Bishop, Warren P.
Section Editor(s): Bijleveld, Charles M.D.; Buller, Hans M.D.; Taminiau, Jan M.D.; Thomas, Adrian M.D.; Lichtman, Steven N. M.D.; Shneider, Benjamin L. M.D.; Bishop, Warren P. M.D.; Haber, Barbara Anne M.D.
Department of Pediatrics; University of Iowa; Iowa City, Iowa, U.S.A.
The lactulose breath hydrogen test and small intestinal bacterial overgrowth. Riordan SM, McIver CJ, Walker BM, Duncombe VM, Bolin TD, Thomas, MC. Am J Gastroenterol 1996;91:1795-1803.
Riordan et al. carefully and critically examined the use of lactulose breath hydrogen testing (LBHT) for small bowel bacterial overgrowth syndrome (SBBO) in a recent article. The authors studied 28 adult subjects, 14 of whom had various underlying conditions predisposing to SBBO. The patients were all experiencing gastrointestinal symptoms. Using a combination of scintigraphy and breath-hydrogen analysis, the authors correlated LBHT measurements with the timing of substrate delivery to the cecum. The authors also tested the hypothesis that SBBO causes a double peak in breath hydrogen after lactulose ingestion: the first is caused by abnormal fermentation in the small bowel, and the second by normal colonic fermentation. The results indicate that this “double peak” criterion for SBBO lacks both sensitivity and specificity.
All subjects received the “gold standard” aerobic and anaerobic culture of an aseptically retrieved small bowel aspirate, the results of which were used to calculate the sensitivity and specificity of the breath-hydrogen results. Small bowel fluid was obtained by endoscopy, using N2 for insufflation to avoid oxygen-mediated killing of anaerobes. Both aerobic and anaerobic cultures were obtained via a sterile, sheathed catheter. Small bowel bacterial overgrowth syndrome was considered to be present when total aerobic or anaerobic counts exceeded 105 colony forming U/ml. Breath-hydrogen testing was performed in fasting subjects. 99mTc-sulfur colloid was administered along with 10 g of lactulose in a 10% solution. End-expiratory breath samples and scintigrapic images were obtained every 10 minutes for 3 hours or until arrival of the radioactive solution at the cecum. A “double peak” (positive) breath hydrogen test was defined as two consecutive measurements at least 10 ppm above baseline that occurred either (a) separately from a later >20 ppm “colonic” peak or (b) commencing within 20 minutes of ingestion and 15 minutes before a recognizable second peak. The correlation of breath hydrogen levels with delivery of the radioactive marker to the colon was also studied. This criterion for SBBO was considered positive if breath-hydrogen concentration rose by at least 10 ppm within 10 or more minutes before the arrival of the isotope in the cecum.
The results indicated that 18 patients had SBBO by bacteriologic criteria. The LBHT identified six patients with the classic double peak pattern, but only three of these six had SBBO. Thus, the sensitivity was only 16% and the specificity was 70%. When scintigraphic oral-cecal transit time was used to sort out the cause of the observed breath-hydrogen peaks, the sensitivity improved to 38.9% and the specificity increased to 100%. Poor sensitivity of the LBHT was not caused by the inability of the small bowel flora to ferment lactulose because all isolates examined exhibited lactulose fermentation in vitro; nor was it caused by unfavorable luminal pH for hydrogen production because no differences in aspirate pH were observed between subjects. The authors concluded that that the LBHT cannot substitute for culture of small bowel aspirates in the diagnosis of SBBO.
Comment: Gastroenterologists have been waiting breathlessly for a reliable noninvasive test for this condition, and will need to wait longer. A variety of breath test substrates have been championed over the years, including lactulose, glucose, 14C-D-xylose, 14C-bile acids, and even rice flour pancakes. The measurement of a fasting breath-hydrogen level (which may be increased in SBBO) has also been studied. Each of these methods has been found to have serious problems with sensitivity and/or specificity when examined more closely. The literature is inconsistent on the use of breath testing for bacterial overgrowth. This well-performed study serves to reinforce suspicions that breath tests are not generally reliable diagnostic tools for SBBO. However, this study only examined one fermentable substrate (lactulose). Can a reliable breath test be devised?
When ordering a breath test, the first decision a clinician must make is which substrate to use. An ideal substrate for small bowel overgrowth determination is fermentable by all potential organisms, is readily absorbed before reaching the colon, and is nontoxic. A useful substrate must also yield an easily measured volatile fermentation product. Nonabsorbable substrates such as lactulose are inherently problematic because they always reach the colon. Colonic fermentation must then be distinguished from SBBO-related bacterial action. As this article proves, such discrimination is not an easy task. The use of absorbable substances (e.g., glucose, xylose, and bile acids) avoids this problem and can therefore (at least in theory) yield a simple positive or negative result.
Of the many potential absorbable substrates, glucose and 14C-D-xylose have been most extensively evaluated as diagnostic agents for SBBO. The hydrogen produced from fermentation of glucose or the 14CO2 produced from xylose can then be measured in the breath. Glucose appeared helpful in at least one recent study (Gastroenterology 1988;95:982-8), with a sensitivity of 93%, a specificity of 78%, and positive and negative predictive values in the high 80s. Another study yielded somewhat lower sensitivity (62%), but similar specificity (Gastroenterology 1990;98:302-309). Glucose clearly appears to be a better choice than lactulose for the diagnosis of SBBO.
The 14C-D-xylose test also seemed promising early on, exhibiting sensitivity and specificity approaching 90% in initial studies (Gastroenterology 1979;77:75-82, Gastroenterology 1986;91:1447-1451). However, several recent studies have reported lower sensitivities, as low as 40% (Gastroenterology 1994:106:615-623, Am J Gastroenterol 1995;90:1455-1460, Scand J Gastroenterol 1993; 28:963-968). An additional problem with D-xylose and all CO2 based tests is the possibility that normal metabolism of absorbed substrates can lead to 14CO2 production and a false positive result (Am J Gastroenterol 1995;90:1455-1460). Even if the 14C-D-xylose breath test was acceptably sensitive and specific, the use of 14C-containing isotopes in children is problematic, because of its long half-life and wide tissue distribution. These concerns can be circumvented by use of the stable isotope, 13C, but stable isotope analysis requires the use of mass spectroscopy. 13C- or 14C-D-xylose breath tests are therefore not demonstrably better than the LBHT for diagnosis of SBBO.
What are the causes of inadequate sensitivity and specificity in breath tests for SBBO? Poor sensitivity is the result of inadequate fermentation by the resident flora. Failure of substrate fermentation can be caused by excessively rapid absorption of the substrate in the proximal gut, presence of bacteria which cannot ferment the substrate (J Pediatr Gastroenterol Nutr 1992;14:192-197), loss of bacterial flora because of recent antibiotic or laxative use, or existence of an acidic environment which inhibits hydrogen production. In contrast, poor specificity is the result of breath hydrogen production from sources other than small bowel bacteria, usually the colon. A fundamental problem in all breath-hydrogen analyses for SBBO is the frequent coexistence of abnormal small intestinal absorptive function, either caused by the SBBO itself or by an underlying illness such as short bowel syndrome. Specificity can also be impaired by excessive substrate dosage, which causes osmotic fluid shifts, shortened transit time, and therefore delivery of substrate to the colon.
Quantitative culture of small bowel aspirates must remain the gold standard when documentation of SBBO is essential. Although many gastroenterologists obtain cultures for SBBO during endoscopy, it is questionable how many have the appropriate equipment and laboratory support to do it optimally. Proper collection and specimen analysis techniques (use of sheathed sterile catheters for specimen collection, anaerobic collection with nitrogen gas insufflation of the bowel, and rapid plating under both aerobic and anaerobic conditions) are not commonly used by most centers and are certainly both labor-intensive and expensive. Trials of a less invasive collection technique, the string test, have shown it to be inaccurate (J Tropical Med Hygiene 1995; 98:117-120, Am J Gastroenterol 1993; 88:925-928).
In the final analysis, clinicians will probably best be served by an empirical trial of enteral antibiotic therapy when the clinical situation favors the possibility of SBBO. Because of its invasive nature and expense, quantitative culture may be best used in special situations when confirmation is required and when specimens can be properly collected and analyzed. When will a sensitive and reliable noninvasive test be available for the diagnosis of SBBO? I wouldn't hold my breath!
Warren P. Bishop
Department of Pediatrics; University of Iowa; Iowa City, Iowa, U.S.A.