CRITICAL APPRAISAL OF THE SIBO HYPOTHESIS AND BREATH TESTING: A CLINICAL PRACTICE UPDATE ENDORSED BY THE EUROPEAN SOCIETY OF NEUROGASTROENTEROLOGY AND MOTILITY (ESNM) AND THE AMERICAN NEUROGASTROENTEROLOGY AND MOTILITY SOCIETY (ANMS)

Abstract

Background:

There is compelling evidence that microbe-host interactions in the intestinal tract underlie many human disorders, including disorders of gut-brain interactions (previously termed functional bowel disorders), such as irritable bowel syndrome (IBS). Small intestinal bacterial overgrowth (SIBO) has been recognized for over a century in patients with predisposing conditions causing intestinal stasis, such as surgical alteration of the small bowel or chronic diseases, including scleroderma and is associated with diarrhea and signs of malabsorption. Over 20 years ago, it was hypothesized that increased numbers of small intestine bacteria might also account for symptoms in the absence of malabsorption in IBS and related disorders. This SIBO-IBS hypothesis stimulated significant research and helped focus the profession’s attention on the importance of microbe-host interactions as a potential pathophysiological mechanism in IBS.

Purpose:

However, after two decades, this hypothesis remains unproven. Moreover, it has led to serious unintended consequences, namely the widespread use of unreliable and unvalidated breath tests as a diagnostic test for SIBO and a resultant injudicious use of antibiotics. In this review, we examine why the SIBO hypothesis remains unproven and, given the unintended consequences, discuss why it is time to reject this hypothesis and its reliance on breath testing. We also examine recent IBS studies of bacterial communities in the GI tract, their composition and functions, and their interactions with the host. While these studies provide important insights to guide future research, they highlight the need for further mechanistic studies of microbe-host interactions in IBS patients before we can understand their possible role in diagnosis and treatment of patient with IBS and related disorders.

INTRODUCTION

Before we attempt to address the morass that SIBO has become, let us trace the origins of this concept and how it morphed into the contentious entity that it is today. A look at the history of SIBO reveals that it originated as a cause of maldigestion and malabsorption among individuals who, because of an altered anatomy, were predisposed to small intestinal stasis and/or the recirculation of colonic contents into the small bowel. Originally often referred to as “the blind loop syndrome,” SIBO was recognized as representing the consequences of bacterial overgrowth in, or “contamination” of, the small intestine leading to a malabsorption syndrome. The earliest reports of the entity, from the late 19^th and early 20^th centuries, described the development of megaloblastic anemia among individuals with small intestinal disease, including strictures (1, 2). Later, additional features such as diarrhea and steatorrhea were added to the symptom list, and other predisposing entities such as various surgical procedures and diseases such as jejunal diverticulosis and scleroderma were recognized (3, 4). In the latter half of the 20^th century, elegant research described how bacteria could disrupt bile acid metabolism, lead to a deficiency of vitamin B₁₂ as well as fat-soluble vitamins and the full spectrum of the clinical, biochemical and pathological sequelae of SIBO, as it was then understood (i.e., as a disorder of maldigestion and malabsorption), were delineated (5–7). Diagnosis rested on the combination of a compatible clinical scenario, the demonstration of increased numbers of bacteria (especially, colonic-type microbes) in aspirates taken from the jejunum (not the duodenum) and a response to antibiotic therapy. The diagnostic cut-offs for bacterial numbers, though much criticized nowadays, were based on compatibility with a real and objectively defined clinical entity and correlated with other tests of malabsorption (such as stool fat) or bacterial activity (such as urinary indican levels) (6, 8). In other words, this original concept of SIBO was based on a definable clinical presentation and underpinned by evidence that could explain the pathophysiology of resultant symptoms and laboratory findings. In research from that era, coliforms loomed large as culprits responsible for many of these disturbances of gut function (6). In the pre-molecular age, conventional bacteriological technologies provided some insights into the bacterial population of the upper gut and identified those factors that normally keep these populations in check (such as gastric acid, an intact neuromuscular apparatus, pancreatic secretions, for example); observations that expanded the list of risk factors for the development of SIBO (5).

Over time, additional clinical scenarios linked to SIBO emerged – chronic, unexplained diarrhea in the elderly, D-lactic acidosis, encephalopathy and protein-losing enteropathy were recognized; in all instances, the clinical features could be explained based on bacteria-host interactions (9). Alternative diagnostic tests were explored, such as breath tests based on bacterial metabolism of bile acids or d-xylose or the measurement of indican in the urine, derived from bacterial metabolism of tryptophan, but despite its invasive nature and many shortcomings, the aspiration and culture of jejunal fluid continued to rule supreme as the “gold standard” for the diagnosis of SIBO (9).

The lactulose hydrogen breath test (LHBT), originally introduced as a measure, an albeit imperfect, of oro-cecal transit time, has since entered the diagnostic armamentarium of SIBO. Its application to clinical conditions far from the clinical spectrum of maldigestion and malabsorption resulted in a dramatic change in the very concept of SIBO, which now came to be defined not based on perturbations of microbiome-host interactions, but on the results of a test which had scarcely been validated for this indication (vide infra). Most consequential was the application of the LHBT to the evaluation of a common disorder of gut-brain interaction, irritable bowel syndrome (IBS), leading to the diagnosis of SIBO in up to 80% of affected individuals (10) and the widespread prescription of broad-spectrum antibiotics to very large numbers of individuals who were so diagnosed. Meanwhile, the field of microbiology has undergone a seismic shift with the advent of molecular microbiological technologies, methodologies which should ultimately provide an answer to a question that is fundamental to our understanding of SIBO – what is the composition and metabolic portfolio of the small intestinal microbiome in healthy human subjects? Rather than directing our efforts to the application of metagenomics, metabolomics and metatranscriptomics to the study of the small intestinal microbiome in health and disease, we continue to apply flawed tests to broad categories of individuals and blithely ascribe positive findings to “bacterial overgrowth.”

Our task in this two-part review is to critically evaluate the status of breath testing in SIBO and the validity of the application of SIBO to irritable bowel syndrome and unexplained gastrointestinal symptoms such as bloating and distension. In Part I., we review the concept of breath testing and its serious limitations. In Part II. we discuss the dynamic state of bacterial communities in the intestine in response to the host factors, highlighting the need for much more study of the complexity of these communities in the small intestine, rather than focusing simply on absolute numbers, in order to understand how their metabolic output could generate symptoms in subsets of patients.

Part I. Breath Testing for SIBO in Disorders of Gut-Brain Interaction (DGBI): A flawed concept

• iThe lactulose hydrogen breath test (LHBT) basic concepts—a measure of orocecal transit time

  The fundamental flaw underlying the use of the LHBT to diagnose SIBO is the impact of the wide variation in transit time through the stomach and small intestine to the cecum. The LHBT was developed by Levitt and colleagues (11) many years ago precisely for this purpose - to measure oro-cecal transit time, not to measure bacterial overgrowth. As shown in Figure 1, lactulose, a non-absorbable sugar, transits through areas of relatively lower bacterial density in the stomach and small intestine (~10¹⁻⁵ duodenum, 10³⁻⁵ jejunum, 10⁶ ileum bacteria, mainly aerobes) until it reaches the cecum. Here, it is exposed to much greater numbers of bacteria, including anaerobes (~10¹² aerobes and anaerobes), which rapidly ferment the sugar and produce H₂ gas (and other gases, including CH₄ and H₂S). This production is the sole source of these gases, and they quickly diffuse into the bloodstream and can readily be captured in expired breath samples. One area of confusion raised by proponents of breath testing for SIBO is the time required for bacteria to ferment the sugars and produce the gases. The original validation studies of the transit time measurements, however, showed that infusion of even minute amounts of the sugar (0.5 g) directly into the cecum led to a rapid rise in H₂ ppm in expired breath samples within just a few minutes of the infusion (11, 12). The time from ingestion to the abrupt rise in H₂ thus represented the transit time from mouth to cecum.

• iiLHBT for diagnosing SIBO

  When applying the LHBT to diagnose SIBO, it was proposed that the overgrowth of sufficient numbers of bacteria in the small intestine results in an early rise in H₂ gases because the time from ingestion of the lactulose to the site of fermentation would be shorter (i.e., now in the small intestine rather than the cecum). It was also originally suggested that the LHBT was advantageous compared to the GHBT because it detected distal SIBO in the ileum, whereas the GHBT could not due to the proximal absorption of glucose (see also GHBT discussion below), a concept that was never validated and clouded by the recognition that increased numbers of bacteria in the ileum compared to the more proximal small bowel is physiologic. The fatal weakness of the LHBT in the diagnosis of SIBO is that the transit time to the cecum in many healthy individuals, as well as those with IBS, and particularly those with diarrhea-predominant IBS (IBS-D), can be remarkably short. The original reports of the application of the LHBT to study small bowel transit (validated by measuring the simultaneous arrival of a polyethylene glycol (PEG) marker in the distal ileum) in healthy volunteers (n=40)(11) found the average transit time was 72 min (with a very wide range extending from 25–118 min) and that lactulose caused a dose-dependent decrease in transit time, a finding mirrored in other studies (13). In this latter study (13) that used combined scintigraphy and lactulose, they also reported short transit times with a mean of 40 +/− 5 min and a recent expert consensus group (14) suggested that the normal oro-cecal transit time based on the LHBT ranged from 50 – 200 min. Studies also showed that comparing healthy volunteers to IBS patients (15), IBS-D patients had shorter small bowel transit times than even healthy volunteers (and IBS-C patients were longer). Thus, the potential for the transit time to be less than the proposed diagnostic cut-off of 90 min for a rise in H₂ as an SIBO diagnostic marker is even greater in IBS-D patients than in the high number of asymptomatic subjects. Consequently, as shown schematically by the dotted lines in Figure 1B, this transit time is faster than all of the cut-off times proposed for the definition of a positive test, thereby leading to a false positive result in a very high proportion of individuals. This fundamental flaw was directly demonstrated using a combined test meal of lactulose and Tc⁹⁹ sulfur colloid in IBS patients (16). Placing a gamma counter over the cecum while simultaneously recording breath H₂ measurements made it possible to determine if the test meal reached the cecum before or after the rise in breath H₂, as shown in Figure 2A. In almost every patient, the Tc⁹⁹ reached the cecum before the rise in H₂, thus confirming that the LBT was measuring oro-cecal transit, not SIBO, as shown schematically in Figure 2B. As shown here and by others (11, 15), the transit time was remarkably short for many patients (Figure 2C), and in almost all cases, the Tc⁹⁹ was found in the cecum at least 10 minutes before the rise in H₂ and greater than 20 minutes in over 40 % of patients. Given that minute amounts of lactulose can cause a significant rise in H₂ within only 2–3 minutes, as described above, there is more than sufficient time for fermentation of lactulose in the cecum.

• iiiThe glucose hydrogen breath test (GHBT) also lacks sufficient accuracy to diagnose SIBO in IBS patients.

  The GHBT has been proposed as an alternative to the LHBT for diagnosing SIBO in IBS patients. Conceptually, the advantage of this test is that glucose is absorbed proximally in the small intestine via Na⁺/glucose cotransporters and hence would not be susceptible to escape into the cecum with shorter transit times, such as may lead to the false positive tests seen with the LHBT. Indeed, the positivity rate with a GHBT is much lower than an LHBT (6–37% positive in IBS vs. 0.7 to 13% in controls, depending on the cut-off value used)(17). However, combined scintigraphy and glucose breath testing showed that glucose can also escape into the cecum and that false positive rates were ~ 10 % with normal anatomy and even higher with prior surgery such as partial gastrectomy (18, 19). Studies using a highly sensitive telemetric capsule to monitor H₂ gas production (20) failed to detect significant H₂ PPM in the small intestine following glucose ingestion but were able to detect increased levels in the colon. Studies examining both GHBT and CFU/ml in jejunal aspirates failed to show a correlation even when the counts were >10⁵ CFU/ml (21–23) implying that elevated bacterial counts in the small intestine in IBS patients may not be sufficient to produce abnormal levels of H₂ and CH₄ following glucose ingestion, whereas in pathological states such as scleroderma where bacterial counts can be much higher, detectable levels of these gases originating from the small intestine may occur. Furthermore, a recent real-world study of over 1000 patients (24) showed that the positivity rate of the test in patients suspected of functional bowel disorders was less than 2%. Based on this discussion and further review of GHBT studies below (see iv, 4.), it is evident that the performance characteristics of GHBT in this patient population are also poor and do not appear to be acceptable for routine clinical use in patients with IBS.

• ivCritical review of IBS-SIBO studies

  Proponents of breath testing for IBS patients have based this recommendation on a number of guiding principles. To assist readers, here we examine four that have either failed to be substantiated in subsequent studies or are often misrepresented by proponents of breath testing for SIBO in IBS patients:

Figure 1. Basic concept underlying the LBT for diagnosing SIBO and negative impact of transit time on validity.

Lactulose, a non-absorbable carbohydrate, is ingested (left) and fermented within a few minutes by the high numbers of microbes upon reaching the cecum. Fermentation produces hydrogen gas (H₂; as well as other gases, including CH₄ and H₂S, not shown), which rapidly diffuses into the circulation and can be collected in exhaled breath samples. The transit time from the mouth to the cecum (right panel, dark blue line) is the time from ingestion to the rise in H2 PPM about the specified cut-off level (e.g., 20 PPM). It is proposed in cases of SIBO that fermentation occurs earlier because the microbes are now increased in the small intestine. The result is an earlier rise in the H₂ in the breath samples (right panel, purple line) above the specified cut-off value. However, transit time is highly variable in healthy subjects, resulting in a “false positive” early rise in H₂ levels that are being attributed to increased bacteria in the small intestine rather than simply a faster transit time to reach the cecum (right panel, light blue lines).

Figure 2. Combined Lactulose – ^99mTc scintigraphy shows the test meal is in the cecum before H₂ rises in IBS patients.

A. Schematic drawing showing gamma counter placed over cecum to detect ^99mTc sulfur colloid in the lactulose test meal, enabling the precise time lactulose has arrived in the cecum to be recorded. The time course panel below shows that in 39 of 40 IBS cases, the ⁹⁹Tc was in the cecum before the H₂ began to rise. This confirmed that lactulose was being fermented in the cecum and thus provided a measure of transit time and not overgrowth of bacteria in the small intestine. If the latter occurred, H₂ gas would have risen before the test meal reached the cecum. B. The transit time is highly variable in IBS patients, and more than half of the patients have a transit time, as measured by the ^99mTc in the cecum, that is less than the specified 90 min for the rise in H2 levels for a positive SIBO test. C. .5 g of lactulose in the cecum (5% of the 10 g test meal) is sufficient to cause a significant rise in H₂ levels within a few minutes (see text). The validation study shows that at least 5% of the meal was in the cecum in all cases before the H₂ levels rose, and in the majority of cases was > 20%. Reproduced with permission from BMJ Publishing Group Ltd. [Simrén M, Barbara G., Flint HJ, Spiegel BMR, Spiller RC, Vanner S, Verdu EF, Whorwell PJ, Zoetendal EG. Intestinal microbiota in functional bowel disorders: a Rome foundation report, Gut, 62(1):159–176, 2013, doi: 10.1136/gutjnl-2012-302167; and Yu D, Cheeseman F, Vanner S. Combined oro-caecal scintigraphy and lactulose hydrogen breath testing demonstrate that breath testing detects oro-caecal transit, not small intestinal bacterial overgrowth in patients with IBS, Gut, 60(3):334–340, 2011, doi: 10.1136/gut.2009.205476].

1). The majority of IBS patients have SIBO based on the LHBT.

The landmark studies (10, 25) suggested that almost 80% of IBS patients had SIBO. When these findings were initially challenged, the IBS-SIBO literature became increasingly difficult to follow as proponents of the LHBT repeatedly changed the diagnostic criteria over the ensuing years, as shown in Figure 3. Regardless of the cut-off value used for H₂ gas, e.g., 20 PPM by 90 or 180 min, multiple studies showed that the LHBT did not discriminate IBS patients from healthy controls at any time point, as shown in the examples in Figure. 4. The additions of methane (CH₄) and, more recently, hydrogen sulphide (H₂S) as diagnostic criteria have never been validated. Methanogens, such as Methanobrevibacter smithii, are found in most healthy controls typically in the colon (26). Thus, knowing where CH₄ originates when detected on a breath test is impossible. The putative “gold standard” for SIBO, culture from jejunal aspirates, was also changed over time from 10⁵ CFU/ml to 10³CFU/ml in the jejunum and then to 10³CFU/ml in the duodenum (Fig. 3) by authors simultaneously refuting the existence of any gold standard (27) and who have not validated this new standard. To quote the North American Consensus, “the optimal criterion to define excessive methane production is not clear.”

Figure 3. Criteria for a positive LBT for diagnosing SIBO: a moving target over time.

Proponents of the LBT have over time changed the criteria for a positive test, presumably to improve the accuracy. None of these criteria have, however, been validated, leading to confusion in the field. In parallel, it has been recognized that the “gold standard” using jejunal cultures with >105 CFU/ml also lacked sufficient accuracy, and proponents have introduced new criteria for culture techniques. However, these too have not been validated.

Figure 4. LBT studies in IBS patients showing that no time point differentiates healthy controls from IBS patients.

Three independent studies, one from Canada, Sweden and the US, show that the recommended 90-minute cut-off for a rise in H₂ does not discriminate between healthy controls and IBS patients and that no time point appears to offer any discriminative value. Reproduced and adapted with permission from BMJ Publishing Group Ltd. [Posserud I, Stotzer P-O, Björnsson ES, Abrahamsson H, Simrén M. Small intestinal bacterial overgrowth in patients with irritable bowel syndrome, Gut, 56(6):802 – 808, 2007, https://gut.bmj.com/content/56/6/802.long]; and from Wolters Kluwer Health, Inc. [Walters B, Vanner SJ. Detection of bacterial overgrowth in IBS using the lactulose H2 breath test: Comparison with 14C-d-xylose and healthy controls, Amer J Gastroenterol, 100(7): 1566 – 1570, https://journals.lww.com/ajg/abstract/2005/07000/detection_of_bacterial_overgrowth_in_ibs_using_the.25.aspx; and Bratten JR, Spanier, J, Jones MP. Lactulose breath testing does not discriminate patients with irritable bowel syndrome from health controls, Amer J Gastroenterol, 103(4):958 – 963, https://journals.lww.com/ajg/abstract/2008/04000/lactulose_breath_testing_does_not_discriminate.21.aspx].

2). Breath testing predicts the response to antibiotics and/or is normalized after antibiotics.

With any test, the critical issue for the clinician is whether the test results will impact clinical care – will it predict prognosis or therapeutic response? In relation to SIBO in IBS, the specific question is whether a positive breath test will predict response to antibiotic therapy. Regrettably, the literature on virtually any form of SIBO therapy, regardless of cause, is limited, and its interpretation is bedeviled by variations in the study population, study design (choice of antibiotic, dose, duration of therapy and follow-up) and clinical outcomes. Furthermore, many studies are observational or adopted an open design; few were placebo-controlled and head-to-head comparisons of different antibiotic regimens are few and far between. As summarized in Table 1, rates of eradication of SIBO or normalization of breath tests vary widely from 7–100%, with symptom response rates showing similar variability. Some studies, indeed, found no benefit from antibiotic therapy over placebo (28, 29). These studies were performed in various study populations – symptomatic individuals with SIBO, subjects with scleroderma and IBS sufferers defined according to various criteria, some enrolled children, and others, adults. Most studies featured rifaximin as the antibiotic intervention, and evidence for a dose-response relationship for breath test normalization does emerge. Study duration has been highly variable – an important issue given the report of Bae and colleagues on the impact of study duration on response (30).

Table 1.

Studies of antibiotic therapy in SIBO and related disorders.

Author	Year	Study Population	N	How SIBO was Diagnosed	Regimens	Eradication Normalization rate	Symptom Response Rates	Comments
Attar (81)	1999	SIBO with diarrhea	20	Hydrogen Breath Test (HBT)	Norfloxacin Amoxicillin/Clavulonic Acid S. boulardii	30% 50%	Both antibiotic arms superior to the probiotic
Pimentel (25)	2000	IBS (Rome I)	202	Lactulose HBT (LHBT) 78% positive	Neomycin and other regimens	Overall 53%	Diarrhea 62% Abdominal Pain 56%	Only 47 subjects had follow-up Treated for 10 days
Di Stefano (82)	2000	SIBO	21	Glucose HBT (GHBT)	Rifaximin 1200 g/dy Chlortetracycline 1g/dy	70% 27%	Symptom responses similar	Treated for 7 days
Castiglione (83)	2003	Crohn’s Disease	29	GBHT	Metronidazole 250 mg t.i.d. Ciprofloxacin 500 mg b.i.d.	87% 100%	85% 83%	Bloating Treated for 10 days
Pimentel (10)	2003	IBS (Rome I)	111	LHBT	Neomycin 500 mg b.i.d. Placebo	20% 2%	35% 11%	Primary outcome was a composite score Treated for 10 days
Lauritano (84)	2005	SIBO	90	GHBT	Rifaximin 600 Rifaximin 800 Rifaximin 1200	17% 27% 60%	Not recorded	All daily doses and treated for 7 days
Biancone (85)	2000	“Inactive” Crohn’s disease	14	HBT	Rifaximin 1200 mg/dy Placebo	100% 30%	No effect	Eradication rates are at 2 weeks – none eradicated at one month. Treated for 7 days.
Sharara (34)	2006	Bloating and flatulence (54–59% Rome II +)	124	LHBT negative in all	Rifaxmin 800 mg/dy Placebo	All negative but hydrogen breath excretion dropped	41% (41% 23% (18%)	Treated for 10 days. Reduction in hydrogen excretion correlated with improvements in bloating and flatulence.
Scarpellini (86)	2007	SIBO	80	GHBT	Rifaximin 1600 mg/dy Rifaximin 1200 mg/dy	80% 58%	Not recorded	Treatment for 7 days
Majewski (87)	2007	SIBO	20	GHBT	Rifaximin 800 mg/dy	54%	Diarrhea 86% Bloating and gas 83%	Treated for 4 weeks
Esposito (88)	2007	IBS (symptom-based)	73	LHBT or GHBT	Rifaximin 1200 mg/dy	53%	Significant reduction in symptoms	Treated for 7 days
Majewski (89)	2007	IBS	93	GHBT	Rifaximin 800 mg/dy	75%	Reduced symptom scores in 88%	Only 8 treated with rifaximin. Treated or 1 month
Lauritano (90)	2009	SIBO	142	GHBT	Rifaximin 1200 mg/dy Metronidazole 750 mg/dy	63% 44%	Not recorded	Treated for 7 days
Furnari (91)	2010	SIBO	77	GBHT	Rifaximin 1200 mg/dy Rifaximin + guar gum	62% 87%	87% 91%	Clinical improvement rates for those who were normalized. Treated for 7 days
Collins (28)	2011	SIBO – children with abdominal pain	75	LBHT	Rifaximin 1650 mg/dy Placebo	20%	No difference	Treated for 10 days
Rosania (92)	2013	SIBO	40	LBHT and GBHT	Rifaximin 400 mf/dy + L. casei Rifaximin 400 mg /dy + FOS		83% 67%	Response rates NS Evaluated at 6 months; treated for 14 days
Scarpellini (93)	2013	Children with IBS	50	LBHT	Rifaximin 600 mg/dy	64%	Symptoms improved	Treated for one week
Tahan (94)	2013	SIBO in children	20	LBHT	Trimethoprim-sulfamethoxazole and metronidazole	95%
Ghoshal (95)	2016	IBS (Rome III) with (+) and without (–) SIBO	80	Aspirate and GBHT	Norfloxacin 800 mg/dy Placebo	100% 0%	87.5% (+) 25% (–) 0% (+) 0% (–)	Treated for 10 days. Symptom responses at one month. Retested at 6 months – no difference
Tuteja (29)	2019	IBS (Rome III)	50	LHBT	Rifaximin 1100 mg/dy Placebo	7% 22%	No impact on symptoms	Gulf War veteran treated for 2 weeks.
Garcia-Collinot (96)	2020	Scleroderma with SIBO	40	HBT	S. Boulardii Metronidazole Both	33% 25% 55%	Diarrhea, bloating etc. improved in Boulardii and combined regimens only.	Treated for 2 weeks.
Zhuang (97)	2020	IBS-D (Rome IV)	78	LBHT	Rifaximin 800 mg	44%	58%	Treated for 2 weeks and followed for 10 weeks.
Kim (98)	2023	Functional dyspepsia	83	GBHT	Mosapride 15 mg/dy Rifaximin 1200 mg/dy Rifaximin + Mosapride	17% 22% 35%	Flatulence and chest discomfort. improved with rifaximin	Treated for 2 weeks

When one focuses specifically on SIBO in the context of IBS, the data shrinks further, and methodological problems persist.

In their original observational study, Pimentel and colleagues noted that 25 of 47 subjects achieved complete eradication of SIBO (using various antibiotics) and that improvements in the symptoms of diarrhea and abdominal pain (but not bloating) were significantly greater than among those who were not eradicated (25). In a subsequent double-blind, placebo-controlled trial, neomycin administered in a dose of 500 mg b.i.d. for 10 days resulted in a 35% improvement in a composite score for IBS symptoms while only an 11.4% improvement was seen in those randomized to placebo (10). In the neomycin-treated group, the improvement rate increased to 75% for those who normalized their breath test following therapy. As shown in Table 1, results since then have revealed variable correlations between breath tests and symptom responses to antibiotic therapy. In ensuing landmark studies showing antibiotic efficacy in IBS, breath testing was either not reported (31) or conducted in only a small segment of the population. In the TARGET phase III studies, which led to FDA approval of rifaximin for the treatment of non-constipated IBS (32), only 98 of the total study population of 1260 patients underwent an LHBT (33). Among responders in this small subset, 59.7% had a positive baseline breath test. While 48% of these patients were defined as overall responders to a 2-week course of rifaximin in a dose of 550 mg t.i.d., normalization of the breath test occurred in only 29%. Not surprisingly, the outcome of the post-treatment breath test was not predictive of response to rifaximin; 76.5% of those who normalized their breath test were deemed responders compared to 56% of those who did not. In summary, the relationships between antibiotic eradication of SIBO/normalization of breath tests and symptom responses are far from consistent or clear.

3). A proportion of IBS patients will have symptom improvement with antibiotic therapy.

The high response rate to antibiotics originally reported (10, 25) has failed to materialize in well-designed clinical trials. Rigorous studies show a therapeutic gain of ~10 % over placebo (32), a response that is similar or less than other therapies targeting neural rather than microbial signaling pathways. Possibly more concerning is that most of the benefit is related to improvement in bloating, a symptom that has not been shown to be predictive of SIBO (34). When this observation is combined with findings on fermentation responses in humans (35, 36), it seems plausible that much of the antibiotic effect could result from the suppression of fermenting bacteria in the colon and not in the small intestine as has been suggested. A colonic source for H₂ production is also suggested by LBT studies showing antibiotics suppressed H₂ production in healthy volunteers (i.e., without symptoms of SIBO) (37) and that H₂ production is markedly reduced in constipated patients following successful treatment of their constipation with PEG (38). Furthermore, Sharara and colleagues (34) reported a response rate to rifaximin that was similar to that seen in the phase III study but in subjects who all had a negative LHBT at baseline.

4). Meta-analyses show a higher proportion of IBS patients have positive breath tests than healthy control.

Multiple meta-analyses (39–41) have been conducted examining studies with breath testing and consistently reported an increased prevalence of a positive breath test in IBS patients compared to healthy controls. This difference has often been suggested by proponents of the SIBO hypothesis to demonstrate a cause and effect between SIBO and IBS patients. A Canadian IBS guideline identified 24 case series involving 2698 IBS patients and found that overall, 25% were GHBT positive (28), although rates varied between 4% (42) and 69% (43). This guideline also reported that 13 case control studies gave a pooled odds ratio of 6.29 (95% CI = 4.55 to 8.68) for being breath test positive in patients with IBS compared to healthy controls (28). Despite this strong association, the guideline did not support routine breath testing of patients with IBS (28). The group came to this decision as the main reasons for doing a breath test are to distinguish IBS from other GI diseases as well as helping to target therapy. As stated above, rifaximin is only modestly effective even in breath test-positive patients and is no more effective than other alternatives. Furthermore, there are many GI diseases associated with a positive breath test where there is no recommendation to conduct breath testing, such as dyspepsia (42, 44–47), gastroesophageal reflux disease (GERD) (48), celiac disease (49) and ulcerative colitis (50). Overall, there was no association between a positive breath test and IBS when disease controls were used rather than healthy controls (Figure 5), except in the case of GERD, where all GERD patients were on a PPI (48). Some patients in the dyspepsia studies were also on a PPI or H2R antagonist. This acid suppression may lead to SIBO, although others have not shown any association between PPI use and SIBO (51). The most likely explanation is that having a positive glucose H₂ breath test is a non-specific indicator of GI disease. Examples of such non-specific associations include motility abnormalities found in peptic ulcer disease (52), before the discovery of H. pylori, or the finding of duodenal eosinophilia in a variety of GI diseases, not just in functional dyspepsia (53). A systematic review of the literature suggests that a positive GHBT is not diagnostically useful for IBS, rather it is abnormal in several different GI diseases and may not help direct therapy.

Figure 5. Case-control studies comparing glucose H₂ breath tests in patients with IBS compared to other GI disease controls.

All of the GERD cases were on a PPI.

• vConflicting guidelines of major societies

The controversy in the field is clearly reflected in the conflicting recommendations on the importance of SIBO and the use of breath tests in different clinical settings, where it is obvious that the interpretation of the existing literature varies considerably among experts. In the North American consensus on hydrogen and methane-based breath testing in gastrointestinal disorders, the experts state that breath testing is a useful, inexpensive, simple, and safe test in the evaluation of common gastroenterology problems. They recommend that a rise in H₂ of ≥20 PPM by 90 min during glucose or lactulose breath testing for SIBO can be considered positive (54). This recommendation is then included in the ACG clinical guideline on SIBO, despite acknowledging that currently employed breath tests have low sensitivity and specificity, that additional validation studies are needed for standardization, and that the evidence base is very low (55). The European guidelines on the use of H₂ and CH₄ breath tests come to somewhat different conclusions and recommend that hydrogenbreath testing can be used until a true gold standard is established, but acknowledges the problems with false positives, in particular with lactulose and, therefore, recommend glucose rather than lactulose, and, if possible, with the addition of simultaneous scintigraphy to reduce the risk of false positives (56). In line with this, the Asian-Pacific guidelines highlight the problems with the tests and recommend glucose and not lactulose as the substrate for the test, and state that the early peak criterion for the lactulose hydrogen breath test, as suggested by the North American consensus, cannot be used given that orocecal transit time with lactulose is frequently shorter than 90 min (57). In the AGA clinical practice update, the authors comment on the inconsistency with which breath tests are performed and interpreted, as well as the wide discrepancies in thresholds for defining a positive breath test result. Based on these issues, the authors do not provide any specific practice guidance on the use of breath tests (58).

The role of SIBO in IBS remains controversial, and a recent systematic review and meta-analysis of case-control studies concluded that the literature suggests a link between SIBO and IBS but that the overall quality of the evidence is low (39). In guidelines, the recommendations regarding the use of breath tests in the diagnostic work-up of patients with IBS differ, with the British and Canadian guidelines clearly advocating against the use of breath tests in IBS (59, 60), whereas recent American guidelines do not include a recommendation either for or against the use of breath tests in IBS (61, 62).

To summarise, the different views on the relevance of SIBO in large groups of patients in gastroenterology and the usefulness of breath testing are clearly reflected in the divergent recommendations in different guidelines and consensus statements.

Part II. A potential role for dynamic microbial communities in the small intestine

Mechanistic studies of microbe-host interactions in the small intestine are needed to better understand a potential role of the microbiome in the generation of IBS symptoms. While much more study is necessary before clinical applications can occur, a number of important findings are emerging that will help to guide these studies.

1. Factors shaping the small intestinal microbiome

   The small intestinal microbiome is largely shaped by diet and the unique physicochemical properties of the luminal environment, which are designed to facilitate digestion and absorption of nutrients. This includes rapid transit despite being one of the longest organs in the body (average length 690.1+−93.7cm (63); measurements from surgical specimens), (median transit time between 196 and 287 minutes; varies depending on methodology and study cohort, e.g., magnetic trackers and scintigraphy) (64–69), facilitated by peristalsis and pancreatic/intestinal secretions, high concentrations of bile which is bacteriostatic, and specialized immune cells (Paneth cells which release antimicrobial peptides, intra-epithelial and lamina propria lymphocytes and IgA producing plasma cells) which conspire to exclude the vast majority of microbes. The low microbial density and diversity in the proximal small intestine restricts carbohydrate fermentation and bile deconjugation by bacteria, allowing for optimal fat and carbohydrate digestion and absorption. The increase in microbial density (increasing from approximately 10^3–5 CFU/mL to 10^7–8 CFU/mL)(27, 69) and diversity (increase in proportion of gram-positive and anaerobic bacteria) towards the distal small intestine is favored by a steady increase in pH (~5.7 −6.4 in the proximal to 7.3–7.7 in the distal small intestine) (70) and bile acid metabolism, along with a decrease in oxygen concentration (71). The few available studies characterizing the small intestinal microbial composition show Actinobacteria and Proteobacteria to be the dominant bacterial groups in the duodenum. In the jejunum, higher levels of Firmicutes are found, accompanied by Proteobacteria, Actinobacteria, and Bacteroidetes, while the ileal microbiota is dominated by Bacteroides, Clostridium, Enterobacteria, Enterococcus, Lactobacillus, and Veillonella (72) A recent study used capsules programmed to collect samples at different times following ingestion to characterize the small intestinal microbiome (73). They found higher levels of Bilophila wadsworthia, the genera Escherichia/Shigella, and Enterococcus within Proteobacteria, Bacteroides within Bacteroidetes and Romboutsia within Firmicutes in the small intestine when compared to stool. Interestingly, there was significant spatial and temporal variability along the intestine, with each sample being dominated often by a single bacterial strain (~40%)(73). There is greater intra- and inter-individual variability in the small intestine compared to the colon, likely due to the dynamic nutrient landscape in the small intestine. In addition to the luminal environment, diet plays an important role in shaping the small intestinal microbiome. A high-fat diet favors the expansion of Clostridia and a concomitant decrease in Bifidobacteria and Bacteroides (74). This microbial profile is associated with increased lipid absorption and increased expression of genes involved in small intestine epithelial lipid transport (74). These effects vary according to dietary fat composition; a diet high in milk-fat favored expansion of bile-tolerant Bilophila wadsworthia, which was associated with the development of colitis in genetically susceptible mice (75). Dietary carbohydrates also affect small intestinal microbial composition and function; unlike the colon, the small intestinal microbiome shows a significantly higher capacity to metabolize simple sugars (increased expression of sugar phosphotransferase systems, pentose phosphate pathway, lactate and propionate fermentation) (76). Thus, the small intestine maintains a fragile equilibrium, wherein the interplay of diet and the luminal environment shapes microbial composition and function. This, in turn, impacts digestion, absorption, and the functioning of the innate and adaptive mucosal immune system.

2. What’s important - small intestinal microbial density or composition or both?

   A change in the physicochemical environment in the small intestine may allow for increased growth of all or some bacteria in the small intestine. Indeed, diarrhea in patients with altered anatomy predisposing them to slower small intestinal transit was attributed to increased microbial density in the proximal small intestine (SIBO), resulting in maldigestion and malabsorption. Since its initial description, additional conditions affecting transit, such as primary or secondary small intestinal dysmotility, increase in pH due to consumption of proton pump inhibitors or achlorhydria, altered bile secretion in cirrhosis, or loss of antibacterial defences, such as in immune deficiency states, have been recognized as factors that alter the small intestinal luminal environment allowing for increased growth of bacteria in the proximal small intestine (69). The increased microbial density in the small intestine is thought to result in fat malabsorption (including fat-soluble vitamins) and increased intestinal secretion due to early deconjugation of bile acids by bacteria, accelerated transit driven by metabolic products from excessive microbial fermentation of carbohydrates and anemia due to increased vitamin B₁₂ utilization by bacteria (77). While these mechanisms are generally attributed to increased bacterial density, they represent distinct microbial functions and hence can also be seen with changes in microbial composition without increased bacterial density. This fundamental concept brings forth an important consideration regarding the different risk factors mentioned above, as each causes a change in different aspects of the luminal milieu (i.e., transit, bile secretion, pH, etc.); hence, they may in fact have distinct effects on small intestinal microbial density and composition. It, however, remains to be seen whether small intestinal bacteria play a role in the generation of these symptoms and, if so, whether increased microbial density or a change in the microbial composition and function or both is responsible.

   One would have expected that with the advent of next-generation sequencing technologies and improved culture methods, the focus would have shifted from increased microbial density towards identifying specific changes in small intestinal microbial composition and function that could explain the different symptom complexes. Instead, considerable effort was dedicated to developing non-invasive surrogates to identify increased bacterial growth in the small intestine (e.g., breath testing). The ease of breath testing led to these tests being used in patients with a wide array of GI symptoms even when none of the typical risk factors for SIBO were present. This is especially important given that small intestinal bacterial density varies among individuals, and a high proportion of healthy asymptomatic individuals have been found to have levels defined as SIBO both by breath tests and culture of small intestinal aspirates (41). While this is not surprising, as several factors, including diet, can influence the small intestinal microbiome without causing symptoms, it brings into question the specificity of testing for microbial density in the absence of risk factors and/or typical symptoms. In a small study of 16 patients, nearly 50% of patients on a habitual high-fiber diet were noted to have SIBO in the absence of GI symptoms, but they developed symptoms of constipation and bloating when the fiber intake was significantly decreased, despite a significant decrease in fermentation end products such as acetate (23). While this small study needs to be replicated in larger cohorts, the findings would argue against increased microbial fermentation by small intestinal bacteria as the determinant of symptoms such as bloating and nausea.

   Only recently have we started appreciating the relevance of small intestinal microbial composition and function and to shift our attention away from absolute numbers of bacteria. In a study of 126 patients (23) with GI symptoms including diarrhea, bloating and abdominal pain, the microbiome-based symptom index but not the presence of SIBO differentiated symptomatic patients from healthy individuals. A subset of symptomatic patients harbored higher levels of genera Erwinia, Escherichia, Bifidobacterium, and Lactobacillus, among others, in duodenal aspirates. There was no significant difference in microbial composition among those with and without SIBO. This is in contrast to another recent study (78) in 98 SIBO-positive and 385 SIBO-negative patients, where they define SIBO by culture and high through-put sequencing, authors found microbial α-diversity progressively decreased, the relative abundance of Escherichia/Shigella and Klebsiella increased, and microbial network connectivity decreased in subjects with ≥10³ but <10⁵ CFU/mL and ≥10⁵ CFU/mL in duodenal aspirates. The differences in findings are largely explained by the differences in criteria for SIBO; the former study used >10⁵ CFU/ml in aerobic/anaerobic rich nutrient media, while the latter study used >10³ CFU/ml in MacConkey media (selective for gram-negative bacteria like Enterobacteriaceae including Escherichia/Shigella and Klebsiella)(54). The latter approach shows that culture-based methods can be useful beyond enumerating bacterial numbers to identify subsets of symptomatic patients with higher levels of specific bacterial families, which may allow for targeted therapies. Despite different methodologies, the increasing focus on characterizing the small intestinal microbiome is a welcome change and will help address an important gap in the field.

3. Small intestinal microbiome-driven mechanisms cannot be extrapolated from those in the colon.

   Most studies have focused on the colon as the transducer of the microbial luminal signals driving GI symptoms, given the large microbiome presence in the colon, a relatively broad understanding of the structure and function of sensory innervation of the colon, and ease of access for testing and sampling bacteria. Short-chain fatty acids such as butyrate are important in promoting serotonergic signaling, maintaining the epithelial barrier, and increasing colonic contractility. For example, tryptamine (produced by decarboxylation of tryptophan) activates serotonin receptor −4 (5-HT4) and increases fluid secretion, while histamine, isovalerate and GABA produced by bacteria can modulate visceral sensation by activating mast cells, enterochromaffin cells (EC cells), or sensory neurons, respectively (79). In addition to producing bioactive molecules, bacteria also play an important role in inhibiting host proteases and deconjugating and decarboxylating host bile acids (79). While there is ample evidence that these mechanisms may play a role in some patients with DGBI, the same metabolites and bacterial functions are also seen in the small intestine, but their physiologic effects have not yet been characterized. This is not only due to a lack of adequate characterization of the small intestinal microbiome but also because the epithelial and neuroepithelial circuits involved in the transduction of luminal signals, as well as the sensory innervation of the small intestine, remain a black box. The small intestine is unique regarding its epithelium, which features the presence of specialized immune cells and a thinner mucus layer. Cells such as EC cells, which serve as sensory and mechanical transducers of luminal signals, while present in both the small intestine and the colon, are still distinct in the two locations. Thus, specific studies investigating the role of the small intestine are warranted, and we cannot simply extrapolate from the colon. This shift to focus on the small intestine is underway, and initial findings highlight the promise of this approach.

CONCLUSIONS

This narrative review outlines compelling evidence that breath testing to diagnose SIBO in patients with IBS and related disorders is inaccurate and should be abandoned as a diagnostic test for this purpose. This message is becoming increasingly time-sensitive, as breath testing is now direct to consumers through industry-sponsored testing in the US and many other countries and no longer requires a physician to order or interpret them. Most recently, home testing devices are being promoted to monitor gas production during and after a meal, for which there is no validation of the test results. As the SIBO-IBS hypothesis continues to be promoted on social media, the number of tests may even increase. This is very concerning as the high number of false positive tests and results which have no clinical foundation can have harmful consequences for our patients. Most importantly, it leads to a SIBO diagnosis for which evidence is lacking, often creating confusion, anxiety and potential loss of trust in the healthcare system. The practical consequences of a positive tests include that it typically leads to one or more courses of potentially harmful antibiotics. It is also important to recognize that mis-diagnosis places a considerable financial burden on the patient (e.g., breath testing can cost up to $300 US and a single course of antibiotics over $1000.00 US; many patients undergo repeat testing and courses of antibiotics).

The challenges of applying the concept of SIBO to DGBI should not undermine confidence in the diagnosis of SIBO in “classical” conditions associated with gastrointestinal dysmotility, such as scleroderma, intestinal stasis secondary small bowel surgery and resection of the ileal cecal valve, with associated signs of malabsorption. In this setting the pre-test probability of the GBT is higher and would increase its diagnostic accuracy. Whether one choses to treat directly with antibiotics or first perform the breath test to guide therapy will depend on a number of factors, including test availability, cost, and patient and physician preference.

This review also does not refute that small and large bowel microbiota could generate IBS-like symptoms in some patients. Rather, it highlights the importance of focusing on the complexities of communities of bacteria and their constant metabolic response to the host, especially in response to diet and related digestive factors such as gastric acid and bile salts, and not just simply measuring absolute numbers of bacteria. The emerging technology to sample the small intestine and colon non-invasively with ingestible and retrievable capsules and the application of high throughput molecular techniques on samples show considerable promise for unraveling this complex field and the opportunity to personalize therapy for affected patients. Future human studies should continue to strive to control for confounders in this complex ecosystem, apply integrative biostatistics in a rigorous fashion that can be replicated by others, and design mechanistic studies to establish causal links for correlative factors (80).

DATA AVAILABILITY STATEMENT

No new data was generated for this review article.