Article Text


Functional dyspepsia: bye-bye to PPIs
  1. O Nyrén
  1. Department of Medical Epidemiology, Karolinska Institutet, Box 281, S-171 77 Stockholm, Sweden; olof.nyren{at}

Statistics from

Absence of therapeutic benefit of proton pump inhibitor therapy among Chinese patients with functional dyspepsia

When specifically asked, a significant proportion of the adult population reports dyspepsia—that is, pain or discomfort in the upper abdomen. The prevalence reported in the literature varies from 25% to 30%. Although only one in four patients with dyspepsia in the UK and USA sees a doctor for their symptoms1–4 dyspepsia accounts for up to 5% of consultations in family practice.5 When appropriately investigated, less than half have an underlying structural explanation for symptoms such as peptic ulcer or reflux oesophagitis. The remainder are classified as having functional dyspepsia. The vast number of patients with functional dyspepsia, and the usually chronic or recurrent nature of this ailment, creates an enormous market for the drug industry. In spite of the obvious difficulties in delimiting the patient population, and the absence of hard end points, a fair number of—mostly industry initiated—controlled trials have been conducted. However, regardless of whether the target has been gastric acid, gastrointestinal dysmotility, or Helicobacter pylori, the treatment effects have generally been unimpressive, and similar to the usually strong effects observed with placebo alone.

Although meta-analyses6–8 seem to indicate an approximately 20–30% advantage for acid suppression therapy (H2 receptor antagonists or proton pump inhibitors (PPIs)) over placebo in functional dyspepsia, the value of these secondary publications may be limited due to possible publication bias and various imperfections in the studies included. The main concern is the admixture of patients with non-erosive gastro-oesophageal reflux disease (GORD), which is known to respond to acid suppressive treatment.9 As large studies are unlikely to remain unpublished even in the event of a negative result, it appears well advised to focus on such studies. It also seems adequate to concentrate on studies involving PPI treatment. Such treatment is expected to maximise the chances of a positive result if gastric acid plays any role in the production of symptoms in functional dyspepsia. Conversely, negative results are likely to be more informative in a study with PPI treatment than in a study with drugs of less antisecretory potency. In the past four years, two large and carefully planned studies of PPI therapy in adequately investigated patients with functional dyspepsia have appeared in the peer reviewed literature.10,11 In this issue of Gut, Wong and colleagues12 report the results from a third large high quality study conducted in Hong Kong [see page 502]. Under double blind conditions, 453 endoscoped patients with functional dyspepsia according to the Rome II criteria were randomised to receive high (30 mg) or low (15 mg) dose lansoprazole, or placebo for four weeks. Irrespective of the outcome measures (proportion of patients with complete symptom relief, change in dyspepsia score, change in SF-36 quality of life score), and regardless of the substrata (subtype of functional dyspepsia, H pylori positive, H pylori negative, moderate/severe dyspepsia, mild dyspepsia), patients who received lansoprazole did not fare better than those who received placebo. The absence of any trend towards an advantage for lansoprazole allays concerns about a possible type II error.

The negative results of this carefully conducted and adequately analysed study are at odds with the two other large and comparable studies of PPIs in properly defined functional dyspepsia.10,11 The German FROSCH study, published two years ago in this journal,11 comparing omeprazole 10 mg and 20 mg with ranitidine 150 mg or placebo, used a slightly different main outcome criterion (disappearance of dyspeptic symptoms requiring further treatment). Overall, the frequency of this outcome did not differ significantly between the omeprazole and placebo treated patients.13 However, complete disappearance of dyspeptic symptoms—a secondary response variable in this study—was observed more often among patients treated with the highest omeprazole dose. In the full paper, the authors explained their results according to H pylori status.11 The bottom line message was an alleged effect modification by H pylori with a positive effect of omeprazole confined to infected patients. However, it appears that this stratified analysis was done ad hoc. This might have increased the risk of chance findings. Moreover, the investigators may have had an incentive for categorising H pylori positive subjects as non-responders.14 Although a non-significant tendency towards a higher response rate was observed by Wong et al among H pylori positive subjects treated with the highest lansoprazole dose—without a dose-response trend—the absence of any statistically significant therapeutic gain among H pylori positive subjects in two of three high quality studies seems to tentatively refute the idea that acid suppressive therapy is particularly indicated in these patients.

The other large high quality study published previously is the Bond/Opera trial of omeprazole 10 mg and 20 mg versus placebo in functional dyspepsia.10 In this study, omeprazole was noted to be modestly superior to placebo but only among patients with ulcer-like and reflux-like dyspepsia. Case definition, inclusion and exclusion criteria, study design, and outcome measures were almost identical to those in the Wong et al study,12 with a few small exceptions: while the Bond/Opera trial required symptoms during the three day period immediately prior to randomisation, symptoms exceeding a certain severity score during the preceding two weeks was sufficient for inclusion in the Wong et al study. The Bond/Opera study had a short run in period during which antacid treatment was allowed whereas no such period was prescribed in the Wong et al study, and no antacid treatment was allowed. It may be important to note that all three high quality studies included patients with heartburn or acid regurgitation as long as these reflux symptoms were accompanied by other dyspeptic symptoms.

How did the very similar Bond/Opera and Wong et al studies arrive at such different conclusions? A closer look at the Bond/Opera study may provide a clue. Interestingly, this study consisted of two combined studies, of which one—the Opera study conducted mainly at specialist departments—was negative. Overall, differences in symptom abolition rates between active treatment and placebo were considerably larger among patients who were enrolled by general practitioners than among those recruited by specialists. This was due mainly to a higher placebo response among the latter patients. The authors speculated that these patients may have felt more reassured than patients primarily managed in general practice. However, if this explains the absence of pharmacological effects in the specialist patients, it means that reassurance and acid inhibition are likely to compete for the same mechanism for symptom relief and that reassurance may be as potent as omeprazole 20 mg daily for four weeks. A more probable explanation is that specialists and general practitioners select different study populations. It seems a reasonable hypothesis that gastroenterology specialists are more apt at distinguishing between “genuine” functional dyspepsia and other syndromes or organic diseases than general practitioners, and that the study populations selected by the specialists are “cleaner”, notably less contaminated with GORD patients. This makes the negative results in the Wong et al study more understandable.12 In this study, specialists selected the eligible patients. Thus Wong's study corresponds more to the Opera study. Possibly more importantly, the study was performed among Chinese patients. For reasons that are only partly known, GORD is very rare among Asians.15 Hence the admixture of GORD patients is likely to have been negligible.

The rarity of GORD in the studied population is one of the attractions of the study of Wong and colleagues12 and makes it an important contribution to the dyspepsia literature. It adds support to the notion that if patients with GORD related dyspepsia can be removed from the study population, the remainder will have true functional dyspepsia, which may have causes other than hypersensitivity to naturally occurring gastric acid. In the other high quality studies, attempts were made to distinguish between genuine GORD and reflux-like dyspepsia, but the criterion used (presence or absence of other dyspeptic symptoms in addition to heartburn and acid regurgitation) may have been inadequate. The assertions in the FROSCH and Wong et al studies that patients with and without heartburn showed similar responses to PPI treatment may indicate that the demarcation line between GORD and functional dyspepsia does not coincide with the presence or absence of this symptom. However, the indirect indications that gastroenterology specialists and general practitioners may differ in their ability to distinguish between GORD and functional dyspepsia suggest that there may be other pieces of information in the patient history that add to the discriminative power. Identification and evaluation of more accurate criteria for classification of dyspeptic patients should perhaps be given high priority, despite previous setbacks. An individualised pharmacological approach, with acid suppressive therapy tentatively reserved for GORD related dyspepsia, might have the potential for considerably improved cost effectiveness in the management of dyspepsia in general practice.

Absence of therapeutic benefit of proton pump inhibitor therapy among Chinese patients with functional dyspepsia


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles