: The relationship between Helicobacter pylori infection and non-ulcer dyspepsia is not established.
: The relationship between Helicobacter pylori infection and non-ulcer dyspepsia is not established.
: To determine whether eradication of H. pylori might be of benefit in non-ulcer dyspepsia patients.
: We randomly assigned 129 H. pylori infected patients with severe epigastric pain, without gastro-oesophageal reflux symptoms, to receive twice daily treatment with 300 mg of ranitidine, 1000 mg of amoxicillin, and 500 mg of clarithromycin for 7 days and 124 such patients to receive identical-appearing placebos.
: Treatment was successful (decrease of symptoms at 12 months) in 62% of patients in the active-treatment group and in 60% of the placebo group (N.S.). At 12 months, the rate of eradication of H. pylori was 69% in the active-treatment group and 18% in the placebo group (P < 0.001). Complete relief of symptoms occurred significantly more frequently in patients on the active treatment (43%) than in placebo-treated patients (31%, P=0.048). Within the active-treatment group, therapeutic success was significantly more frequent in the non-infected patients (84% vs. 64%, P=0.04).
: Although eradicating H. pylori is not likely to relieve symptoms in the majority of patients with non-ulcer dyspepsia, a small proportion of H. pylori-infected patients may benefit from eradication treatment.
Among the functional gastrointestinal disorders, non-ulcer dyspepsia is one of the most prevalent and consequently results in substantial healthcare costs. About 20% of Western adults have persistent and recurrent upper-abdominal symptoms.1–3 Non-ulcer dyspepsia is usually defined as persistent or recurrent upper abdominal pain or discomfort in patients without detectable abnormalities in structural or biological examinations.3, 4 The pathogenesis of non-ulcer dyspepsia remains poorly understood, although some motility disorders as well as disturbances of visceral perception have been identified in some of these patients.4, 5
Between 30 and 70% of patients with non-ulcer dyspepsia have a Helicobacter pylori infection.6 Due to the lack of satisfactory treatment in non-ulcer dyspepsia, an antibacterial therapy for those patients who have both non-ulcer dyspepsia and H. pylori infection may be relevant. In fact, although some therapeutic trials support a role for H. pylori in non-ulcer dyspepsia, an analysis of the published trials has shown that among 16 trials, eight reported that anti-H. pylori therapy was efficacious, whereas eight failed to detect a statistically significant benefit.7 This analysis identified several serious methodological weaknesses in these trials. Despite the limitations of available studies, several groups of experts have endorsed H. pylori eradication strategies for patients with non-ulcer dyspepsia.8, 9 In particular, one of the weaknesses of the studies was that the follow-up period was not long enough, because the resolution of H. pylori gastritis may take several months. Recently, four controlled trials with an adequate 1-year follow-up have been completed, suggesting no role (or a weak one) of H. pylori infection in the pathogenesis of non-ulcer dyspepsia.10–13 However, a recent meta-analysis of nine randomized, double-blind, controlled trials with a follow-up of 3–12 months (including preliminary results of the present study published in abstract form) revealed a small but significant benefit of H. pylori eradication therapy.14, 15 We report here the results of a large, randomized, placebo-controlled trial that assessed the effect of eradicating H. pylori in patients with painful non-ulcer dyspepsia.
This randomized, double-blind and multicentre study involved 41 French centres and was conducted between September 1996 and December 1998. The study protocol as well as the patient information and consent form had been previously approved by the local ethical committee (Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale, CCPPRB) des Pays de la Loire no. 2. Written informed consent was obtained from patients before enrolment.
Patients who were between 18 and 80 years of age were invited to participate if they had intermittent or persistent epigastric pain for at least 3 months with a severity score of 3 or more on a 5-point Likert scale and occurring at least 3 days during the week before inclusion. This scale evaluates the severity of symptoms (maximal score, 4), with 0 being no discomfort; 1, mild pain or discomfort (can be forgotten); 2, moderate (cannot be forgotten without influencing daily activities); 3, severe (influencing concentration or daily activities); and 4, very severe (stopping daily activities and needing rest). Inclusion criteria included no history of peptic ulcer disease, normal endoscopic findings and H. pylori-positive on analysis with the rapid urease test (Clo-Test). Patients with predominant symptoms suggestive of gastro-oesophageal reflux disease (heartburn and/or regurgitation and/or epigastric burning) were not included. For inclusion into the study, H. pylori infection had to be confirmed on histological examination. Treatment with antibiotics or antisecretory drugs during the 15 days before entry into the study was forbidden, as was the use of non-steroidal anti-inflammatory drugs (NSAID) during the 7 days before entry. Likewise, patients with previous treatment for eradicating H. pylori or antibiotic allergy were excluded.
After inclusion, patients were allocated according to a computer-generated randomized list (1:1) to receive either twice daily treatment with 300 mg of ranitidine, 1000 mg of amoxicillin, and 500 mg of clarithromycin (RAC group) or twice daily treatment with identical appearing placebos (PLA group) for 7 days.
The patients returned at day 8 (i.e. at completion of the treatment) and 3, 6, 9 and 12 months after cessation of treatment. A standardized questionnaire was completed by the investigator at every visit which recorded the occurrence and the severity of dyspeptic symptoms during the last 7 days. In addition, quality of life forms using the SF36 questionnaire were filled out by the patients at inclusion, 3, 6, 9 and at the 12-month visits. Urea breath testing was performed at the 3-month visit. Upper endoscopy with biopsies (two from the antrum and two from the corpus) were performed at baseline and at the 12-month visit. Compliance was assessed by tablet count at 8 days and adverse events were recorded over the entire period of the study.
When needed, rescue symptomatic medications could be prescribed from day 8 until the end of the study provided they were not antisecretory drugs (proton pump inhibitors, anti-H2-receptor antagonists) or sucralfate. Likewise, antibiotics drugs were not permitted for 3 months after treatment.
Treatment success was defined as a decrease of at least 2 points on the Likert scale between randomization and the 12-month follow-up visit. The two groups were compared in terms of percentage of patients responding to this primary outcome criterion. Secondary outcome criteria were: the proportion of patients reporting a decrease of 2 points or more on the Likert scale between inclusion and each visit (day 8 and 3, 6, 9 months); and the proportion of patients with complete resolution of dyspeptic symptoms at 12 months (Likert score of 0). Also considered were: healing of chronic gastritis at 12 months; the global symptom score and the score of each symptom at each visit; the change in each symptom between inclusion and 12 months; the change in quality of life scores; H. pylori eradication at 3 (breath test) and 12 months (histology); and the need for a rescue treatment.
The two antral and two corpus biopsy specimens taken at the pre-treatment and at the 12-month endoscopy were examined by the same experienced histopathologist who was unaware of the patients’ treatment assignment (JFF).
The study was designed to enrol 244 patients. This number of patients would detect a 20% difference in the 12-month success rate assuming a success rate of 60% in the RAC group and 40% in the PLA group, with α=0.05 and β=0.20 and by assuming a 25% drop out rate of patients.
The analysis was based on the intention-to-treat population, considering all patients with missing data at the 12-month visit as failures. The χ2 and exact Fisher tests were used to compare proportions between the two arms. For analysing categorical variables, the Cochran–Mantel Haenszel and Wilcoxon rank sum tests were used. For quality of life, analysis was performed using two aggregated physical and mental health summary scores according to Ware et al.18 Per protocol analysis was performed when there were at least 20% of patients with major protocol violations excluded from the intention-to-treat population. A logistic–regression analysis was carried out to determine which characteristics of the patients were predictive of the resolution of symptoms. Data were analysed by using SAS software (SAS Institute 6.12 program, Cary, North Carolina, USA).
A total of 253 patients were included in the intention-to-treat population; 129 received the RAC treatment and 124 the PLA treatment. The two groups were well matched with respect to age, sex, alcohol and tobacco consumption, frequency, duration and severity of symptoms, as well as quality of life scores (Tables 1 and 2). A total of 101 RAC patients (78%) and 100 PLA patients (81%) were followed-up for 1 year. Compliance to treatment assessed by investigator was considered good in 95% of the RAC patients and 98% of the PLA patients.
To constitute the per protocol population, 58 patients were excluded (30 in RAC group). The main reasons for exclusion were: lack of confirmation of H. pylori infection (n=24); treatment with antibiotic during the first 3 months of the study (n=10); and duration of trial treatment less than 5 days (six patients, all of them in the RAC group). Overall, 99 RAC patients (77%) and 96 PLA patients (77%) were considered for the per protocol analysis.
According to the intention-to-treat analysis, treatment was successful (decrease of at least 2 points on the symptom scale at the 12-month visit) in 62% of the patients in the active treatment group (80 out of 129) and in 60% (74 out of 124) of the PLA patients (N.S.). There were no differences in baseline characteristics nor clinical history and symptoms between patients classified as a success (n=154) or a failure (n=99). The per protocol analysis did not yield significantly different results (respectively 77% and 71%, N.S.).
The complete relief of symptoms (i.e. score of 0) occurred significantly more frequently in the RAC group (55 out of 129, 43%) than in the PLA group (38 out of 124, 31%; P=0.048). The rate of patients with a decrease of symptom score > 2 between initial score and the intermediary scores (i.e at day 8, and at months 3, 6 and 12) was not significantly different between both groups at any time (Figure 1). The course of every individual symptom was not clearly different between groups (Table 2). The need for rescue treatment during the study was not significantly different between groups (36% of RAC patients vs. 40% of PLA patients).
The eradication of H. pylori was observed in 63 out of 91 patients (69%) in the RAC group and in 15 out of 82 patients (18%) in the PLA group at 12 months (P < 0.001). At 3 months, urea breath testing was negative in 70% of patients in the RAC group and in 9% of PLA group patients. Among the 15 PLA patients negative for H. pylori at 12 months, eight were already negative at month 3. Only one of these 15 patients received antibiotic treatment during follow-up. At 12 months, 26% of the RAC patients (24 out of 91) and 79% of the PLA patients (65 out of 82) had both antral and fundic H. pylori infection (P < 0.001).
The course of chronic gastritis in the antrum and in the fundus is given in the Table 3. The prevalence of antral and fundic chronic gastritis was not significantly different between both groups. However, the severity of inflammation as well as the activity of the gastritis were significantly decreased in the RAC patients at 12 months (Table 3). As far as RAC group is concerned, both antro-fundic inflammation and activity, respectively, 16% and 0% for moderate or severe forms, were significantly decreased in H. pylori-eradicated patients, compared with non-eradicated patients, respectively, 82% and 64% (P < 0.001).
A significant improvement within each dimension of the SF 36 score was observed between month 12 and pre-treatment in each group. For two dimensions (physical health and general health), the improvement was significantly greater in RAC patients (Table 4). In addition at 12 months, the physical aggregated score was significantly greater (P=0.02) in the RAC group (0.1 ± 0.8) than in the PLA group (– 0.2 ± 0.8). However, when comparing the variations of scores between month 12 and pre-treatment for each item between eradicated and non-eradicated patients, no statistically significant differences could be demonstrated (data not shown).
In the RAC group at 12 months, therapeutic success was significantly more frequent in the H. pylori-negative patients (53 out of 63, 84%) than in the patients with persisting infection (18 out of 28, 64%, P=0.04). In the PLA group, 11 out of the 15 H. pylori-negative patients at 12 months and 48 out of the 67 H. pylori-positive patients were a therapeutic success (P=N.S.).
At 12 months, activity as well as inflammation of gastritis (moderate or severe grade) were not significantly different between patients with therapeutic success (respectively 37% and 53%). Neither were they significantly different between patients with failure (respectively 51% and 57%), as well as between patients with a null score (respectively 37% and 51%) and those with a non-null score (respectively 44% and 57%, comparisons with a χ2-test).
Clinical and histological initial characteristics of patients were not different according to their score of dyspepsia at 12 months (data not shown). Logistic regression models were used to examine whether any characteristic of the patient at baseline could be of interest to predict complete relief of symptoms. Only RAC treatment and a low degree of inflammation of chronic gastritis at inclusion were significantly associated at a 10% level with a higher rate of complete relief of symptoms at 12 months (respectively P=0.08 and P=0.04 using log-likelihood test).
Twenty-eight per cent of RAC patients and 10% of PLA patients experienced at least one adverse event considered as study treatment related. The most common drug-related adverse events were diarrhoea and taste disorders. Eighteen patients experienced at least one serious adverse event, not considered as study treatment related. Six RAC (5%) and no PLA patient experienced at least one adverse event (headaches, digestive or allergic disorders) leading to treatment being discontinued.
This randomized controlled study shows that a 1-week H. pylori eradicating treatment does not significantly increase the number of patients with a decrease in intensity of their dyspeptic symptoms at 1 year.
Patients included in the study had to be clearly symptomatic with symptoms severe enough to influence daily activities. Table 2 shows that the basal predominant symptoms were epigastric pain and severe epigastric pain in most of the patients studied; about half of them experienced night-time epigastric pain. These characteristics show that patients included really were dyspeptic. We aimed to carefully avoid patients with gastro-oesophageal reflux disease by using selective symptomatic exclusion criteria. It can be seen in Table 2 that the number of patients with symptoms suggestive of gastro-oesophageal reflux disease was in fact very low, indirectly demonstrating that patients had been correctly included. However, at the completion of the 1-week placebo treatment, about two-thirds of the patients were largely or completely relieved of their symptoms. Although the dramatic effect of placebo is frequently reported in non-ulcer dyspepsia, it was not observed to the same extent in three of the four other previous large studies.10, 12, 13, 19–21 Characteristics of patients as well as criteria chosen to define therapeutic response are of pivotal importance in the response rate to placebo. Most of the patients included were not recruited from hospital clinics but from gastroenterologists in private practice. Although these patients cannot be considered as primary care referrals, they represent non-ulcer dyspepsia patients referred by primary care physicians mainly for endoscopies in order to confirm the diagnosis of non-ulcer dyspepsia. As far as criteria of therapeutic response are concerned, the sensitivity of the scale for evaluating symptoms is a likely key-point.21 Hence among the five large published studies (including the present one), higher rate responses were observed in the two studies using the less discriminant scales with a 4-point scale in the study of Talley et al. and a 5-point scale in the present study.11 In our study, the main criterion had been defined as a decrease of at least 2 points in the symptom score. Consequently, successful treatment could be observed in patients moving from an initial score of 4 to a final score at 12 months of 2 (i.e. moderate symptoms that cannot be forgotten without influencing daily activities). This is unlikely to be sufficient reduction to consider this change as a success. In three other studies reporting lower response rates, more detailed scales were used, i.e. a 7-point scale or even a larger one in the study of McColl et al.10, 12, 13 These differences, and especially the less specific criterium for defining therapeutic success, probably explain the high percentage of response to both treatments in our study compared with the other previous randomized, double-blind, placebo- (or omeprazole-) controlled trials with 1 year of follow-up (Table 5). This high placebo response makes it difficult to demonstrate a therapeutic benefit unless a much larger number of patients are included. A 7-day placebo run-in period, as performed in three of the four previous studies, is probably required in future trials to decrease the rate of placebo responders, although this last point has been recently challenged.20, 21
However, when using a stronger efficacy criteria, i.e. the complete relief of dyspeptic symptoms at 12 months, the proportion of healed patients was significantly greater in those receiving eradication treatment than in placebo-treated patients (43 vs. 31%). Using this analysis, percentages of therapeutic success observed in the eradication group as well as in placebo-treated patients are close to those of three other studies (Table 5).10, 12, 13 Although the difference is small, this last result does suggest that eradicating H. pylori may be of benefit in a subset of patients with non-ulcer dyspepsia.1 This is reinforced by a recent meta-analysis revealing a small but significant benefit of H. pylori eradication with a mean response rate at 1 year of 36% in the eradication therapy group vs. 28% in the placebo group.15 In addition, and though using a non-specific quality of life instrument, there was a significant improvement in two out of the eight dimensions in patients receiving eradication treatment. In our study, observed rates of H. pylori eradication were 70% at 3 months and 69% at 12 months in the RAC group. These figures are within the range of the rates observed in previous eradication studies conducted in France (between 61 and 77% according to the digestive disease), either with a proton pump inhibitor or H2-receptor antagonist.22–25 This relatively low rate of eradication at first seems to dilute the response. In fact it does yield an interesting result showing clearly that therapeutic success in the RAC group occurred significantly more frequently in eradicated patients (84%) than in patients with persisting infection (64%).
In contrast, the high rate of H. pylori eradication in the PLA group (9% at 3 months and 18% at 12 months) was rather unexpected. Although some antibiotic treatments during the study as well as some false negatives of the histological analysis at 12 months cannot be totally excluded, we do not have a clear explanation for this high percentage.
Unfortunately we were unable to detect any factor likely to predict ‘responders’. Although presence and severity of gastritis do not appear to be related to dyspeptic symptoms, the results of logistic regression analysis suggest that a low degree of inflammation of chronic gastritis would be associated with a more frequent total relief of symptoms. This point needs to be confirmed in future studies. Likewise, this analysis confirmed that the RAC treatment was significantly associated with a higher rate of complete relief of symptoms at 12 months. As indicated before, this result is at least partially related to H. pylori eradication.
Our results also show that chronic gastritis is still present in many patients at 12 months despite H. pylori eradication. This is especially true for antral gastritis. However, as expected, the severity, and especially the activity of the gastritis, decreased in patients who had received eradication treatment (Table 3).26–28 In addition, our results show that there is no relationship between the activity of gastritis and symptoms because the number of patients with a symptom score of 0 or 1 was not significantly different between both groups.
In conclusion, our study confirms that eradicating H. pylori does not constitute the key treatment of non-ulcer dyspepsia. However, the significant difference observed in patients with complete relief of symptoms does show that a subset of these patients can probably get a benefit from eradication treatment. Further studies should now mainly focus on the definition of characteristics (symptoms, histology, characteristics of bacteria, immune response of the host…) enabling us to predict a profile of potential responders to eradication therapy.
We are indebted to the following investigators for recruiting patients for the study:
C. Barberis (Talence); J. Benisty (Lunel); S. Beorchia (Lyon); R. Beyssac (Talence); E. Bismuth (Athis Mons); P. Bontemps (Meyzieu); P. A. Bounin (Cagnes sur Mer); J. M. Canard (Paris); M. Charbit (Levallois Perret); R. Colin (Rouen); P. Colonna (Marseille); G. Coulanjon (Issoire); P. Courtial (Nimes); J. L. Daigre (Loudeac); J. Y. Danielou (Les Sables d’Olonne); O. Delette (Lille); J. Diaz (Toulouse); C. Dray (Lunel); O. Duhamel (Béziers); E. Echinard (Bayonne); M. Escartin (Saint Jean de Luz); G. Fratini (Cagnes sur Mer); C. Fromenteau (Saint Jeande Luz); J. Gislon (Beziers Cedex); B. Humeau (Les Sables D’olonne); J. L. Kolopp (Metz); P. Kuder (Issoire); O. Lalaude (Rouen); B. Lamiraud (La Roche sur Yon); H. Lamouliatte (Bordeaux); J. Lapuelle (Toulouse); M. Le Rhun (Nantes); A. Le Texier (Cannes); P. Lelong (Lille); C. Lenoir (Nevers); M. Lhermie (Douai); H. Magois (La Roche sur Yon); A. Martin (Lyon); M. Mesner (Paris); M. Mignon (Paris); M. F. Napoléon (Tournefeuille); L. Parois (Villemoisson sur Orge); D. Pospai (Paris); C. Revol (Paris); G. Rozental (Montauban); D. Schmitz (Forbach); T. Spreux (Cannes); P. Talbi (Bordeaux); J. Therond (Saint Germain en Laye); C. Tournier (Limoges); G. E. Tubiana (Le Cannet); B. Vernisse (Saint Germain en Laye); G. Vigreux (Marseille); P. Zavadil (Loudeac).