No randomized controlled trial has compared all the licensed standard dose proton pump inhibitors in the healing of reflux oesophagitis.
To compare the effectiveness of esomeprazole with licensed standard dose proton pump inhibitors for healing of reflux oesophagitis (i.e. lansoprazole 30 mg, omeprazole 20 mg, pantoprazole 40 mg and rabeprazole 20 mg).
Systematic review of CENTRAL, BIOSIS, EMBASE and MEDLINE for randomized controlled trials in patients with reflux oesophagitis. Searching was completed in February 2005. Data on endoscopic healing rates at 4 and 8 weeks were extracted and re-analysed if not analysed by intention-to-treat. Meta-analysis was conducted using a fixed effects model.
Of 133 papers identified in the literature search, six were of sufficient quality to be included in the analysis. No studies were identified comparing rabeprazole with esomeprazole. A meta-analysis of healing rates of esomeprazole 40 mg compared with standard dose proton pump inhibitors gave the following results: at 4 weeks [relative risk (RR) 0.92; 95% CI: 0.90, 0.94; P < 0.00001], and 8 weeks (RR 0.95; 95% CI: 0.94, 0.97; P < 0.00001). Publication bias did not have a significant impact on the results. The results were robust to changes in the inclusion/exclusion criteria and using a random effects model.
Esomeprazole consistently demonstrates higher healing rates when compared with standard dose proton pump inhibitors.
In 2001, we published a systematic review1 that was the first to assess the five available proton pump inhibitors (PPIs) licensed in Europe for the healing of reflux oesophagitis: esomeprazole 40 mg, lansoprazole 30 mg, omeprazole 20 mg, pantoprazole 40 mg and rabeprazole 20 mg. At that time, there were no randomized controlled trials available which compared all of the PPIs and this gap in the evidence base remains.
In the 2001 review, omeprazole 20 mg was chosen as the common comparator because, at the time, it was the most commonly used PPI in randomized controlled trials with other PPIs and provided the largest pool of trials for review. The analysis concluded that esomeprazole 40 mg was the only PPI to demonstrate significantly higher healing rates than omeprazole 20 mg at 4 and 8 weeks.1 Since that initial review, other assessors of the literature have arrived at the same conclusion.2, 3
The implication from the 2001 review was that esomeprazole could be considered the most effective of all of the available PPIs for the healing of reflux oesophagitis. Over the subsequent 5 years, research has been conducted to compare esomeprazole with the other PPIs in head-to-head randomized controlled trials. The primary objective of the current review was to evaluate the direct comparative clinical trials with esomeprazole in the healing of reflux oesophagitis.
Materials and methods
CENTRAL, BIOSIS, EMBASE and MEDLINE were searched for abstracts and papers. All searching was completed in February 2005. The following search terms were included in the search strategy: the drug names [esomeprazole and (lansoprazole or omeprazole or pantoprazole or rabeprazole)], healing (or initial or short-term treatment) of gastro-oesophageal reflux disease (GERD) and clinical trial (or clinical trial, phase III or clinical trial, phase IV or controlled clinical trial or randomized controlled trial). The search was restricted to English-language publications.
The criteria for selection of trials for inclusion in the review were a study design that was a randomized controlled trial, direct comparison of any European licensed standard dose of PPI with esomeprazole 40 mg, which contained endoscopic healing data at 4 and/or 8 weeks across comparable grades of oesophagitis [i.e. Los Angeles (LA) classification4 A–D or grades 2–4 in older classification systems such as the Savary–Miller classification]. As the literature search progressed, a change in the protocol was made to look at endoscopic healing rates by baseline grade of oesophagitis as a secondary analysis.
The main criterion for determining the quality of trials was by an assessment of the method of randomization and concealment of allocation of the treatments used as this has been shown to be the aspect of a randomized controlled trial's design likely to introduce the most bias.5, 6 This was scored as, ‘A = clearly adequate’, ‘B = possibly adequate’ or ‘C = clearly inadequate’. Two reviewers independently reviewed the papers with any differences in opinion adjudicated by a third party.
Endoscopic healing data at 4 and 8 weeks were taken from qualifying trials and recalculated if not presented in an intention-to-treat (ITT) format. For the purposes of this review, we defined ITT as, ‘patients being analysed in the treatment arm that they entered at randomization, regardless of whether they drop-out, receive the incorrect treatment or withdraw before completion of the trial’. If a different definition of ITT was used in the trial being analysed then the results were recalculated where possible. A single reviewer initially extracted data before being checked independently by two others.
Quantitative data synthesis
The summary effect estimate was calculated using a fixed effects model using the Mantel–Haenszel method.7 This was calculated as a relative risk (RR) with 95% confidence intervals (95% CI), where a confidence interval that crosses the value 1.00 indicates no statistically significant difference in treatment effects. A value <1.00 shows a benefit to esomeprazole while a value >1.00 shows a benefit for the comparator PPI. A fixed effects, rather than a random effects, model was used in the analysis, because a random effects model gives undue weighting to smaller trials.8
Heterogeneity was assessed using a chi-squared test for each meta-analysis performed. If significant heterogeneity was detected, a subgroup analysis was performed in order to identify which trial(s) were responsible for the heterogeneity.
All analyses were performed using RevMan version 4.2.8,9 except for the assessment of publication bias, where a regression of normalized effect vs. precision was calculated using StatsDirect version 2.5.610 using the method proposed by Egger et al.11
The following alternative analyses were planned: (i) as quality scoring of randomized controlled trials requires a degree of subjectivity it was planned to assess what effect including all trials regardless of quality assessment would have on the primary analysis; and (ii) using a random effects model using the DerSimonian and Laird method12 for the primary meta-analyses.
Results from the search strategies are presented in Figure 1. As the systematic review aimed to capture the maximum number of qualified trials available for analysis, no restriction was placed on the type of publication included (i.e. full publications and abstracts were considered) if they met the quality criteria.
Of the eight trials identified (Table 1), all were published in full. Only six of the eight papers met the inclusion criteria. The trials were randomized, double-blind, double-dummy in design and had an appropriate account of all patients after randomization. They were judged to be of sufficient quality to have data extracted for inclusion in the analyses. There were no disagreements between the two reviewers. The two trials excluded at this stage were rejected because they focused on a restricted range of oesophagitis (i.e. only LA grades C and D15 or B and C18).
|Study||Dose||Grades of reflux oesophagitis||n||Intention-to-treat|
|4 weeks||8 weeks|
|Healed||Unhealed||% Healed||Healed||Unhealed||% Healed|
|Castell et al. (2002)13||LAN30||LA Grades A–D||2617||1877||740||71.7||2204||413||84.2|
|Howden et al. (2002)14||LAN30||Grades 2–4*||143||107||36||74.8||127||16||88.8|
|Fennerty et al. (2005)15||LAN30||LA Grades C and D||502||238||264||47.4||367||135||73.1|
|Kahrilas et al. (2000)16||OME20||LA Grades A–D||650||399||251||61.4||529||121||81.4|
|Richter et al. (2001)17||OME20||LA Grades A–D||1209||805||404||66.6||978||231||80.9|
|Gillessen et al. (2004)18||PAN40||LA Grades B and C||113||55||58||48.7||94||19||83.2|
|Labenz et al. (2005)19||PAN40||LA Grades A–D||1596||1156||440||72.4||1413||183||88.5|
|Schmitt et al. (2006)20†||OME20||LA Grades A–D||572||379||193||66.3||491||81||85.8|
Quantitative data synthesis
Data for standard dose PPIs were combined and compared with esomeprazole. This meta-analysis of healing rates of reflux oesophagitis showed a significant benefit in favour of esomeprazole at 4 weeks (RR 0.92; 95% CI: 0.90, 0.94; P < 0.00001; Figure 2) and 8 weeks (RR 0.95; 95% CI: 0.94, 0.97; P < 0.00001; Figure 3).
The analysis comparing esomeprazole 40 mg with lansoprazole 30 mg shows a significant difference in favour of esomeprazole at 4 weeks (RR 0.95; 95% Cl: 0.95, 0.98; P = 0.001), and 8 weeks (RR 0.96; 95% CI: 0.94, 0.99; P = 0.0009).
When comparing omeprazole 20 mg with esomeprazole 40 mg, there is a significant difference in favour of esomeprazole at 4 weeks (RR 0.88; 95% Cl: 0.85, 0.91; P < 0.00001), and 8 weeks (RR 0.93; 95% CI: 0.91, 0.95; P < 0.00001).
Similarly, there is a significant difference with esomeprazole, when pantoprazole 40 mg is compared with esomeprazole 40 mg at 4 weeks (RR 0.93; 95% Cl: 0.89, 0.96; P = 0.0002), and 8 weeks (RR 0.97; 95% CI: 0.95, 1.00; P = 0.03).
There were no randomized controlled trials identified comparing rabeprazole 20 mg with esomeprazole 40 mg.
As all eight trials identified met the quality assessment, no additional analysis was possible based on quality scoring. However, when the two trials15, 18 excluded due to the limited grades of oesophagitis allowed in the studies were incorporated (Table 2), the effect of using all eight trials made small numerical differences to the overall estimates but did not change the direction or make a significant difference non-significant. Similarly, using a random effects model as the basis for the meta-analysis makes little difference at 4 weeks (RR 0.92, 95% CI: 0.87, 0.96; P = 0.0001) and 8 weeks (RR 0.96, 95% CI: 0.93, 0.98; P = 0.003).
|Comparator with esomeprazole||Summary estimate (relative risk; 95% confidence interval)|
|4 weeks||8 weeks|
|Standard dose PPIs||0.91 (95% CI: 0.90, 0.93)||P < 0.00001||0.95 (95% CI: 0.94, 0.97)||P < 0.00001|
|Lansoprazole 30 mg||0.94 (95% CI: 0.91, 0.97)||P < 0.0001||0.96 (95% CI: 0.94, 0.98)||P = 0.0002|
|Omeprazole 20 mg*||0.88 (95% CI: 0.85, 0.91)||P < 0.00001||0.93 (95% CI: 0.91, 0.95)||P < 0.00001|
|Pantoprazole 40 mg||0.92 (95% CI: 0.88, 0.96)||P < 0.0001||0.98 (95% CI: 0.95, 1.00)||P = 0.05|
The analysis comparing standard dose PPIs with esomeprazole by baseline grade of oesophagitis (Figure 4) shows a similar trend at 4 and 8 weeks (Table 3). That is, in LA grade A there is a trend in favour of greater healing with esomeprazole but this is not statistically significant. However, at 4 and 8 weeks, for the more severe grades of oesophagitis the additional benefit gained with esomeprazole over standard dose PPIs becomes statistically significant and increases in magnitude as severity increases.
|Los Angeles grade of oesophagitis||Summary estimate (relative risk; 95% confidence interval)|
|4 weeks||8 weeks|
|A||0.98 (0.95, 1.01)||P = 0.12||0.99 (0.97, 1.00)||P = 0.12|
|B||0.93 (0.90, 0.96)||P < 0.00001||0.97 (0.96, 0.99)||P = 0.008|
|C||0.83 (0.79, 0.88)||P < 0.00001||0.90 (0.88, 0.93)||P < 0.00001|
|D||0.73 (0.64, 0.83)||P < 0.00001||0.86 (0.80, 0.93)||P = 0.0001|
A chi-squared test was carried out to investigate the possibility of statistical heterogeneity in the primary analysis. Significant heterogeneity was detected at 4 weeks (P = 0.002) and 8 weeks (P = 0.0009). By sequentially excluding the included trials, we identified that the bulk of the heterogeneity was associated with the inclusion of one study.17 Excluding this trial, the test for homogeneity is passed at 4 weeks (P = 0.19) and 8 weeks (P = 0.23) but statistical heterogeneity should not be used as a post hoc method to exclude trials without an associated clinical reason for heterogeneity.
The limited number of trials in each of the comparisons makes assessment of publication bias difficult. However, in the current analyses, there was no apparent asymmetry in the funnel plots at 4 or 8 weeks, and an assessment of regression of normalized effect vs. precision did not detect any significant publication bias at 4 weeks (bias −0.544, 95% CI: −6.598, 5.510; P = 0.82) or 8 weeks (bias −6.682, 6.715; P = 0.99).
This systematic review confirms the implication of the 2001 review1 that esomeprazole 40 mg has significantly higher healing rates at 4 and 8 weeks compared with the standard dose PPIs with which it has been compared within randomized controlled trials. The secondary analysis demonstrates that magnitude of benefit esomeprazole offers over standard dose PPIs increases as the severity of the underlying reflux oesophagitis increases.
It is unfortunate that no data exist comparing rabeprazole 20 mg and esomeprazole 40 mg in the healing of reflux oesophagitis. However, it should be noted that the previous review demonstrated no significant difference with rabeprazole 20 mg compared with omeprazole 20 mg at 4 weeks (RR 1.00; 95% CI: 0.87, 1.14; P = 0.95) or at 8 weeks (RR 0.98; 95% CI: 0.91, 1.05; P = 0.55).
A definitive cause could not be found for the statistical heterogeneity identified in the primary analysis. A detailed study of the three trials comparing esomeprazole and omeprazole16, 17, 20 did not reveal any apparent differences in how they were designed or carried out. They were all double-blind, randomized controlled trials that were carried out using an identical protocol and in similar patient populations. As such a clinical explanation of why the results from one of the studies17 is different enough for the test for homogeneity to fail remains elusive.
A potential limitation of the research is that it was restricted to English-language publications. However, the lack of asymmetry in the test for publication bias provides some confidence that this limitation is not causing a significant skew on the data included in the analysis.
Today's healthcare providers have to balance the benefits gained with interventions against the costs associated with providing effective treatments. However, as the healing of reflux oesophagitis for most patients requires only 4–8 weeks of treatment, the financial impact of using any of the available PPIs on prescribing budgets is likely to be low. It is in the maintenance of patients with healed reflux oesophagitis that the largest spend is likely to be borne. The choice of which PPI is the most appropriate for the maintenance of patients with healed reflux oesophagitis is beyond the scope of the current research.
In the UK, clinicians have the additional complexity of new guidelines recommending treatment without a definitive diagnosis except when a patient has alarm signals.21 As the severity of symptoms associated with reflux oesophagitis does not necessarily correlate with the level of oesophageal damage,22–24 primary care clinicians have to treat patients ‘blindly’. In such a situation, a prudent course of action might be to treat with the most effective PPI to prevent under treatment of patients with severe disease.
The evidence suggests that esomeprazole 40 mg is the most effective treatment currently available for the healing of reflux oesophagitis. Of the other PPIs, there is no evidence to suggest any significant difference in the healing of reflux oesophagitis compared with omeprazole.1 As such, in a situation where PPI choice is restricted (such as on formulary listings), the choice of esomeprazole and the least expensive alternative PPI would appear to be the best supported by the available evidence.
Esomeprazole demonstrates higher healing rates when compared with the other available PPIs. The magnitude of benefit esomeprazole offers over other standard dose PPIs increases with the severity of the underlying reflux oesophagitis.
This publication was supported by AstraZeneca. SJE and TL are full-time employees of the company.