Article Text

Download PDFPDF

Small intestinal bacterial overgrowth, intestinal permeability, and non-alcoholic steatohepatitis
  1. S M Riordan1,
  2. V M Duncombe1,
  3. M C Thomas1,
  4. A Nagree1,
  5. T D Bolin1,
  6. C J McIver2,
  7. R Williams3
  1. 1Gastrointestinal and Liver Unit, Prince of Wales Hospital, Sydney, Australia
  2. 2Department of Microbiology, Prince of Wales Hospital, Sydney, Australia
  3. 3Institute of Hepatology, University College London Medical School and Hospitals, London, UK
  1. Associate Professor S M Riordan, Gastrointestinal and Liver Unit, Prince of Wales Hospital, Barker Street, Randwick, 2031, NSW, Australia.
  1. A J Wigg4,
  2. A G Cummins5
  1. 4Department of Gastroenterology and Hepatology, Flinders Medical Center, Bedford Park, Adelaide, 5042, South Australia, Australia
  2. 5Department of Gastroenterology, Queen Elizabeth Hospital, 28 Woodville Rd, Woodville South, Adelaide, 5011, South Australaia, Australia

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

In a recent issue, Wigg and colleagues (Gut 2001;48:206–11) reported that small intestinal bacterial overgrowth (SIBO), as diagnosed by a combined 14C-d-xylose/lactulose breath test, is significantly more common in patients with non-alcoholic steatohepatitis (NASH) than in control subjects without liver disease. The authors investigated the possible pathogenic significance of this observation by examining whether intestinal permeability and circulating levels of endotoxin and tumour necrosis factor α are increased in NASH patients with SIBO compared with those without. No significant differences in any of these parameters could be demonstrated in the two groups.

An important factor influencing the validity or otherwise of these findings is the diagnostic accuracy of the 14C-d-xylose and lactulose breath tests for SIBO. Our experience, using a sterile endoscopic technique to sample small intestinal secretions under direct vision, is that these breath tests lack sensitivity and specificity for culture proven SIBO.1,2 Endogenous CO2 production and colonic metabolism of d-xylose are important factors inherently limiting the accuracy of the 14C-d-xylose breath test for SIBO.1 Furthermore, reliance on the finding of “double peaks” in serial breath hydrogen or methane levels after ingestion of lactulose to improve the accuracy of the 14C-d-xylose breath test, or as a diagnostic marker in its own right, is problematic. In a study in which a scintigraphic tracer was administered concurrently with lactulose, we found that each of the double peaks in breath hydrogen values may occur after the arrival of the test meal at the caecum, paralleling delivery patterns of fermentable substrate to caecal bacteria. A caecal source of each peak was suggested on 50% of occasions, rather than the first peak necessarily reflecting small intestinal metabolism of lactulose by overgrowth flora as purported. Conversely, a single rise in breath hydrogen levels commencing before the test meal reached the caecum was evident in 22% of subjects with culture proven SIBO. Thus both false positive and false negative diagnoses of SIBO may result.2 Indeed, as pointed out in the accompanying commentary (Gut 2001;48:148–9), the prevalence of SIBO, as diagnosed by breath testing, in control subjects in Wigg et al's study seems remarkably high.

Rather than seeking to establish the prevalence of SIBO in patients with NASH, as in the study of Wigg et al, we have investigated the prevalence of liver damage, as reflected by elevated liver enzyme levels in serum, in patients with culture proven SIBO.3 Biochemical evidence of liver injury was found in 0/11 patients with SIBO with salivary type bacteria only, 0/21 patients with SIBO with facultative anaerobic (Enterobacteriaceae) but not obligate anaerobic (Bacteroides spp) colonic type bacteria, and 1/8 patients with SIBO including Bacteroides spp. Alkaline phosphatase and gamma glutamyl transferase levels were elevated in this patient, although liver ultrasonography and cholangiography revealed no abnormality. Small intestinal permeability was increased and, together with liver enzyme abnormalities, normalised following eradication of SIBO with a metronidazole based antibiotic regimen. We concluded that liver injury, reversible with antibiotic treatment, occurs uncommonly in patients with SIBO, and only when the overgrowth flora includes obligate anaerobes such as Bacteroides spp, in keeping with earlier findings implicating such flora in the pathogenesis of liver injury associated with experimental SIBO in rodents.4 Liver injury associated with SIBO with Bacteroides spp was not a necessary consequence of increased small intestinal permeability, which was also evident in 50% of patients with SIBO with Bacteroides spp who had no evidence of liver damage.

Based on these observations, we suggest that future studies examining the prevalence of SIBO in patients with NASH and its possible pathogenic significance should use culture of small intestinal aspirate rather than breath testing as the diagnostic modality and focus on the presence or absence of overgrowth with obligate anaerobic flora such as Bacteroides spp. Such an approach would be preferable to simply assessing for any improvement in NASH following a therapeutic trial of metronidazole, as SIBO with Bacteroides spp and other obligate anaerobes is not always eradicated by a single course of antibiotic treatment. As small intestinal colonisation with Bacteroides spp depends on underlying intestinal dysmotility,5 factors other than SIBO are likely responsible for NASH in patients in whom intestinal motility is normal.


Authors' reply

We thank Riordan et al for their own observations concerning the diagnosis of small intestinal bacterial overgrowth (SIBO) and liver injury associated with SIBO.

We agree with the suggestion that Bacteroides may be more important than other bacterial species in causing liver injury. This may be an explanation for our failure to detect endotoxin elevations in those patients diagnosed with both NASH and SIBO (endotoxin is derived only from Escherichia coli bacteria and not Bacteroides spp) (Gut 2001;48:206–11).

We must however correct their statement that no significant differences were found in our study between NASH patients and control subjects for tumour necrosis factor α. A statistically significant difference was found between these groups (p<0.001) (Gut 2001;48:206–11).

Their comments highlight the longstanding difficulty in gastroenterology of diagnosing SIBO. Although the traditional 14C-d-xylose breath test has been associated with a high sensitivity in some studies,1,2 the specificity of this test is unacceptable in our experience. The high false positive rate associated with this breath test probably relates to catabolism of unabsorbed 14C-d-xylose by the colon, resulting in 14CO2 expiration. In an attempt to retain sensitivity and improve specificity, we have developed a combined 14C-d-xylose-lactulose breath test. Lactulose, which is not absorbed and requires large bacterial concentrations for its catabolism to H2 and CH4, acts as an internal transit marker of colonic metabolism. Smaller bacterial concentrations only are required for the catabolism of 14C-d-xylose to 14CO2. Thus in SIBO, catabolism of 14C-d-xylose by small intestinal bacteria will result in an early 14CO2 peak prior to the colonic H2 and CH4 peaks. Specificity is improved because 14CO2 peaks due to colonic metabolism of unabsorbed 14C-d-xylose can be identified when they rise simultaneously with H2 and CH4 colonic peaks.

When both the 14C-d-xylose breath test and the combined breath test were done in a group of 11 patients, only four had positive combined tests compared with nine positive 14C-d-xylose breath tests (Gut 2001;48:206–11). This suggests that the combined test has achieved a greater specificity. We feel that the combined 14C-d-xylose-lactulose breath test is a sensitive and specific non-invasive alternative to culture of small intestinal aspirates.

We note the concern of Riordan et al on the use of double H2 peaks for the diagnosis of SIBO, based on their observations with scintigraphic studies. In very severe SIBO, a double peak of H2 and CH4 may be produced due to lactulose catabolism by bacteria in both the small intestine and colon. As suggested by Riordan et al, double peaks may reflect catabolism of lactulose by colonic bacteria rather than by bacteria in the small intestine and then the colon. Diagnosis of SIBO in our study was based on early 14CO2 expiration before the appearance of a H2 or CH4 peak in all cases. Double H2 or CH4 peaks were observed in only one of the 16 breath tests recorded as positive in our study. In this patient, significant 14CO2 was expired prior to the first peak.

The studies quoted by Riordan et al have used cultures of small intestinal aspirates as the gold standard for the diagnosis of SIBO. Is aspiration of intestinal secretions under sterile conditions a satisfactory gold standard for the diagnosis of SIBO? This diagnostic method is not universally accepted. It is likely that the small volume of proximal intestinal contents aspirated does not accurately represent the bacterial flora of the entire small intestine. This may explain the problems with sensitivity and reproducibility described by some investigators.1,3,4 Lack of standardisation of specimen collection and the invasive nature of the test are further problems, particularly in the setting of studies involving a healthy control population. The use of culture of small intestinal aspirates as a gold standard to assess the performance of breath tests may therefore not be valid.

In view of the difficulties associated with diagnosing SIBO, the association of SIBO with NASH found in our study requires confirmation by other investigators. Studies using culture of small intestinal aspirate to diagnose SIBO, which can also provide qualitative bacterial information, will be complimentary to our study using a combined 14C-d-xylose-lactulose breath test.