There is a complex interplay between the gastrointestinal (GI) tract and the bacterial flora it contains. In health, commensal bacteria in the GI tract participate in the digestion of nutrients, boost host immunity, and help control the growth of potentially pathogenic organisms. This equilibrium can be disrupted and once-beneficial bacteria become pathogens. Small bowel bacterial overgrowth (SBBO) is a condition of intestinal bacterial disequilibrium. SBBO is defined as a high bacterial burden in the small intestine (duodenum, jejunum, and/or ileum) so that samples of intestinal fluid for culture yield >106 colony-forming units/1 mL (1). Symptoms of SBBO can be subtle and include diarrhea, abdominal pain, bloating, and flatulence. The morbidities associated with SBBO include anemia, malabsorption, and malnutrition. Certain conditions predispose the intestine to SBBO: stasis from slow motility, a blind loop or a dilated loop of intestine. Some immune deficiencies weaken the ability of the intestine to control bacterial growth (2). In addition, increased bacterial load from gastric bacterial overgrowth may overwhelm intestinal defenses, as seen in gastric achlorhydria and acid blockade by medications (3–5).
Patients with cystic fibrosis (CF) have defects in the CF transmembrane regulator (CFTR) gene. This results in dehydrated and inadequately alkalinized mucus membrane secretions in the lungs, biliary tract and intestines. Patients with CF who have severe GI disease produce thickened, acidified pancreatic secretions. They develop progressive fibrosis in the pancreas, pancreatic exocrine insufficiency, and often diabetes mellitus. Abnormal biliary secretions and associated progressive liver damage result in multilobular cirrhosis in some patients (6). In addition, the intestine also is profoundly affected in CF. Intestinal motility is slowed, and constipation is common (7,8). Obstruction may occur secondary to inspissated intestinal secretions that take the form of meconium ileus in neonates and distal intestinal obstruction syndrome in children and adults (9). The use of medications to suppress gastric acid is common in patients with CF to treat symptoms of gastroesophageal reflux or to potentiate the action of supplemental pancreatic enzymes. Therefore, patients with CF have multiple risk factors that predispose them to SBBO. Awareness of this possibility by physicians could lead to earlier diagnosis, and CF patients could benefit from both prevention and effective treatment of SBBO.
Supporting evidence comes from a recent study in CF knockout mice that found a >40-fold increase in bacterial 16S genomic DNA in the small intestine of the CFTR knockout mice compared with wild-type mice (10). Prior studies in adults and children with CF also have suggested a high prevalence of SBBO (11,12). This study explores the prevalence of SBBO in CF patients, the risk factors for SBBO, and the correlation of SBBO with GI symptoms.
PATIENTS AND METHODS
Twenty-five subjects were recruited from the CF Center of the Lucile Packard Children's Hospital at Stanford. The center follows 232 patients with CF determined by a positive sweat test and/or genotype analysis. Exclusion criteria left a pool of 139 eligible patients. Patients with insulin-dependent diabetes mellitus and other non–CF-related conditions likely to influence GI motility and those taking narcotics were excluded. Patients who had received solid organ transplants also were excluded. No subject had taken daily oral or intravenous antibiotics for 30 d before the study. However, participants were not excluded if their usual regimen included inhaled tobramycin or azithromycin taken 3 times per week. No medications were changed or added in the preceding 30 d before the study procedure was performed. Of the 55 approached in regards to the study, 25 completed the study.
All of the recruited patients were pancreatic insufficient and were prescribed daily pancreatic enzymes. The subjects' medical records were reviewed; data were taken from the most recent CF clinic visit and the most recent laboratory tests. The collected data included age, gender, current medications, pulmonary function, past medical history, genotype, complete blood count, chemistry panel, glycosylated hemoglobin, and sputum culture.
Twenty-five breath tests (BTs) from age-matched GI patients were selected for comparison. These adults and children were referred to our pediatric GI clinic for BTs for suspected SBBO. Consecutive age-matched studies performed during the previous ≈2 y were selected.
Study and control patients underwent a hydrogen BT after an overnight fast. Breath samples were collected at baseline (fasting). Patients then drank a 20% dextrose solution dosed at 2 g/kg to a maximum dose of 80 g. If subjects chose to combine the BT with an oral glucose tolerance test, then a lower dose of 1.5 g/kg to a maximum of 50 g dextrose was given. Samples were collected at 0, 5, 10, 15, 20, 25, 30, 40, 50, 60, 75, 90, 105, 120, and 150 min for a total of 15 samples. The breath samples were collected with a 20-mL syringe and 3-way stopcock attached to a nasal prong as previously described (13). Samples of exhaled breath were collected during quiet respiration, ≈5 mL per breath for a total of 4 to 5 breaths per syringe. Two syringes were taken at each collection point; 1 sample from each time point was analyzed. If a sample differed from the observed trend, then the second syringe from that time point was analyzed, and the higher value was used. The samples were analyzed with the QuinTron MicroLyser DP (QuinTron, Milwaukee, WI) for hydrogen and methane. The machine has sensitivity to 1 ppm hydrogen (H2) and methane (CH4), an accuracy of ±2 ppm, and a linear range of 2 to 200 ppm H2 and CH4 and requires a 15-mL sample size. Twelve of the 25 control samples were analyzed on an earlier gas chromatographer, and hydrogen only was analyzed. Criteria for a positive BT for SBBO included a fasting hydrogen ≥15 ppm (14), a rise of ≥10 ppm hydrogen over the baseline sample at any time during the test (13,15,16), or a doubling of baseline methane excretion at any time during the test (17). Methane excretors were defined by a methane level of >2 ppm in any sample.
Each study subject and/or parent filled out a 1-page quality-of-life questionnaire during the BT. The questionnaire was modified from the Cystic Fibrosis Questionnaire to focus on GI issues (18). The modified questionnaire has 3 dimensions: vitality, digestion, and ability to eat. The questionnaires are constructed so that a higher score equals a higher quality of life. Three versions of our Cystic Fibrosis Questionnaire were offered: a teen/adult version (≥14 y old) filled out by the adolescent/adult; a child version (12–13 y old) filled out by the child; and a parent/child version (6–13 y old) filled out by the parent.
The Stanford University Internal Review Board Panel on Medical Human Subjects approved this study. All of the subjects or their legal guardians gave informed consent, and children 7 to 17 y old gave their assent.
The prevalence data were analyzed for statistical significance with χ2 analysis or Fisher exact test performed with NCSS (Hintze J. Number Crunching Statistical Software, Kaysville, UT, 2001). Paired Student t tests were used in Microsoft Excel 2000 (Microsoft Corp, Redmond, WA) to analyze continuous variables. The level of significance was set at P < 0.05. All tests were 2-tailed. This was a pilot study; thus, no sample size calculations were performed given the lack of existing data.
Of the study patients recruited, 60% (15 of 25) were male. Ages ranged from 6 to 46 y (mean, 17 y) (Fig. 1). Sixty percent (15 of 25) were homozygous for the CFTR ΔF508 mutation. Sixty percent (15 of 25) had normal lung function defined by a forced expiratory volume in 1 second (FEV1) of >80% predicted. Of the control patients, 72% (18 of 25) were male. The mean age was 15 y (range, 7–50 y) (Fig. 1).
Three study patients chose to combine the BT with an oral glucose tolerance test taking a lower glucose load. The remaining patients took the full glucose dose. The prevalence of SBBO was higher in the CF study group (56%, or 14 of 25) than in the control group of at-risk GI patients (20% or 5 of 25) (P = 0.02). In the study group, all patients diagnosed with SBBO had a fasting hydrogen level of ≥15 ppm, whereas only 2 of the 5 diagnosed with SBBO in the control group had a high fasting hydrogen. Of the patients with a positive BT, a rise in hydrogen of ≥10 ppm over fasting was seen in 3 of 14 study patients and 4 of 5 control patients. The rise occurred a mean of 47 min after ingestion of glucose in the CF patients and after 30 min in the control patients (not significantly different). A fasting level only, which was >15 ppm, was obtained for 1 study patient.
The mean fasting hydrogen level of all of the patients in the study group was 22 ppm (±22), which was significantly higher than the 5 ppm (±4) in the control group (P = 0.0001) (Fig. 2). The pattern of hydrogen excretion seen in the patients with positive BTs differed between the study and control groups (Fig. 3). No patient was diagnosed with SBBO on the basis of a rise in methane. Few patients excreted methane: 8% (2 of 25) in the study group and 4% (1 of 13) in the control group.
No statistical differences were detectable between the BT-positive and BT-negative study patients as determined by the symptoms questionnaire. The mean quality-of-life scores for the BT-positive study patients compared with BT-negative patients were 69.6 (±15.0) and 69.4 (±8.0) for the vitality dimension (P = 0.97), 74.6 (±16.0) and 69.2 (±14.5) for the digestion dimension (P = 0.37), and 81.5 (±18.7) and 83.3 (±16.2) for the eating habits dimension (P = 0.79).
The use of low-dose (Monday, Wednesday, Friday) azithromycin as an anti-inflammatory was associated with increased odds of a positive BT. The use of daily laxatives and inhaled ipratropium was associated with decreased odds of a positive BT. Other medications, patient characteristics, and laboratory values examined did not reach significance. See Table 1 for summarized results.
In this study, most patients were diagnosed as having SBBO on the basis of a high fasting hydrogen level. This phenomenon has been noted in CF patients in previous studies (7,8,19). This high level presumably is due to low-level fermentation of endogenous carbohydrates such as glycoproteins by gut microflora in the upper small bowel (20). Fasting hydrogen levels in healthy children average 7.1 ± 5.0 ppm (21). High fasting levels of hydrogen are associated with bacterial overgrowth (14,16) and with the presence of fermentable carbohydrates in the colon (22). The cutoff value for defining a high fasting hydrogen varies considerably between studies (14,19,22,23). We chose a hydrogen level of 15 ppm as our cutoff on the basis of the curve patterns of hydrogen excretion in our study and control groups and our review of the literature. It is important to note that our comparison control group was an at-risk group with a high prevalence of positive BTs of 20% rather than normal controls with no risk factors and presumably a lower prevalence of SBBO.
When fasting hydrogen levels in BTs are interpreted, other factors such as slow transit of complex carbohydrates through the intestine must be considered. In 1 study of CF patients, 6 of 7 with fasting hydrogen levels >20 ppm after a 12-h fast decreased their fasting hydrogen to <20 ppm after prolonging the fast to up to 23 h (19). Other motility studies in patients with CF also showed prolonged intestinal transit, with a mean transit time of 136 versus 88 min for control subjects (24). Another possible cause for high fasting hydrogen levels may be the presence of malabsorbed simple carbohydrate in the colon. We did not assess this in our study, but prior studies have shown that pancreatic enzyme supplementation results in normal digestion of starch (25).
BTs demonstrate an early peak in hydrogen excretion following an oral sugar load to diagnose SBBO. We chose glucose rather than lactulose as our carbohydrate substrate because of the more accurate interpretation of the glucose BT (26). With a glucose BT, any rise in hydrogen is abnormal. With a lactulose BT, the early rise in SBBO must be distinguished from the late colonic rise, which may be difficult when intestinal motility is disordered as in CF. The rise is early because the abnormally high number of bacteria present in the proximal small intestine ferment the glucose more rapidly than it can be absorbed. In this study, only a small number of CF study patients showed a rise in hydrogen after the glucose challenge. The hydrogen peak may have been blunted by the effects of abnormal ileal pH. Low colonic pH has been shown to cause false-negative lactose and lactulose hydrogen BTs because the pH is outside the optimal level for the bacteria to ferment the sugar (27,28). Because pancreatic-insufficient patients also lack pancreatic bicarbonate secretion, gastric acid is inadequately neutralized, and ileal pH is significantly lower in patients with CF (29,30). Another potential cause of a failure to note a postglucose rise in hydrogen levels in CF patients with SBBO may be unusually rapid uptake of glucose from the small intestine in patients with CF. Rapid uptake, shown in in vivo and in vitro studies, would limit the availability of glucose for fermentation by bacteria (31,32). Blunting of the early rise in hydrogen also may be seen in the setting of delayed gastric emptying. Although gastric emptying was not assessed in this study, patients with diabetes requiring insulin therapy were excluded because of concerns about gastroparesis in this subgroup.
With regard to predisposing factors for SBBO in pancreatic-insufficient CF patients, 1 factor may be the low level of pancreatic secretions. Pancreatic secretions have intrinsic antibacterial activity and participate in the control of intestinal flora. The antibacterial activity is independent of digestive enzyme activity and is thought to be due to the activity of a polypeptide present in the secretions (33,34). Supporting evidence comes from studies showing that pancreatic juice from patients with chronic pancreatitis has reduced in vitro bacterial-killing properties (35). In rats, bacterial counts in the cecum increased after obstruction of the pancreatic duct (36). Pancreatic-insufficient CF patients take pancreatic enzyme replacement therapy to supply exogenous digestive activity, but undoubtedly beneficial substances in native pancreatic secretions are not being replaced by the currently prescribed porcine-derived enzymes.
The abnormally thick intestinal mucus layer noted in CF may also predispose these patients to SBBO. Increased production of luminal mucus is described as a possibly protective response to bacterial overgrowth in rats (37). However, the nature of mucus in CF is altered, and the thickened layer, rather than being protective, may contribute to bacterial overgrowth by blocking natural defenses. In the CFTR knockout mouse intestine, mucus filled the crypts and the area between the villi; this effect was not seen in the wild-type mice (10). The luminal mucus was filled with bacteria in the CF mice only. Staining showed that Paneth cell lysozymes appeared to be trapped in the crypts of the CF mouse intestine. These lysozymes are normally released into the gut to participate in the defense against bacteria. In addition, the stasis resulting from poor motility in CF may contribute to SBBO by impairing the clearance of mucus, which provides a reservoir for bacteria.
We found no statistically significant difference in body mass index or transaminase or hematocrit levels between BT-positive and BT-negative patients, although such differences had been noted in previous studies (38–41). However, our study results suggest that laxative use may have a protective effect (polyethylene glycol PEG 3350 or docusate sodium). This may be due to the shortened transit time that results from laxative use. We also noticed a protective role of inhaled ipratropium (Atrovent, Boehringer Ingelheim Pharmaceuticals, Ridgefield, CT). This finding was unexpected because this product is used for its respiratory anticholinergic effects. In the GI tract the anticholinergic effects may be expected to cause smooth muscle relaxation and stasis, favoring SBBO. However, the protective role may be mediated through its effects on goblet cells because anticholinergics decrease goblet cell secretion and therefore mucus production (42). Finally, the use of low-dose azithromycin as an anti-inflammatory was associated with a positive BT. This broad-spectrum macrolide antibiotic may inhibit the growth of beneficial commensal bacteria in the gut and allow proliferation of resistant bacteria.
Despite studies supporting the role of proton pump inhibitor use in gastric bacterial overgrowth, we found no correlation between the use of a gastric acid suppressor and a positive BT in CF patients. Some physicians believe SBBO in CF is related to swallowed bacteria-laden respiratory secretions rather than overgrowth of colonic-type flora. The role of gastric acid would be more pertinent to overgrowth with respiratory flora as a result of failure of gastric bacterial killing (43). The lack of association noted in this study favors overgrowth with colonic-type flora, and mouse studies have shown a predominance of Enterobacteriacae in the CFTR knockout mouse intestine (10).
This pilot study demonstrates a high prevalence of positive BTs among CF patients. The findings suggest a high prevalence of SBBO. Multiple confounding variables should be considered in attempts to interpret the results of BTs in this patient population. Historically, the gold standard for the diagnosis of SBBO has been culture of jejunal fluid. However, this technique excluded analysis of unculturable bacteria and may be surpassed by bacterial DNA analysis in the future. Either way, this study needs to be confirmed by analysis of jejunal samples in a comparable population. Although the CF population has many risk factors for SBBO, the clinical significance is uncertain, especially among those who respond well to pancreatic enzyme replacement therapy. However, certain medications were seen to confer an additional risk for or apparent protection from a positive BT. The use of acid-suppressing medications had no effect. This small pilot study is not conclusive but points to the dearth of clinical research in intestinal issues of CF patients. Some medications used in CF maybe beneficial for the lungs but have a deleterious effect on the GI tract, whereas others may improve GI health. As CF patients enjoy improved pulmonary health, intestinal issues deserve closer investigation.
The authors wish to thank Mariska Anderson, LRSA1, the staff of the LPCH CF Center at Stanford, and Robert C. De Lisle, PhD, for his helpful comments.
1. Hamilton JD, Dyer NH, Dawson AM, et al
. Assessment and significance of bacterial overgrowth in the small bowel. Q J Med 1970; 39:265–285.
2. Pignata C, Budillon G, Monaco G, et al
. Jejunal bacterial overgrowth and intestinal permeability in children with immunodeficiency syndromes. Gut 1990; 31:879–882.
3. Fried M, Siegrist H, Frei R, et al
. Duodenal bacterial overgrowth during treatment in outpatients with omeprazole. Gut 1994; 35:23–26.
4. Thorens J, Froehlich F, Schwizer W, et al
. Bacterial overgrowth during treatment with omeprazole compared with cimetidine: a prospective randomised double blind study. Gut 1996; 39:54–59.
5. Armbrecht U, Eden S, Seeberg S, et al
. The value of the hydrogen (H2
) breath test for the diagnosis of bacterial overgrowth in gastric achlorhydria. Hepatogastroenterology 1987; 34:219–222.
6. Sokol RJ, Durie PR. Recommendations for management of liver and biliary tract disease in cystic fibrosis Cystic Fibrosis
Foundation Hepatobiliary Disease Consensus Group. J Pediatr Gastroenterol Nutr 1999; 28(Suppl 1):S1–S13.
7. Escobar H, Perdomo M, Vasconez F, et al
. Intestinal permeability to 51
Cr-EDTA and orocecal transit time in cystic fibrosis
. J Pediatr Gastroenterol Nutr 1992; 14:204–207.
8. Dalzell AM, Freestone NS, Billington D, et al
. Small intestinal permeability and orocaecal transit time in cystic fibrosis
. Arch Dis Child 1990; 65:585–588.
9. Rubinstein S, Moss R, Lewiston N. Constipation and meconium ileus equivalent in patients with cystic fibrosis
. Pediatrics 1986; 78:473–479.
10. Norkina O, Burnett TG, De Lisle RC. Bacterial overgrowth in the cystic fibrosis
transmembrane conductance regulator null mouse small intestine. Infect Immunol 2004; 72:6040–6049.
11. Lewindon PJ, Robb TA, Moore DJ, et al
. Bowel dysfunction in cystic fibrosis
: importance of breath testing. J Pediatr Child Health 1998; 34:79–82.
12. O'Brien S, Mulcahy H, Fenlon H, et al
. Intestinal bile acid malabsorption in cystic fibrosis
. Gut 1993; 34:1137–1141.
13. Karcher RE, Truding RM, Stawick LE. Using a cutoff of <10 ppm for breath hydrogen testing: a review of five years' experience. Ann Clin Lab Sci 1999; 29:1–8.
14. Kerlin P, Wong L. Breath hydrogen testing in bacterial overgrowth of the small intestine. Gastroenterology 1988; 95:982–988.
15. Riordan SM, McIver CJ, Walker BM, et al
. The lactulose breath hydrogen test and small intestinal bacterial overgrowth. Am J Gastroenterol 1996; 91:1795–1803.
16. Corazza GR, Menozzi MG, Strocchi A, et al
. The diagnosis of small bowel bacterial overgrowth: reliability of jejunal culture and inadequacy of breath hydrogen testing. Gastroenterology 1990; 98:302–309.
17. Corazza GR, Benati G, Strocchi A, et al
. The possible role of breath methane measurement in detecting carbohydrate malabsorption. J Lab Clin Med 1994; 124:695–700.
18. Quittner AL. Measurement of quality of life in cystic fibrosis
. Curr Opin Pulm Med 1998; 4:326–331.
19. Bali A, Stableforth DE, Asquith P. Prolonged small-intestinal transit time in cystic fibrosis
. BMJ 1983; 287:1011–1013.
20. Perman JA, Modler S. Glycoproteins as substrates for production of hydrogen and methane by colonic bacterial flora. Gastroenterology 1982; 83:388–393.
21. Perman JA, Modler S, Barr RG, et al
. Fasting breath hydrogen concentration: normal values and clinical application. Gastroenterology 1984; 87:1358–1363.
22. Brummer RJ, Armbrecht U, Bosaeus I, et al
. The hydrogen (H2
) breath test: sampling methods and the influence of dietary fibre on fasting level. Scand J Gastroenterol 1985; 20:1007–1013.
23. Riordan SM, McIver CJ, Bolin TD, et al
. Fasting breath hydrogen concentrations in gastric and small-intestinal bacterial overgrowth. Scand J Gastroenterol 1995; 30:252–257.
24. Gregory PC. Gastrointestinal pH, motility/transit and permeability in cystic fibrosis
. J Pediatr Gastroenterol Nutr 1996; 23:513–523.
25. Amarri S, Harding M, Coward WA, et al
C and H2
breath tests to study extent and site of starch digestion in children with cystic fibrosis
. J Pediatr Gastroenterol Nutr 1999; 29:327–331.
26. Romagnuolo J, Schiller D, Bailey RJ. Using breath tests wisely in a gastroenterology practice: an evidence-based review of indications and pitfalls in interpretation. Am J Gastroenterol 2002; 97:1113–1126.
27. Perman JA, Modler S, Olson AC. Role of pH in production of hydrogen from carbohydrates by colonic bacterial flora: studies in vivo and in vitro. J Clin Invest 1981; 67:643–650.
28. Vogelsang H, Ferenci P, Frotz S, et al
. Acidic colonic microclimate: possible reason for false negative hydrogen breath tests. Gut 1988; 29:21–26.
29. Gilbert J, Kelleher J, Littlewood JM, et al
. Ileal pH in cystic fibrosis
. Scand J Gastroenterol Suppl 1988; 143:132–134.
30. Barraclough M, Taylor CJ. Twenty-four hour ambulatory gastric and duodenal pH profiles in cystic fibrosis
: effect of duodenal hyperacidity on pancreatic enzyme function and fat absorption. J Pediatr Gastroenterol Nutr 1996; 23:45–50.
31. Frase LL, Strickland AD, Kachel GW, et al
. Enhanced glucose absorption in the jejunum of patients with cystic fibrosis
. Gastroenterology 1985; 88:478–484.
32. Baxter P, Goldhill J, Hardcastle J, et al
. Enhanced intestinal glucose and alanine transport in cystic fibrosis
. Gut 1990; 31:817–820.
33. Laubitz D, Zabielski R, Wolinski J, et al
. Physiological and chemical characteristics of antibacterial activity of pancreatic juice. J Physiol Pharmacol 2003; 54:283–290.
34. Rubinstein E, Mark Z, Haspel J, et al
. Antibacterial activity of the pancreatic fluid. Gastroenterology 1985; 88:927–932.
35. Marotta F, Tajiri H, Li ZL, et al
. Pure pancreatic juice from patients with chronic pancreatitis has an impaired antibacterial activity. Int J Pancreatol 1997; 22:215–220.
36. Runkel N, Moody FG, Miller TA, et al
. Promotion of bacterial translocation by altered biliary and pancreatic secretions. Gastroenterology 1994; 106:A763.
37. Sherman P, Fleming N, Forstner J, et al
. Bacteria and the mucus blanket in experimental small bowel bacterial overgrowth. Am J Pathol 1987; 126:527–534.
38. Lichtman SN, Keku J, Clark RL, et al
. Biliary tract disease in rats with experimental small bowel bacterial overgrowth. Hepatology 1991; 13:766–772.
39. Lichtman SN, Sartor RB. Hepatobiliary injury associated with experimental small-bowel bacterial overgrowth in rats. Immunol Res 1991; 10:528–531.
40. Lichtman SN, Keku J, Schwab JH, et al
. Hepatic injury associated with small bowel bacterial overgrowth in rats is prevented by metronidazole and tetracycline. Gastroenterology 1991; 100:513–519.
41. Smyth RL, Croft NM, O'Hea U, et al
. Intestinal inflammation in cystic fibrosis
. Arch Dis Child 2000; 82:394–399.
42. Furuya S, Naruse S, Hayakawa T. Intravenous injection of guanylin induces mucus secretion from goblet cells in rat duodenal crypts. Anat Embryol (Berl) 1998; 197:359–367.
43. Husebye E. The pathogenesis of gastrointestinal bacterial overgrowth. Chemotherapy 2005; 51(Suppl 1):1–22.