Digestive Diseases and Sciences

, Volume 55, Issue 8, pp 2302–2308

Proton Pump Inhibitor Therapy Does Not Affect Hydrogen Production on Lactulose Breath Test in Subjects with IBS

Authors

  • David Law
    • GI Motility Program, Cedars-Sinai Medical Center
    • GI Motility Program, Cedars-Sinai Medical Center
Original Article

DOI: 10.1007/s10620-009-1010-2

Cite this article as:
Law, D. & Pimentel, M. Dig Dis Sci (2010) 55: 2302. doi:10.1007/s10620-009-1010-2

Abstract

Background

Evidence suggests a role for small intestinal bacterial overgrowth (SIBO) in IBS. Recently, the question has arisen whether the lactulose breath test (LBT) is abnormal in IBS subjects due to overlapping GERD and proton-pump inhibitor (PPI) usage.

Aim

The aim of this study was to compare the prevalence of an abnormal LBT in IBS patients either receiving or not receiving PPI therapy.

Methods

Consecutive Rome I positive IBS patients referred for LBT completed a questionnaire regarding their symptoms and medication use. All subjects then underwent an LBT. The prevalence of abnormal breath test results and hydrogen production were compared based on PPI usage.

Results

Of a total of 555 (429 female) subjects, 106 (19.1%) subjects reported current PPI use. Among those on PPI, 46.2% had a positive LBT. This was not different from the 56.3% positive LBT in non-PPI subjects (OR = 0.67, CI = 0.436–1.017, P = 0.06). No differences in hydrogen parameter were seen with PPI. The average amplitude of rise (first peak) in PPI users was 28.0 ± 35.3 ppm from baseline; in non-PPI users it was 27.5 ± 29.1 ppm (P = 0.89). The average rise in the second peak in PPI users was 48.5 ± 43.8 ppm from baseline; in non-PPI users, it was 49.3 ± 37.6 ppm (P = 0.87). The time to first peak in PPI users was 56.4 ± 23.0 min; in non-PPI users, it was 58.2 ± 26.1 min (P = 0.58). However, among subjects receiving PPI only 7.5% had methane detection on LBT, which is significantly different from the 15.4% of subjects not taking PPI.

Conclusion

PPI therapy does not effect hydrogen production on lactulose breath tests in IBS patients. However, there may be an effect on methane.

Keywords

Proton pump inhibitorIBSLactulose breath test

Introduction

Irritable bowel syndrome (IBS) is a common gastrointestinal ailment that affects a large portion of the world population [16]. The Rome I classification of irritable bowel syndrome initially defined the disorder as abdominal discomfort or pain associated with alteration in stool form and function [7]. While the cause of IBS is not certain, recent data suggest that gut bacteria may play a role.

The bacterial hypothesis for IBS suggests that a portion of IBS patients have excessive endogenous bacteria within the small intestine. This is commonly referred to as small intestinal bacterial overgrowth (SIBO). In an initial study, as many as 76% of IBS subjects were deemed to have SIBO based on a lactulose breath test [8]. The use of breath testing in IBS has been validated in case-control studies (age- and gender-matched) [912]. Based on this work, controlled trials of antibiotics have demonstrable benefit in patients with IBS [9, 1315], especially when symptom resolution coincides with bacterial eradication [9].

Two controversies have emerged in the use of breath testing in IBS. First, opponents of breath testing believe that bacterial levels are similar between IBS subjects and controls. The first controversy is that some argue the breath test does not distinguish IBS from healthy controls [16]. This study is problematic since it was not age- or gender-matched and the control subjects were so much younger than the IBS subjects such that the interpretation is difficult. Studies that have used age- and gender-matched controls suggest a difference [912]. Furthermore, recent data from Posserud et al. [17] suggest that there are indeed elevations in coliform counts among IBS patients compared to controls [17]. While coliform counts in the proximal intestine were significantly higher in IBS subjects compared to controls, there was no difference in the number of subjects with counts exceeding 105 cfu/ml.

The latest controversy is the effect proton pump inhibitors may have on breath testing. There are many studies that suggest a clear overlap between IBS and gastroesophageal reflux disease (GERD). In fact, this was recently summarized in a systematic review of the data [16]. With a high degree of overlap between these conditions, it is expected that many IBS subjects are receiving acid suppressive therapy. Recently, Spiegel et al. [18] raised this issue in an opinion piece, highlighting that the incidence of concurrent proton pump inhibitor (PPI) usage in IBS patients is typically greater than the general population [18]. Based on previous literature, they discuss data supporting that PPI therapy may promote SIBO, and by consequence result in a positive lactulose (and/or glucose) breath test. They term this the “PPI hypothesis.” Furthermore, Spiegel et al. [18] raise concern that the existing studies linking SIBO to IBS have not adjusted for or excluded the use of PPI therapy.

In this paper we perform a retrospective analysis on a prospectively collected dataset to explore the use of PPI usage and its effect on SIBO, and specifically the lactulose breath test findings in IBS patients.

Methods

Patient Population

Subjects referred to a tertiary care medical center for a lactulose breath test (LBT) were prospectively recruited to participate in a study to database subjects referred for breath testing. This prospectively collected dataset was now retrospectively used to assess the role of PPI therapy in IBS. Patients were eligible for this study if they met Rome I criteria for IBS [7]. Subjects were excluded if they had a history of inflammatory bowel disease, previous intestinal resection, unstable thyroid disease, diabetes or chronic narcotic medication use. The study was approved by the institutional review board and all subjects were enrolled under informed consent.

Primary Study Design

Subjects presented to the medical center having fasted for 12 h. All subjects were asked to abstain from legumes on the evening before testing and to refrain from smoking on the day of testing. Subjects were encouraged to brush their teeth on the morning of the LBT. Prior to conducting the breath test, subjects were given a questionnaire that included symptoms, past medical history, current medication usage, and demographic information. This was done before breath testing to avoid any sensation from the lactulose (e.g. bloating) from influencing their symptom reporting in the questionnaire. Once the survey was completed, a baseline breath sample was obtained followed by ingestion of 10 g of lactulose (Inalco, Milano, Italy; packaged by Xactdose, South Beloit, IL) with up to 250 ml of water. Breath samples were then collected at 15-min intervals for 180 min. End expiratory breath samples were taken to ensure alveolar gas sampling. This was achieved via a 750-ml gas collection bag (Quintron Instrument, Milwaukee, WI). Samples were analyzed for hydrogen, methane, and carbon dioxide, reported in parts per million (ppm), using a Model SC, Quintron gas chromatograph (Quintron Instrument). Carbon dioxide measurements were used to correct for the quality of alveolar sampling. Measurements were plotted graphically.

Breath Test Analysis

The breath tests were analyzed in a comprehensive way so as to evaluate any possible effects of the PPI. For instance, the PPI may not change whether a breath test is positive or not, but simply result in the hydrogen rise occurring earlier in these subjects. First, a breath test was considered positive if there was a rise in breath hydrogen gas ≥20 ppm above baseline at or before 90 min from the time of ingestion of lactulose. Two additional criteria were also considered. These included an absolute rise of hydrogen gas of ≥20 ppm irrespective of baseline and the number of subjects with ≥20 ppm at any point during the first 180 min. In addition to this, each breath test was evaluated for multiple parameters as outlined in Fig. 1. This included, among others, variables such as the time to first hydrogen peak and the amplitude of hydrogen production (Fig. 1). The entire cohort of IBS patients was compared based on the presence or absence of PPI use for all LBT parameters. Subsequently, a second comparison was made among only those IBS subjects with a positive breath test. In addition, a final comparison was made to determine if there was a difference in the presence of methane on breath testing based on the presence of absence of PPI usage.
https://static-content.springer.com/image/art%3A10.1007%2Fs10620-009-1010-2/MediaObjects/10620_2009_1010_Fig1_HTML.gif
Fig. 1

Anatomy and analysis of the lactulose breath test. (a) Amplitude of baseline hydrogen production (ppm). (b) Amplitude rise from baseline for first peak (ppm). (c) Amplitude rise from baseline for second peak (ppm). (d) Time to first peak (a rise of ≥5 ppm over baseline) in hydrogen production (min). (e) Time to maximum point of hydrogen production (min)

Statistical Analysis

The number of positive LBTs in PPI users was compared with the number of positive LBTs in non-PPI users using a chi-square test expressed as an odds ratio and confidence interval. A 95% confidence interval was used to determine significance. To compare the different mean values for the various qualities of the breath test results between the two groups, a Wilcoxon rank sum analysis was conducted.

Results

Subject Characteristics

A total of 555 IBS subjects were eligible for analysis based on the Rome I criteria and did not have the previously mentioned exclusion criteria including unstable thyroid disease, diabetes, IBD, chronic narcotic use, or previous intestinal resection. Among these subjects, 106 (19.1%) were actively taking a PPI medication at the time of breath testing. The demographic data among patients on or off PPI were not different, as seen in Table 1.
Table 1

Subject demographics

Demographic

Non-PPI use (%) (N = 449)

PPI usage (%) (N = 106)

P-value

Male

98 (21.8)

28 (26.4)

0.31a

Female

351 (78.2)

78 (73.6)

 

Mean age (years)

43.1 ± 15.8

46.1 ± 16.0

0.069

PPI proton pump inhibitor

aThis P-value reflects the analysis of gender and PPI use

PPI Versus Non-PPI in All IBS Subjects

Out of the 555 subjects, 106 reported taking a PPI at the time of their breath test. Among these 106 subjects, 49 (46.2%) fulfilled the criteria for a positive breath test based on a 20 ppm rise in hydrogen at or before 90 min. This was not significantly different from the non-PPI group, where 253 of the 449 (56.3%) had a positive breath test (OR = 0.67, CI = 0.44–1.02) (Table 2). Using the criteria of any rise of 20 ppm within the 180 min test, 88 out of 106 (83.0%) of PPI using IBS patients met the criteria compared to 349 out of 449 (77.7%) non-PPI using IBS subjects (OR = 1.40, CI = 0.81–1.42).
Table 2

Effect of PPI use on the proportion of subjects with SIBO and amplitude of hydrogen production

Parameter

Non-PPI

PPI

P-value

Δ20 ppm by 90 mina

 n (%)

253 (56.3)

49 (46.2)

0.07

 Mean amplitude

45.5 ± 27.0

53.0 ± 38.5

0.20

≥20 ppm by 90 minb

 n (%)

281 (62.6)

64 (60.4)

0.74

 Mean amplitude

49.9 ± 28.6

51.6 ± 38.5

0.73

≥20 ppm any timec

 n (%)

349 (77.7)

88 (83.0)

0.29

 Mean amplitude

61.2 ± 34.2

59.8 ± 43.3

0.77

PPI proton pump inhibitor, SIBO small intestinal bacterial overgrowth

aSubjects with ≥20 ppm rise in hydrogen production over baseline

bSubjects with any value of ≥20 ppm before 90 min

cSubjects with hydrogen levels ≥20 ppm at any point during the 180 min test

On comparing each breath test parameter from Fig. 1 between subjects using and not using PPI, no difference was seen with any parameter (Table 3). In particular, there was no difference in the time to first peak hydrogen production to suggest a more proximal bacterial proliferation. Furthermore, the amplitude of the first peak was not different either.
Table 3

Comparison of breath test parameters from Fig. 1 based on use of acid suppressive therapy

Breath test parameter

Non-PPI users (n = 449)

PPI users (n = 106)

P-value

(a) Baseline breath H2

6.52 ± 6.4

6.41 ± 7.5

0.88

(b) Measure 3 in all patients

27.5 ± 29.1

28.0 ± 35.3

0.89

(c) Amplitude of second H2 peak (ppm)

49.3 ± 37.6

48.5 ± 43.8

0.87

(d) Time to first H2 peak (min)a

56.4 ± 23.0

58.2 ± 26.1

0.58

(e) Time to maximum rise in H2 (min)b

99.1 ± 28.4

95.8 ± 27.6

0.29

PPI proton pump inhibitor, H2 hydrogen, ppm parts per million

aOnly among subjects with a rise of greater than 5 ppm above baseline ≤90 min (n = 313 in non-PPI and 76 in PPI users)

bOnly in subjects with a detectable peak (n = 406 for non-PPI and 98 for PPI users)

PPI and the Prevalence of Methane on LBT

Despite the relative lack of effect of PPI on hydrogen during the lactulose breath test, the presence of PPIs seemed to be associated with a change in prevalence of methane on the breath test. Among the 555 subjects, 78 subjects (14.1%) had detectable methane. Among the subjects taking PPI, only eight (7.5%) were methane producers compared to 70 (15.6%) among non-PPI users (OR = 0.39, CI = 0.18–0.82) (Fig. 2). While the presence or absence of methane was different between users and non-users of PPI, the amplitude of methane production (ppm) among methane producers between the two groups was not different.
https://static-content.springer.com/image/art%3A10.1007%2Fs10620-009-1010-2/MediaObjects/10620_2009_1010_Fig2_HTML.gif
Fig. 2

Comparison of methane producers between PPI users and PPI non-users (P = 0.01)

Discussion

In this study, we demonstrated that there is no statistically significant difference between prevalence of a hydrogen positive LBT in IBS patients who report taking a PPI and those not taking one. After an extensive analysis of the breath tests, there is also no change in any of the dynamics of hydrogen production between those using and not using PPI. However, there is an association between the use of PPI and the presence or absence of methane on the breath test.

In the last decade, there has been a growing interest in the role of gut bacteria in the pathophysiology of IBS. Specifically, initial studies in this area used lactulose breath testing to determine the presence of bacterial overgrowth [8]. This area of work remains controversial for the sole reason that there is no gold standard test for the diagnosis of SIBO [19]. Despite this problem, numerous studies have conducted breath testing in IBS. While the results are mixed, nearly all of the studies which use age- and gender-matched controls demonstrate a significant difference between IBS subjects and controls [912]. Adding to the possibility that coliforms in the small bowel may be a cause of symptoms in IBS, a recent study by Posserud et al. [17] suggested that while the number of coliforms was not greater than 105cfu/ml, increased numbers of coliforms were seen in the small bowel of IBS subjects compared to controls [17].

Further indirect evidence for a gut flora problem in IBS is data suggesting that IBS subjects respond to antibiotics. There have now been four double-blind randomized controlled trials in which antibiotics have demonstrated significant efficacy over placebo in overall IBS improvement [9, 1315].

A commonly recognized association in IBS is its high rate of comorbidity with other functional GI disorders. These include an overlap with functional dyspepsia and GERD [19, 20]. The challenge is that both IBS and GERD are common in the population and thus they may overlap because they are both so common. However, in a recent systematic review of IBS literature, it seems that the IBS and GERD overlap cannot be easily explained based on simple common prevalence [19].

In the pretext that IBS and GERD co-exist, many subjects with IBS are likely to be taking PPI therapy. It has been suggested that proton pump inhibitors, by increasing stomach pH, can precipitate an increase in foregut contamination [21]. Based on this concept, Spiegel et al. [18] recently opined that perhaps PPI use is a cause of abnormal breath test findings and antibiotic response in IBS. In one study, culture of the gastric fluid demonstrated that 11 out of 30 subjects (37%) on PPI had colony counts suggestive of SIBO [22]. In a similar work, duodenal colony counts were also elevated in subjects taking PPI [23]. However, in this study, the relevance to the current discussion may not be accurate since the PPI use was for the treatment of duodenal ulcers, which depending on their size and nature could potentially lead to duodenal stasis. Despite studies suggesting an increase in bacterial colonization while taking PPI, breath testing has been less impacted by PPI use. In a study of elderly subjects taking PPI, duodenal bacterial counts were elevated in six out of 14 subjects on PPI [24]. However, none of the subjects had a positive lactulose breath test. In one further study, there appeared again to be elevated foregut bacteria in subjects on PPI [25]. In this study, multiple doses of PPI were examined and dose did not appear to be a factor in the development of increased bacterial flora. Based on the finding that PPI do raise bacterial counts in the foregut, it is possible to surmise that the breath test findings in IBS subjects could be related in some instances to the use of PPI. One difficulty with this concept is that the use of PPI appears discordant with findings on breath test [23, 24].

Besides the potential argument that PPI could raise the bacterial counts in subjects, Spiegel et al. [18] further suggest that perhaps this is associated with a known side effect of diarrhea among PPI users [18]. This concept is contradictory to their suggestion that the SIBO hypothesis is limited based on its inability to explain other proposed models of IBS such as the biopsychosocial, visceral hypersensitivity and inflammatory models. First, none of these proposed models of IBS (including SIBO) have been definitively proven as the ultimate cause of IBS. So, is the “PPI hypothesis” being suggested as the culprit in abnormal breath testing? Or, is this “PPI hypothesis” being charged as the cause of IBS since the authors declare it can cause diarrhea. If it is the latter or both, perhaps patients with IBS on PPI should be excluded from all clinical trials since we cannot explain the confounding effect of PPI on intestinal symptoms. In the case of bacterial overgrowth, a recent review of the literature suggests that although PPI affect the bacterial populations, there is only rarely a clinical change [21].

In the discussion above, it appears that previous studies in small numbers of patients demonstrate that PPI do not seem to affect the lactulose breath test. In our study we confirm this finding. Unlike the previous small studies, this study included a large number of subjects. Ironically, although not significant, it appears that there is a lower prevalence of positive lactulose breath testing among IBS subjects using PPI therapy (46.2%) compared to those not on PPI (56.3%). To further delineate the effect (or lack thereof) of PPI on breath testing, we conducted an extensive analysis of the breath test profile. This was done specifically to see if PPI might (by increasing proximal counts of bacteria) result in earlier production of hydrogen, among other factors (Fig. 1). This too demonstrated no difference between IBS subjects with or without PPI (Tables 2 and 3).

One difference was noted in the use of PPI and breath test. There appears to be a higher prevalence of methane production among IBS subjects not taking PPI. While it is known that methane on breath testing is associated with C-IBS [9, 26] and that methane may have physiologic effects on transit [27], the methane association with lack of PPI cannot be easily explained. In fact, studies suggest that there is no difference in the prevalence of GERD among diarrhea and constipation forms of IBS [28].

Why would PPI increase the level of foregut bacteria but not affect the breath test in IBS subjects? This may be difficult to answer. One issue may be the type of flora affected by PPI usage. In most of the studies, bacterial overgrowth precipitated by PPI use consists of oral flora excess [22, 24, 29]. In the case of SIBO in IBS, the hypothesis is that of coliform excess, not oral flora excess, as confirmed by Posserud et al. [17]. Another explanation may be the duration and breadth of bacterial colonization. In the case of coliform SIBO, the excessive colonization is presumed to involve the full 15 feet of small intestine. This is in contrast to the 8–12 inches of foregut that may be affected by PPI therapy. Since the pancreatic juices are both bacteriostatic and cidal, these secretions may confine the SIBO associated with PPI use to the proximal duodenum and stomach [30]. A final explanation is the effect of acid. How long do bacteria need to be exposed to acid in order to curb their number? Acid will either be bactericidal or dramatically inhibit the metabolic activity of non-acidophilic bacteria within minutes of exposure. Certainly, breakthrough acid production is seen with PPI therapy [31]. Perhaps this breakthrough production of acid (however brief) may suffice in reducing upper gut flora. This could explain a lack of PPI effect in this study.

A final important area of research has suggested that entry of pathogens might be aided by the use of a PPI medication. These include important diseases such as Clostridium difficile colitis, Campylobacter, and Salmonella [3234]. The difference here is that the organism may have been given the opportunity to transit through the stomach during times of maximal acid suppression (no breakthrough). Since the tendency for both PPI to be given before a meal and pathogens enter during a meal, the timing of these may be ideal.

The study has some limitations. While the results of this study suggest that PPI do not influence the production of hydrogen as detected by breath testing, the LBT remains an imperfect surrogate for SIBO. Future studies may need to examine the effect of PPI therapy on small bowel aspirates and culture in IBS subjects. Secondly, our questionnaire did not specifically ask subjects to report the length of time they had been taking their PPI or their daily dosages. This may be important in the development of a positive breath test. Finally, as this was a retrospective analysis of a prospectively-collected dataset, we were unable to establish a control group of solely GERD patients without IBS to assess if they had abnormal breath tests. Further studies will need to examine the effect of PPI on breath testing in healthy subjects and patients with GERD who do not concurrently suffer with IBS.

In conclusion, our results support the notion that proton pump inhibitors have no effect on hydrogen production as detected by lactulose breath tests in IBS patients as measured by several factors including positivity, baselines gas production, average peak values, and average times to such peaks. Although LBT remains an imperfect surrogate for SIBO, the utility of LBT is not dampened by the use of PPI medication and future studies may not need to exclude such patients from study. Finally, there seems to be a link between PPI use and the lack of methane on lactulose breath test. While this does not prove that SIBO is involved in the pathogenesis of IBS, it does imply that PPI therapy does not confound the results of previously published work on breath testing in IBS.

Acknowledgments

We would like to thank the Beatrice and Samuel A. Seaver Foundation as well as the Singleton Foundation for their financial support of this study.

Copyright information

© Springer Science+Business Media, LLC 2009