The management of functional dyspepsia (FD) remains challenging.1 2 FD, defined as unexplained epigastric pain or meal-related symptoms, may affect as many as one in six in the general population.3 Furthermore, approximately 5% of primary care practice deals with dyspepsia, of which the majority fall under the category FD.3 Following a normal endoscopy, treatment options for patients with FD include acid suppression therapy or Helicobacter pylori eradication, but the majority of patients fail to respond to both approaches.4 5 Prokinetic therapy has appeared promising in FD, although the efficacy of available therapies has been controversial.6 In a meta-analysis, cisapride appeared to be superior to placebo, but concerns have been raised about possible publication bias, and cisapride has been withdrawn because of safety concerns.6 7 The prokinetic tegaserod, a 5-HT₄ agonist, was used off-label for dyspepsia, but efficacy data have been mixed; it was recently withdrawn because of concerns about an increased incidence of cardiovascular ischaemic events (0.1%).8 Despite continued off-label use in FD, the efficacy of metoclopramide or domperidone is not established in this condition.9

Itopride is a benzamide derivative that acts via dopamine D2 receptor antagonism and acetylcholinesterase inhibition.10 In a dog model of pharmacologically induced gastric hypomotility, itopride restored motor activity in a dose-dependent manner.11 Because it is highly polar, it does not cross the blood–brain barrier in any significant amount, and the drug does not alter the Q-T interval like cisapride (data on file, Axcan Pharma). Itopride is currently marketed in several countries including Japan. A phase IIb multicentre, randomised, double-blind, placebo-controlled trial compared three doses of itopride (50, 100 and 200 mg) with placebo12; in 548 white patients from Germany, there was a significantly higher proportion of responders based on a global efficacy measure versus placebo for all doses.

The aim of these two phase III clinical trials was to assess the efficacy and safety of itopride in FD. One study was conducted primarily in North America and an identically designed study was performed mostly in Europe. We hypothesised that 100 mg three times daily of itopride would be superior to placebo in FD, both in terms of a global assessment of efficacy as well as in terms of specific dyspepsia symptoms.

METHODS

Two similar placebo-controlled, randomised controlled trials were conducted (fig 1). The International trial was conducted in Germany, France, The Netherlands, Belgium, Poland, the UK, the USA and Canada. The North America trial was conducted primarily in the USA and Canada, with additional sites in Poland and Germany. Patients were selected essentially from first tier gastroenterologists, with 14 out of the 170 sites being academic sites, the rest being private practice or hospital-based sites. At the end of the double-blind phase, patients were allowed to continue being treated in an open-label extension phase for a maximum of 2 months following the end of the double-blind phase. The results presented here cover the double-blind period only.

Study population

Males and females, 18–65 years of age, were eligible for enrolment. All patients were required to sign an informed consent form before study evaluations. Patients were required to have a diagnosis of FD by Rome II criteria.13 This was defined as the presence of persistent or recurrent dyspepsia (pain or discomfort centred in the upper abdomen). Discomfort could be characterised by or associated with upper abdominal fullness, early satiety, bloating or nausea. Patients were required to have no evidence of organic disease that was likely to explain the symptoms, and no evidence of dyspepsia that was exclusively relieved by defecation or associated with a change in stool frequency or stool form, to exclude irritable bowel syndrome.

Inclusion and exclusion criteria

Patients were required to have a negative upper endoscopy which excluded oesophagitis, peptic ulcer disease and oesophageal or gastric neoplasia. The endoscopy was required to be completed within 1 month prior to enrolment. Prior to endoscopy, patients were required to be off any antisecretory therapy for at least 2 weeks. Patients were further required to have no or infrequent heartburn (defined as burning, retrosternal pain). If they did have heartburn, this could not exceed a frequency of one episode per week and had to be subordinate to their abdominal pain or discomfort.

To be eligible, patients had to have baseline severity of at least moderate symptoms on the Leeds Dyspepsia Questionnaire (LDQ; total score of ⩾9, as described below).14 Patients were required not to receive any non-steroidal anti-inflammatory drugs in the 4 weeks prior to randomisation, antisecretory medicine within the last 20 days prior to randomisation or any other treatment for FD. Patients were further required to be H pylori negative, confirmed with a 13C urea breath test within 2 weeks prior to enrolment. Patients suffering mainly or exclusively from symptoms corresponding to reflux disease were excluded. As the evaluation of symptoms needed to be standardised at all sites, an algorithm was provided to sites in order to guide them in the patient selection process; the symptoms reported by the patient on the LDQ and Nepean Dyspepsia Index (NDI; see below) needed to be consistent. The algorithm allowed the sites to cross-check the patient responses, to make sure that no exclusionary symptoms were reported.

Other exclusion criteria included patients with any clinically relevant ECG abnormalities, inflammatory bowel disease or coeliac disease, any severe hepatic, renal, cardiac, metabolic, haematological or malignant disease, and trimethylaminuria. Those who changed their smoking status, pregnant or lactating women, or those with a specific food intolerance which was relieved by changing their diet were also excluded. Alcoholism or drug abuse and active psychiatric disease were also exclusions.

Randomisation, concealed allocation and blinding

All patients had a four-digit number assigned. The randomisation code was generated by Quintiles Inc. using a computer-generated program. At the baseline visit, eligible patients were assigned a randomisation number according to the predetermined list at each site. These numbers were allocated in sequential order and registered in the patient enrolment list. This ensured appropriate concealed allocation. Patients took identical active or placebo medication, and patients and investigators were blinded at all sites. Emergency envelopes were provided to the investigators with the study code and randomisation, and these were examined at the end of the study to ensure the study blind being maintained.

Compliance

At baseline and 1 month postrandomisation, a sufficient supply of study medication was given. Pill counts were used to check compliance. Follow-up was obtained at 2, 4 and 8 weeks after randomisation, and compliance was verified at week 4 and 8 only.

Outcome measures

Global patient assessment (GPA) of efficacy12

The patient was asked to answer the question “Please rate the strength of your upper abdominal complaints in the past 14 days. Compared to the condition at the onset of treatment, how much have they changed? Please mark the statement that best applies to you: symptom-free, markedly improved, slightly improved, unchanged, worse”. Symptom-free or markedly improved was defined as a responder.

Leeds Dyspepsia Questionnaire (LDQ)1415

This measures both the presence and severity of upper gastrointestinal tract symptoms and was administered in a face to face interview. It comprises eight questions asking about the frequency and severity of symptoms and generates a global dyspepsia score by summing the severity responses of the eight symptoms. The LDQ has been shown to have a sensitivity and specificity of 80% and 79%, respectively. The test–retest reliability κ has been reported to be 0.83, with internal consistency by Cronbach α of 0.68. LDQ questions 1 and 8, measuring pain in the upper abdomen and feeling of fullness, respectively, were the primary end point questions.

Nepean Dyspepsia Index (NDI)1617

This is a self-administered questionnaire with 25 questions assessing disease-specific quality of life and a separate symptom checklist which measures the frequency, intensity and bothersomeness of 15 upper gastrointestinal symptoms. The quality of life part of the NDI has good internal consistency (Cronbach α 0.81–0.96), as well as good discriminative validity and convergent validity. Worse quality of life correlates strongly with a higher symptom score.

Primary efficacy end points

There were two co-primary efficacy end points defined a priori: pain in the upper abdomen and a feeling of fullness from the LDQ, and the GPA of efficacy. Both variables were evaluated in terms of responder and non-responder status at the end of 8 weeks. To be classified as a responder, patients had to be either symptom free or markedly improved according to the GPA of efficacy, and to have shown improvement with respect to the LDQ grading for severity of pain and fullness (each scored on a 5-point scale rating from very mild, mild, moderate, severe to very severe). A priori, a different cut-off point was used as a co-primary outcome measure for the International trial versus the North American trial with respect to these LDQ subscores. In the International trial, a 1-point improvement on either pain or fullness, without deterioration of the other, was used as one of the co-primary outcomes. In the North American trial, a 2-point improvement on either pain or fullness, without deterioration of the other, was used as the co-primary outcome.

Secondary end points

These were defined as change from baseline and the overall severity of FD using the LDQ severity score at weeks 4 and 8, change from baseline in the patient’s quality of life as measured by the NDI at weeks 2, 4 and 8, change from baseline in the overall severity of FD as measured by the NDI symptom score at weeks 2, 4 and 8, and GPA of efficacy at weeks 2 and 4.

Safety evaluations included adverse event and serious adverse event reporting, and physical examination including vital signs, laboratory and 12-lead ECG testing. In addition, the North American trial included a Data Safety Monitoring Board (DSMB) for the assessment of cardiac safety. The DSMB was comprised of three independent cardiologists. Their mission was to conduct ongoing evaluation of all cardiac-related events and make an assessment about the overall safety pattern observed in the trial. The data comprised:

• ECG data: 12-lead ECG, and 24 h Holter monitoring on a subset of 30 patients.

• Laboratory data: the normal range for prolactin levels was dependent on sex and was defined as 42–423 mIU/l for males and 42–613 mIU/l for females.

• A listing of specified cardiac-related adverse events: syncope, heart failure, angina, myocardial infarction, stroke, arrhythmias, palpitations, seizures or ECG abnormalities.

Statistical analysis

An intent-to-treat (ITT) analysis was conducted on all patients randomised to therapy who received at least one dose of study medication and had a postbaseline assessment of at least one of the co-primary efficacy parameters. A revised ITT analysis using all randomised patients was also carried out post hoc on the primary efficacy outcome measures only. A per protocol analysis was conducted consisting of all patients who completed the protocol without major protocol violations, including those taking at least 80% of the tablets prescribed.

A last observation carried forward (LOCF) analysis was conducted with classification of a non-responder for any missing data. The overall difference between the two treatments was tested using a general Cochrane–Mantel–Haenszel raw mean score test adjusting for centre.

Secondary end points were analysed by a covariance (ANCOVA) model controlling for baseline and centre for quantitative end points. For the GPA of efficacy measured at weeks 2 and 4, a Cochrane–Mantel–Haenszel raw mean score test to compare the two treatment groups was conducted. Subgroup analyses were conducted to evaluate the comparability of the treatment groups including those patients who suffered from heartburn and those who did not suffer from heartburn.

Study power

Based on the phase IIb clinical trial, the proportions of the placebo responders was calculated to be about 40% and 60% for the GPA of efficacy score and the LDQ items of pain and fullness, respectively.12 In order to detect a difference of at least 15% with respect to the pain and fullness items from the LDQ versus placebo, a total sample size of 216 patients per group would be required, based on the significance level of 5% and a power of 90%. Using the same assumptions, a sample size of 140 patients would be required to detect a difference of 20% with respect to the GPA of efficacy score. A dropout rate of 15% was included in the calculation, giving a total sample size of 500 patients for each of the two clinical trials.

RESULTS

A total of 1170 patients were randomised across the two clinical trials (fig 2). Baseline characteristics were similar in the two trial populations. Furthermore, there were no major differences between those randomised to itopride or placebo across both trials combined (table 1).

Primary efficacy

The GPA of efficacy on itopride and placebo in the North American trial was 37.8% and 35.4%, respectively; in the International trial the GPA was 45.2% and 45.6%, respectively (both p = NS) (table 2). Based on an LDQ improvement by at least 2 points in a LOCF ITT analysis, there was a significant difference observed in the International (itopride 62% vs placebo 52.7%, p = 0.04), but not the North American trial (itopride 46.9% vs 44.8% placebo) (table 2). Using an end point of LDQ responder with improvement by at least 1 point, no significant differences were observed (International trial, itopride 80.4% vs placebo 76.6%; North American trial, itopride 73.8% vs placebo 69.8%). The results were not altered by considering only patients randomised (table 3).

Secondary end points

There was no difference in the NDI or quality of life scores in either clinical trial. Similarly, the LDI total score, and GPA scores at week 2 and week 4 did not show any significant differences (table 4). The subgroup analysis comparing patients with heartburn versus no heartburn at baseline did not identify any significant treatment effects in either of these populations (table 5).

Compliance, safety and tolerability

Compliance was adequate in both trials, with 80% of patients being at least 80% compliant in both trials. Itopride was comparable with placebo in terms of adverse events. There was a modest prolactin elevation on itopride versus placebo. A total of 18 patients (3.1%) on itopride in both trials combined had an increase in prolactin levels versus one case in the placebo group (0.1%). None of these prolactin elevations was classified as severe. No galactorrhoea was reported.

The ECG data showed no evidence of any significant change in ECG intervals, outliers or new clinically relevant morphological changes. One case of atrial fibrillation was reported in the itopride group but was considered highly unlikely to be drug related by the DSMB. A total of 9 patients in the itopride group reported serious treatment-emergent adverse events (2 possibly related, 4 unlikely, 3 not related) vs 3 in the placebo group.

DISCUSSION

In two adequately powered, randomised, double-blind, placebo-controlled trials, itopride was not superior to placebo in terms of symptom improvement in FD in the population studied. These results fail to confirm the positive efficacy findings for itopride in a smaller dose-ranging randomised controlled trial conducted in Germany.12

Reasons for the discrepancy with the phase II study12 may include heterogeneity of the condition, differences in patient selection, suboptimal evaluation of symptom intensities or lack of true efficacy of the drug. FD is known to be a heterogeneous syndrome and several potential pathophysiological abnormalities have been identified.18^–20 These have included both slow and accelerated gastric emptying, gastric dysaccommodation and gastroduodenal hypersensitivity.18^–20 No pathophysiological testing was conducted in the current trial for practical reasons, in part because there is no consensus on how best to do this in large clinical trials. In a randomised study of healthy volunteers, itopride impaired gastric volume responses after a meal versus placebo, implying that it may worsen gastric accommodation.21 Other gastric motor and sensory functions, including symptoms induced by the nutrient drink challenge to the point of full satiation, gastric emptying and small bowel transit, were also not altered by itopride in healthy volunteers.21 The few studies that have attempted to stratify patients by slow or normal gastric emptying have generally failed to show that this approach confidently predicted prokinetic treatment efficacy, so this seems unlikely to be the explanation for the negative results.22^–24

Notably, however, the patient population in the current trials differed from that in the phase II study12 where H pylori status was not exclusionary and there was no restriction for heartburn or LDQ score at baseline. In the current trials, all patients had to be H pylori negative at baseline, and the LDQ score at baseline needed to be ⩾9. The LDQ is a validated instrument14 and is composed of three domains: dyspepsia symptoms (epigastric pain, fullness and belching), oesophageal symptoms (heartburn, regurgitation and dysphagia) and feeling sick (nausea and vomiting). As the inclusion and exclusion criteria implied that oesophageal symptoms needed to be almost absent (patients were required to have no or infrequent heartburn, no more than one episode per week and subordinate to abdominal pain or discomfort), high intensity scores for the typical symptoms of pain and fullness were needed for the patient to reach the eligibility threshold of ⩾9 on the LDQ. High baselines scores for these symptoms may have contributed to the high placebo response rate when improvement of 1 or 2 points on the LDQ was considered as an end point.

The exclusion of heartburn may also have had an impact on the response rates in the present studies compared with the phase II trial. Indeed, a post hoc analysis we conducted on the phase II data, looking at the efficacy of itopride on every symptom taken individually, revealed that heartburn responded best to treatment of all the symptoms assessed (55.8% with heartburn responded to itopride applying a 2-point rule on the LDQ vs 38.6% on placebo; p = 0.009).12 Notably, in the phase II study, 51% of the patients enrolled suffered from at least mild heartburn.12 In the phase III studies, any evidence of gastro-oesophageal reflux disease (GORD) led to exclusion. While a subanalysis of the phase III data, looking at the efficacy in patients suffering from minor heartburn at baseline versus those who did not, failed to show any benefit of treatment in either category, the proportion of patients suffering from heartburn was low as expected (34%). The frequency of heartburn was also limited to once a week or less. In the phase II study, a higher response rate of heartburn to the 50 mg dose was also noted (65.6% vs 38.6%, p = 0.001).12 It is conceivable that the restriction imposed on heartburn in the present phase III programme may have obscured a true drug effect. Excluding heartburn cases from FD trials was not considered appropriate by the Rome III consensus, published after this trial programme commenced, although this stance is controversial.25 However, it is notable that overlap with GORD symptoms may well be present in the majority of dyspeptic subjects.26 On the other hand, the available literature suggests that co-existing heartburn is a predictor of responsiveness to proton pump inhibitors in FD patients without predominant heartburn.5 It is our view that further trials with itopride in patients with FD without exclusion of co-existing heartburn would be useful to explore these issues further.

Although the same scales were used in the phase II and the phase III clinical trials, important differences need to be pointed out as relates to the definition of a responder. In the phase II study, the total score of the LDQ was used as one of the co-primary end points. Using this approach, the maximum change in the LDQ total score between active treatment and placebo in phase II was 1.77 (6.25 for the 200 mg dose vs 4.5 on placebo).11 Although this result was statistically significant, the clinical relevance of a 1.7 point difference on a 40-point scale may be questioned. While the LDQ scale is fully validated, its relevance in the evaluation of drug efficacy in FD patients may be argued, as many of the questions relate to reflux symptoms. In the phase III programme, an attempt was made to target core symptoms of FD better using this scale, and, therefore, specific questions in the LDQ related to pain and fullness were used as primary end points. Using this approach, placebo and treatment were generally similar, with the exception of one analysis, using a 2-point difference in score (table 2). It could be argued that the GPA of efficacy was weighted towards a positive response which may have exaggerated the placebo response, which ranged from 35% to 46% in the current trials. However, this would be unlikely to have changed the absolute therapeutic gain of itopride. Another potential limitation is what represents the minimal improvement with clinical significance on these scales is not established, although a 1-point change on a 5-point graded scale would seem reasonable.

The present clinical trials had a number of strengths. Validated measures of outcome were applied, and the results were consistent across the measures used. Both trials applied rigorous approaches to randomisation, blinding and concealed allocation. Compliance was assessed and appeared to be adequate in both trials. Centre effects were adjusted for in the analyses. Multicentre trials can lead to heterogeneous patient populations being included and, therefore, may theoretically dilute treatment efficacy, but realistically, in order to recruit such a large clinical programme, multicentre trials remain the only viable way to do the research. Whether the results of the current clinical trial programme would have been more positive if patients with just meal-related symptoms had been enrolled, is unknown. Another hypothesis worthy of testing may be that the dose chosen for the phase III programme was not optimal in a Western patient group despite the phase II findings.12

In conclusion, itopride was not superior to placebo for the treatment of FD in this study. There was a higher response with respect to the LDQ in the International trial, but this was not confirmed by the North American trial. Although itopride was well tolerated, the studies suggest that this drug is not efficacious in the tested patient population at the dose used. However, further studies in patients with dyspepsia and reflux symptom overlap are probably warranted.