatments (10 and 20 mg) were effective over 5 years; but unlike the present study, in which rabeprazole 20 mg was significantly more effective at preventing relapse than the 10-mg dose, in the 5-year European study, the two rabeprazole doses were equally effective.6 One potential reason for the observed difference is that the prevalence of H. pylori infection was drastically different between the two studies. In this study, H. pylori infection was present in corpus biopsy specimens at baseline in 16–18% of patients, with no significant difference in distribution between the treatment groups. In the Thjodleifsson study, 36–46% of the corpus biopsies were H. pylori positive at baseline.6Helicobacter pylori infection may be associated with increased intragastric pH and better response to PPI.13 This explanation is speculative however, as the analysis of H. pylori was not a primary objective of either of these studies. Overall, the reason for the observed difference between the European and the US studies in the results for rabeprazole 20 mg and rabeprazole 10 mg is not clear.

Apart from the aforementioned study, few trials of PPIs in the maintenance of healed erosive GERD have been conducted for longer than 1 year. The longest study to date evaluated omeprazole treatment in patients with reflux oesophagitis who responded poorly to previous treatment with an H2RA. Patients received an initial maintenance dose of omeprazole 40 mg, which was reduced to 20 mg after acute healing of erosions. This study found that omeprazole was effective for up to 11 years in preventing relapse in patients with severe reflux oesophagitis.14 In a study of pantoprazole 40 mg, the endoscopic remission rate after 2 years was 76% in patients with healed reflux oesophagitis.15

Serum gastrin elevation and ECL cell hyperplasia are considered normal physiological responses to gastric acid suppression. Although studies in rats have shown that hypergastrinaemia and ECL cell hyperplasia can cause gastric carcinoid tumour formation, clinical data on long-term PPI use have not indicated this in humans.14, 16 In the current study, the frequency of ECL cell hyperplasia increased in the rabeprazole group, but no patient progressed beyond a grade of micronodular. No biopsy in the 5-year study period showed adenomatoid, dysplastic or neoplastic changes. In the European 5-year study, hyperplasia was also mild, with no progression to dysplasia or neoplasia.6 In addition, although median serum gastrin values for the rabeprazole treatment group were 10–20% higher at study end compared with at baseline, these increases remained within the normal range. Increases in serum gastrin levels are a normal response to gastric acid suppression. The moderate rise observed in this study is consistent with the pharmacology of PPIs and does not present a safety concern.6, 17–20

Managing GERD in older patients poses special problems. Paradoxically, serious oesophageal erosions are more likely, but the severity of heartburn often underestimates the severity of erosive oesophagitis with increasing age.21, 22 Older patients with GERD tend to have more nocturnal symptoms than younger patients and worse health-related quality of life than those without nocturnal symptoms.21 Older patients are thus more likely to require aggressive therapy both for relief of symptoms and healing of erosions, regardless of the reported severity of presenting symptoms.21, 22 In the current study, small numbers in the older age groups placed limits on the strength of the findings; however, neither relapse rates nor adverse drug-related events had a greater frequency in patients aged ≥65 years compared with younger patients.

Long-term maintenance treatment with PPIs is necessary to control symptoms and help prevent relapse of erosive GERD. Available studies, most notably the multinational, cross-cultural Domestic/International Gastroenterology Surveillance Study, indicate that upper gastrointestinal symptoms significantly impair quality of life.23 Well-being assessments are an important measure of treatment efficacy in studies of gastrointestinal disorders. However, long-term quality of life data pertaining to the use of PPIs are lacking. This 5-year maintenance study found minimal interference with quality of life in rabeprazole-treated patients.


Overall, compared with placebo, this study demonstrated the safety and efficacy of 20- and 10-mg rabeprazole in maintaining GERD healing for up to 5 years. Both doses were effective in preventing the relapse of GERD, heartburn frequency and daytime heartburn severity, with the 20-mg dose also effective in preventing night-time heartburn severity. Rabeprazole treatment vs. placebo improved patients’ quality of life. Both rabeprazole doses were well tolerated and equally safe.


This research was supported by Eisai Inc., Teaneck, NJ, USA, and by Janssen Pharmaceutica Inc., Titusville, NJ, USA.