SEARCH

SEARCH BY CITATION

Keywords:

  • meta-analysis;
  • non-erosive reflux disease;
  • proton pump inhibitor;
  • treatment

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Background:  A lack of published data remains on factors associated with the treatment response to proton pump inhibitors (PPI) in patients with non-erosive reflux disease (NERD).

Methods:  Studies on effectiveness of PPI in patients with NERD were identified using the MEDLINE database and manual searches of the available literature. Twenty publications that comprised a total of 22 study arms were eligible, and meta-analyses of these 22 study arms were carried out.

Results:  Overall improvement rate in patients with PPI was 68%. Univariate meta-regression analysis was used to explore the source of heterogeneity, P-value for males was 0.192, and in the presence of Helicobacter pylori (H. pylori) infection was 0.186, indicating that these factors are associated with treatment effectiveness. Thus, as the frequency of male and H. pylori-infected patients increased, the effectiveness of PPI against NERD improved. By multivariate meta-regression analysis, the P-value for male and H. pylori infection decreased further (P = 0.002 and P = 0.003, respectively), indicating little interaction between male sex and H. pylori infection.

Conclusion:  These results suggest that H. pylori infection and patient sex may be associated with the effectiveness of PPI against patients with NERD. The reasons are still unclear, and further examinations are needed to clarify them.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Gastroesophageal reflux disease (GERD) is defined as a condition that develops when reflux of stomach contents causes troublesome symptoms or complications.1 Both reflux esophagitis (RE) and non-erosive reflux disease (NERD) are included in GERD; RE can be characterized by endoscopically detectable mucosal damage, whereas mucosal breaks are not seen endoscopically in NERD, despite the presence of symptoms.2 The majority of patients with GERD have NERD, not only in the West but also in Asia.3,4

Proton pump inhibitors (PPI) are widely given to patients with GERD.5 Systematic review data suggest that patients with NERD respond less well to acid suppression therapy than patients with RE.3 Complete resolution of symptoms such as heartburn has been reported to occur in 57–80% of the patients after PPI administration. This indicates that there is a considerable group of patients, approximately 30%, who may not have true acid-induced reflux symptoms in this condition. It would be clinically useful and potentially cost-effective to be able to predict the patients with NERD that would be highly likely to respond and those that would not respond to treatment with PPI; however, there remains a lack of published data on factors that are associated with treatment response to PPI in patients with NERD. Meta-analysis is a statistical method used to integrate and analyze previous research results, and it can identify factors that are associated with the effectiveness of PPI as sources of heterogeneity.6,7 Thus, meta-analyses were carried out in order to identify the factors associated with treatment response in patients with NERD.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Selection of studies

Studies on the effectiveness of PPI in patients with NERD were identified by computerized and manual searches of the available literature. The MEDLINE database (January 1951–March 2008) was searched using the following medical subject headings: non-erosive reflux disease, endoscopy-negative reflux disease and symptomatic reflux disease. Only papers in English were considered. Data available in the full paper were reviewed. Additionally, a recursive manual search of cited references in published studies was carried out to locate other studies.

Studies were eligible for inclusion in the meta-analysis if they met all of the following criteria: (i) patients included were diagnosed with NERD; (ii) PPI was given to patients; (iii) the study was conducted as either a randomized, double-blind trial (RCT) or a non-randomized trial (NRT); (iv) criteria for successful treatment were clearly defined, and treatment success was not measured in terms of illness severity scores nor the intensity of each symptom; and (v) treatment lasted for 2 weeks or more.

The search yielded a total of 115 publications, 66 of which were excluded by screening of the titles, and 29 that were excluded by screening the abstracts. Meta-analyses were carried out on the 20 remaining publications, which comprised a total of 22 study arms.8–27 One publication contained three study arms.17Table 1 shows all the 22 study arms, 12 arms were NRT and 10 arms were RCT.

Table 1.  Twenty-two study arms on which the meta-analysis was carried out
AuthorCountryPPI arm (improved/treated)Placebo arm (improved/treated)YearJournal
  1. PPI, proton pump inhibitors.

Bate CM8UK56/9821/1111996Aliment Pharmacol Ther
Watson RGP9UK12/18 1997Gut
Lind T10Denmark223/40425/1051997Scand J Gastroenterol
Lind T11Denmark, Sweden215/28180/1431999Aliment Pharmacol Ther
Richter JE12USA168/22426/1132000Arch Intern Med
Talley NJ13Denmark, Finland, Norway, Sweden144/17083/1722001Aliment Pharmacol Ther
Miner P Jr14USA66/1279/642002Am J Gastroenterol
Talley NJ15UK, Ireland, Canada418/47310/712002Eur J Gastroenterol Hepatol
Katz PO16USA211/45147/2272003Aliment Pharmacol Ther
Armstrong D-117Canada, UK, Ireland749/1282 2004Aliment Pharmacol Ther
Armstrong D-217France Germany, Switzerland479/693 2004Aliment Pharmacol Ther
Armstrong D-317Denmark, Finland, Norway, Sweden407/670 2004Aliment Pharmacol Ther
Bytzer P1814 European countries241/27994/1392004Aliment Pharmacol Ther
Ponce J19Spain17/18 2004Dig Dis Sci
Ang TL20Singapore144/216 2005World J Gastroenterol
Kahrilas PJ21USA34/1054/1062005Dig Dis Sci
Fock KM22Singapore64/74 2005World J Gastroentenrol
Fujiwara Y23Japan43/50 2005Aliment Pharmacol Ther
Kinoshita Y24Japan19/25 2006J Gasroenterol
Fernandez MC25Spain41/51 2006Rev Esp Enferm Dig
Talley NJ26USA1549/2458 2006Aliment Pharmacol Ther
Miwa H27Japan35/97 2007Aliment Pharmacol Ther

Data abstraction

To avoid bias in the data abstraction process, two investigators (T. H. and Y. U.) independently abstracted the data from the publications and compared the results. All data were checked for internal consistency and disagreements were resolved by discussion. The details abstracted from the reports included the name of the first author, year of publication, location of the study, age of participants, male-to-female ratio, number of participants, status of Helicobacter pylori (H. pylori) infection and number of treatment responses.

Quantitative data synthesis

All statistical analyses were conducted with STATA ver. 10 software (College Station, TX, USA). A P-value of less than 0.05 was accepted as a statistically significant test result for the summary effect of a drug. Adjustments were not made for multiple comparisons as the study lacked statistical power and we had an a priori hypothesis.28–30

The probability of attaining symptom relief was used as the summary statistic for the treatment effect. In the ith trial, the treatment effect of the drug was defined as the probabilities (Pi), ratio of number of subjects who attained symptom relief defined in the study and number of subjects who received PPI treatment in the study (Nitotal). Variance of Pi was calculated as inline image.

A random effects model was used to calculate the summary statistics. The between-study variation (τ2) was derived from the Q statistic according to a previously described method.31 The I2-statistic, which was an indicator for heterogeneity, was also reported. Meta-regression analysis was used to detect the sources of heterogeneity.32 Factors examined in the meta-regression analysis were male sex, the presence of hiatal hernia, H. pylori infection, tobacco smoking, alcohol use, Asian race and study design (RCT or NRT).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

The 22 study arms included in the analyses are summarized in Table 1. The studies were conducted in Europe (n = 10), North America (n = 5), Asia (n = 5) and both Europe and North America (n = 2).

First, we carried out meta-analyses using NRT and RCT (Fig. 1). The calculated improvement rate of the NRT was 64%, and that of the RCT was 68%; the overall improvement rate was 68%. The results of meta-analysis with all the trials are indicated in Fig. 2.

image

Figure 1. Results of meta-analysis using non-randomized trials and randomized-controlled trials.

Download figure to PowerPoint

image

Figure 2. Results of meta-analysis with all trials.

Download figure to PowerPoint

The results of univariate meta-regression analysis, exploring the source of heterogeneity, are shown in Table 2. The P-value for the male sex is 0.192, and that for the presence of H. pylori infection is 0.186, indicating that these factors are associated with treatment effectiveness. Thus, as the frequency of male and H. pylori-infected patients increased, the effectiveness of PPI against NERD improved.

Table 2.  Results of univariate meta-regression analysis exploring source of heterogeneity
 CoefficientP-valueN
  1. RCT, randomized,double-blind trial.

Male sex (%)0.73730.19222
Hiatal hernia (%)−0.26260.7077
H. pylori infection (%)0.49900.18616
Tobacco smoking (%)0.04920.94312
Alcohol use (%)−0.42770.31610
Asian race (yes vs no)0.02640.77322
RCT (yes vs no)−0.04390.56422

Table 3 shows the results of multivariate meta-regression analysis. The P-value for male and H. pylori infection decreased further in this analysis (P = 0.002 and P = 0.003, respectively), indicating little interaction between male sex and H. pylori infection. To confirm the results, we carried out a meta-analysis, stratified by frequency of male sex (Fig. 3). In the studies in which the frequency of male sex was at the median or below, the improvement was 65%. In the studies in which the frequency of male sex was above the median the improvement rate was 69%. The improvement rate of the group where frequency of male sex was higher than the median was certainly higher than that of the group that was equal to or below the median. Next, we carried out a meta-analysis, stratified by the frequency of H. pylori infection (Fig. 4). The improvement rate of the studies in which the frequency of H. pylori infection was at the median or below was 57%, and that of those in which the frequency of H. pylori infection was above the median was 72%. In regard to H. pylori infection, the improvement rate of the above median group was clearly higher than that of the group that was equal to or below the median.

Table 3.  Results of multivariate meta-regression analysis exploring source of heterogeneity
 CoefficientP-valueN
Male sex (%)2.20290.00216
H. pylori infection (%)1.14270.003 
image

Figure 3. Results of meta-analysis, stratified with frequency of male sex.

Download figure to PowerPoint

image

Figure 4. Results of meta-analysis, stratified with frequency of H. pylori infection.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

In the present study, it was shown that H. pylori infection and patient sex were associated with the effectiveness of PPI in patients with NERD. Thus, as the frequency of H. pylori-infected and male patients increases, the effectiveness of PPI against NERD improves.

To date, there are few published data on factors associated with treatment response in patients with NERD. Regarding the association between H. pylori infection and efficacy of treatment, Fujiwara et al. reported that famotidine significantly improved reflux symptoms in H. pylori-positive patients, but not in H. pylori-negative patients, whereas omeprazole significantly improved reflux symptoms irrespective of H. pylori infection.23 These results suggest that H. pylori infection may affect the efficacy of acid-suppressive drugs, especially histamine2-receptor antagonist (H2RA). In contrast, Fernandez et al. reported that H. pylori infection had no significant effect on response to treatment.25 There have been no other reports on the association between H. pylori infection and efficacy of acid-suppressive drugs in patients with NERD. From the Fujiwara et al. and Fernandez et al. reports, it is difficult to draw any conclusion; however, in the present study, through the use of meta-analyses, the association between H. pylori infection and treatment effectiveness was clearly indicated, as was an association between H. pylori infection and treatment effectiveness, suggesting that the presence of the infection may be one of the predictors of treatment success. The reason is unknown. H. pylori infection influences gastric secretion, and a change in gastric secretion might be one reason.

As for the association between patient sex and efficacy of treatment, there are no reports. This study is the first report to demonstrate the association. The possible reason is that the causative role of acid in female NERD may be less than in male NERD, suggesting that other factors such as motility disorder and psychological factors may have a much greater association with female NERD. It has been reported that NERD patients are more likely to be female than RE patients. Compared to RE, NERD has a number of different characteristics to RE, including low frequency of hiatal hernia, high frequency of H. pylori infection, severe glandular atrophy of the gastric mucosa, and frequent resistant to PPI therapy.3 These findings suggest that NERD may be a distinct entity from RE with a different pathophysiology. Although RE is a male dominant and acid-related disease, NERD can be a complex and multifactorial disease.

Regarding factors associated with PPI treatment response in patients with NERD, Talley et al. reported that assessment of heartburn resolution during the first week of therapy was the best predictor of treatment success at 4 weeks, and that patient demographics, duration of symptoms, medication used, other symptoms, and body mass index were not.26 These results seem to be convincing, because the acid suppression effect of PPI is observed within 1 week of starting daily treatment.33 However, it is noteworthy that the authors described that heartburn relief in any period during the first week of acid suppression therapy remained inadequate to enable a decision on whether to continue or cease therapy after 1 week.33

In the present study, the overall improvement rate in patients treated with PPI was 68%; more than 30% of patients did not have heartburn resolution with PPI therapy. A higher dose of PPI did not appear to provide any additional benefit in terms of symptom response, despite the fact that better acid suppression should be obtained with such dosing.5 Underlying pathophysiological disturbances causing heartburn, such as esophageal hypersensitivity, may partly account for the non-response, although a full explanation is currently lacking.34 NERD, therefore, remains a potentially difficult management problem. At present, PPI-based step-down treatment is recommended for GERD patients.35,36 It has been reported that PPI is a more effective treatment than H2RA.37 PPI treatment can eliminate NERD symptoms faster than H2RA treatment and has been reported to be more cost-effective than other treatments.33

GERD is more common in Western countries than in Asian countries, such as China, Korea and Japan. Epidemiological studies show a prevalence of GERD symptoms in Western countries ranging from 20% to 40% and in Asian countries ranging from 5% to 17%.3 One of our interests is whether there are any differences in treatment with PPI between Western and Asian patients with NERD. In other words, whether Western guidelines on GERD (NERD) are applicable to Asian patients without modification. In the present study, patient race is not a source of heterogeneity, indicating that there are no significant differences in treatment with PPI between Western and Asian patients with NERD. Thus, Western guidelines on GERD (NERD) are applicable to Asian patients without modification. The differences between Western and Asian countries on NERD may simply be in prevalence, rather than in the treatment efficacy of PPI.

We have proposed a new strategy for the treatment of NERD in Asia that is based on the basic idea of step-down therapy.3 The recommended first-choice treatment is giving PPI in combination with a prokinetic agent such as mosapride. PPI can cure only one-third of NERD patients, a prokinetic agent in conjunction with the PPI can increase the efficacy. NERD is frequently associated with functional dyspepsia that can be treated with prokinetic agents. Due to the rather low quality of life of NERD patients, quicker and more effective treatment options are needed.38 Some NERD patients are refractory to PPI and/or prokinetics. In these patients, psychological factors may be associated with symptoms and giving an antidepressant and/or minor tranquilizer should be considered. However, evidence of the benefits of these agents in treatment of NERD is weak,39 and further studies are needed to clarify the effects of such medications. For NERD patients with infrequent symptoms of heartburn, on-demand therapy with PPI (and/or prokinetics) is proposed as the best treatment option.40,41 Additional studies of the effectiveness of this treatment regimen are needed. At present, there is limited evidence regarding effective treatments for the disease; therefore, it is best to try several different treatment strategies in order to find the best treatment for each patient.

In conclusion, in the present study, we showed that H. pylori infection and patient sex might be associated with the effectiveness of PPI in patients with NERD through the use of meta-analysis. H. pylori infection was shown to have a greater association with treatment effectiveness than did patient sex. The reasons are still unclear, and further examinations are needed to clarify them.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References