Background
Several studies published the last few years support an association between small intestinal bacterial overgrowth (SIBO) and irritable bowel syndrome (IBS) [
1‐
4]. SIBO represents the overgrowth of bacterial species that usually predominate in the large bowel in the proximal small intestine. As a result of SIBO, fermentation of dietary carbohydrates by these bacteria leads to overproduction of gas and to the generation of symptoms of IBS. Most of existing studies linking SIBO and IBS use for the diagnosis of SIBO lactulose or glucose breath tests [
1‐
4]. However the gold standard of diagnosis is the culture of the content of the proximal part of the small intestine. In a recent survey of our group, 320 consecutive patients undergoing upper GI tract endoscopy were studied. Fluid was collected from the third part of the duodenum and quantitatively cultured. Using a cut-off of 10
3 cfu/ml of colonic type bacteria, the frequency of SIBO was 19.4 %. SIBO was significantly linked with odds ratio 5.64 with the presence of IBS (
p < 0.0001) [
5].
To dates, a number of papers suggest that the linkage between SIBO and IBS could be an epiphenomenon of the chronic intake of proton pump inhibitors (PPIs) leading to changes of the intestinal pH and promoting the colonization by large intestinal flora [
6,
7]. In a recent meta-analysis, an association was found between intake of PPIs and development of SIBO only for studies comparing incidence of SIBO before and post-treatment with PPIs within the same population but not when comparing with an independent control population [
8]. Available studies do not comment on the type of PPI intake, the dose regimen and the existence of an association between IBS and PPI intake [
6‐
8]. As a consequence, great heterogeneity is generated if indeed an association between SIBO and PPI intake exists. Our cohort of 320 patients using forward step-wise logistic regression analysis showed that the linkage between SIBO and IBS was independent from the intake of PPIs [
5]. Over the years, this prospective cohort has been expanded to 904 patients. We aim, using this cohort of 904 patients including the previously published 320 patients, to provide a definitive answer on the association between SIBO and PPI based on the gold standard for SIBO and to determine the association between IBS and PPIs as well.
Methods
Study design
This prospective study took place during the period September 2009 to March 2013. Patients who were subject to upper GI tract endoscopy in the Department of Gastroenterology of Sismanogleion General Hospital of Athens were eligible for the study. Every patient was allowed to be enrolled once after written informed consent. The study protocol was approved by the Ethics Committee of the Sismanogleion General Hospital of Athens.
Inclusion criteria were: a) age ≥ 18 years; b) written informed consent; and c) clinical indication for upper GI tract endoscopy. Exclusion criteria were: a) infection by the human immunodeficiency virus; b) chronic infection by the hepatitis B and hepatitis C viruses; c) Child Pugh liver cirrhosis stages 2 and 3; d) active GI tract bleeding; e) gastroesophageal reflux disease (GERD); f) systemic sclerosis; g) any antibiotic intake the last one month prior to endoscopy; and h) inflammatory bowel disease.
Fluid from the third portion of the duodenum was collected during upper GI tract endoscopy. In case of patients with little fluid in the duodenum, the endoscope was approached close to the intestinal wall to allow the aspiration of the biggest possible amount of fluid. Never water was flushed in the duodenal lumen before completion of fluid aspiration. The samples were immediately transported and quantitatively cultured using serial dilutions in sterile NaCl 0.9 % under aerobic conditions. An aliquot of 0.1 ml was plated onto MacConckey agar (Becton Dickinson, Conckeysville Md) and incubated for 18 h at 35
0C. Bacterial growth was determined after multiplying the number of isolated bacteria with the respective dilution factor. Identification of bacteria was done as described previously [
5].
The following information was recorded for each patient: age, gender, height, weight, reason for endoscopy, endoscopic findings, other diseases and intake of any medication. Specifically, full information regarding the type and the duration of PPI intake was registered. A patient was considered to have history of PPI intake if he/she had been administered one PPI for at least the last one month on a daily basis by his/her case-history. Adherence to PPI treatment was based on case-history and daily dose and start and stop dates were confirmed by the information provided in the prescription system for every single patient.
Based on their history, patients were classified as suffering from IBS if they met the following criteria: recurrent abdominal pain or discomfort at least 3 days per month in the previous 3 months associated with 2 or more of the following [
9]: i) Improvement with defecation; ii) onset associated with a change in frequency of stool; and iii) onset associated with a change in form (appearance) of stool.
Patients with IBS were further sub-classified based on their symptoms into three bowel habit subtypes ie diarrhea-predominant (IBS-D); constipation-predominant (IBS-C); and alternating bowel habit (IBS-A) according to the Rome criteria [
10].
The primary study endpoint was the relationship between SIBO and IBS in light of the history of PPI intake. The secondary study endpoint was the impact of PPI intake on the characteristics of SIBO and IBS. Since a total of 320 patients were needed in a previous study of our group to disclose a significant relationship between SIBO and IBS [
5] and making the hypothesis that 30 % of totally enrolled patients will have history of PPI intake, to analyse the association between SIBO and IBS separately among patients without PPI intake and among patients with PPI intake, it was assumed that a total of 900 patients should be enrolled.
Statistical analysis
In order to explore the primary study endpoint, SIBO was defined in three different ways using three different cut-offs of concentrations of colonic type bacteria in the duodenal aspirate ie >103, >104 and >105 cfu/ml. Comparisons of qualitative variables between patients with SIBO and those without SIBO was done by the Chi-square test. Odds ratios (ORs) and 95 % confidence intervals (CIs) were calculated by the Mantel-Haenzel’s statistics. ORs were compared by the Tarone’s and Breslow-Day’s tests. Comparisons of quantitative variables were done by the Student’s “t-test”. In order to define a cut-off value for a quantitative variable that can discriminate SIBO from non-SIBO with specificity greater than 90 %, receiver operator curve (ROC) analysis was done. Then forward step-wise logistic regression analysis was performed taking into consideration all variables differing between SIBO and non-SIBO patients at a p value lower than 0.100 and the intake of PPIs as well. Regarding the secondary endpoints, patients were divided into those with history and into those without history of PPI intake. Comparisons between these two subgroups comprised the counts of proximal intestine colonizers where the type of PPI, the duration of PPI intake and the clinical subtype of IBS were taken into consideration. 2-sided P values less than 0.05 were considered significant.
Discussion
Current findings challenge the concept that intake of PPIs is favoring the overgrowth of bacteria in the proximal small intestine leading to symptoms compatible with SIBO like bloating and diarrhea. The rate of PPI intake was similar between patients with SIBO and patients without SIBO as SIBO was determined by small intestinal aspirate culture. This was also the case with the absolute number of isolated bacteria whereas intake of PPIs was also not associated with the likelihood of SIBO. Instead, the logistic multiple regression analysis showed that the only factors independently associated with SIBO were the presence of IBS, T2DM and age ≥60 years. Gastritis and BMI ≥22 kg/m2 were protective from SIBO.
There is a traditional concept that alteration of gastric pH, as achieved after long-treatment with PPIs, can prime bacterial overgrowth and lead to SIBO. Despite this concept, existing evidence on the association of SIBO with PPI intake is not-clear cut. Quantitative culture results of the duodenal aspirates from 675 subjects who underwent upper GI endoscopy were retrospectively analyzed in relation with the history of PPI intake. Aspirates were cultured both for aerobe and anaerobe bacteria and patients were divided into those with negative aspirates, with intermediate aspirates yielding less than 10
5 cfu/ml and with abnormal aspirates yielding more than 10
5 cfu/ml. History of PPI intake was positively linked with the presence of intermediate aspirates; surprisingly it was not associated with abnormal aspirates [
6] showing that many factors other than intake of PPI prime the development of SIBO. In another study of 150 participants, a prospective design was followed. Participants underwent not only quantitative culture of their duodenal aspirate for the diagnosis of SIBO but manometry as well to identify intestinal dysmotility. Dysmotility and PPI intake were independently associated with a greater frequency of SIBO. However, the investigators did not provide data of an interaction between PPI intake and dysmotility for the generation of SIBO so it cannot be disclosed whether patients under PPIs were suffering from dysmotility as well [
11].
Prospective treatment of 52 patients with GERD with esomeprazole for 6 months was associated with the development of symptoms compatible with SIBO like bloating, flatulence and abdominal pain. In these same symptom-positive patients, SIBO was more frequent by the glucose breath test. Although findings suggest an association between PPI intake for 6 months and development of SIBO [
12], it should be outlined that the study did not have a comparator arm. Moreover, patients with GERD were excluded from our study so that direct comparisons cannot be done.
It should be underscored that one main advantage of the presented analysis is the consideration of time of PPI treatment that is missing from all former publications. Three main limitations of the study should be reported. The first is the lack of culture for anaerobes. The second is the small subgroups of patients with SIBO and IBS despite the large number of performed endoscopies. The third is the lack of use of a method for prevention of contamination during endoscopy. However, the aerobic bacteria isolated from the duodenum were mainly species usually inhabiting the large bowel. This makes the chance for contamination unlikely.
The lack of association between intake of PPIs and SIBO reinforces the recently developed therapeutic concept for SIBO eradication as a means for management of IBS. Although our study failed to identify an association between PPI intake and SIBO, it clearly showed a positive association between SIBO and IBS particularly with the IBS-D subtype. Although factors like T2DM, age above 60 years, gastritis and BMI above 22 kg/m
2 may modulate SIBO, the association between SIBO and IBS is independent from these factors. As a consequence, eradication of SIBO may be part of the algorithm for the management of IBS. A recent meta-analysis has confirmed treatment benefit in IBS with antibiotic treatment; this is pronounced with the non-absorbable rifaximin antibiotic [
13]. Results of the big TARGET1 and TARGET2 randomized clinical trials were not included in this meta-analysis. In these two trials, oral rifaximin 550 mg tid for ten days improved considerably signs of IBS compared with placebo when given in patients with IBS without constipation (40.8 % versus 31.2 %,
p = 0.01 in TARGET 1 trial; and 40.6 % versus 32.2 %,
p = 0.03 in TARGET 2 trial) [
14]. The in vitro activity of rifaximin was studied against the bacteria isolated from 117 of the patients included in our prospective cohort. To study the over-time killing effect of rifaximin, bile salts were added in the growth medium to simulate the intestinal environment. Pronounced time-kill effect was found at a concentration of 500 μg/ml against isolates of
Escherichia coli, Klebsiella pneumoniae and
Enterococcus faecalis that caused SIBO in our patients [
15]. This concentration is considerably lower than the concentration of the drug found in the stool [
16].
Conclusions
Analysis of a cohort of 897patients with prospective quantitative culture of the duodenal aspirates did not find any association between SIBO and intake of PPI. Instead the association between IBS and SIBO was completely independent from PPI intake. Although gastritis was protective against SIBO, results show that PPI intake cannot prime SIBO.
Abbreviations
BMI, body mass index; CFU, colony forming units; CI, confidence interval; GERD, gastroesophageal reflux disease; IBS, irritable bowel syndrome; IBS-C, irritable bowel syndrome predominant constipation; IBS-D, irritable bowel syndrome predominant diarrhea; IBS-M, irritable bowel syndrome mixed-type; OR, odds ratio; PPI, proton pump inhibitors; SIBO, small intestinal bacterial overgrowth; T2DM, type 2 diabetes mellitus
Acknowledgements
We thank Dr. Miltiades Kyprianou for his critical statistical review.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (
http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.