Small intestinal bacterial overgrowth (SIBO) is a common cause of chronic diarrhea and malabsorption. Morphologic changes associated with this condition have not, to our knowledge, been studied in detail.
To better characterize the histopathologic changes associated with SIBO by comparing the clinicopathologic features of patients with SIBO (duodenal aspirate cultures with ≥105 colony-forming units [CFUs]/mL) to controls with cultures found to be negative (<105 CFUs/mL).
We included 67 consecutive patients with SIBO and 55 controls in the series. Each duodenal biopsy was assessed for the following features: villous to crypt ratio, intraepithelial lymphocytosis, crypt apoptoses, basal plasmacytosis, cryptitis/villitis, peptic duodenitis, erosions/ulcers, eosinophilia, and absence of goblet and Paneth cells; and correlated with clinical features and culture results.
Decreased villous to crypt ratio (<3∶1) was more frequent in SIBO than controls (24% versus 7%; P = .01). Duodenal biopsies from patients with SIBO were slightly less likely to be judged within reference range than were controls (52% versus 64%; P = .27). There were no significant differences in any of the other histologic features. Clinically, patients in the SIBO group were older than the age of controls (mean, 60 years versus 52 years; P = .02), and they were more likely to have one of the known predisposing factors for bacterial overgrowth (66% versus 36%; P = .002). Other clinical features, including presenting symptoms, were similar.
Villous blunting is the only feature more common to SIBO than to controls. More than half of biopsies from SIBO patients are histologically unremarkable. Therefore, SIBO needs to be considered as a potential etiology for gastrointestinal symptoms even when duodenal biopsies are found to be normal.
Small intestinal bacterial overgrowth (SIBO) is a common cause of chronic diarrhea.1 Small intestinal bacterial overgrowth results from colonization of the proximal small bowel by gram-negative aerobic and anaerobic bacteria that are normally restricted to the colon or, less frequently, from overgrowth of oropharyngeal flora. A predisposition to SIBO exists in diverse conditions where there is altered anatomy from prior surgery (eg, blind loop syndrome) or stricture or where there is impaired gut motility and prolonged orocecal transit time. Small intestinal bacterial overgrowth has also been reported in progressive systemic sclerosis, jejunal diverticulosis, 33% of patients with portal hypertension, 36% of those with chronic renal failure, 59% of uncomplicated acute colonic diverticulosis, 54% of overt hypothyroidism, 44% of acromegaly, and 96% of Crohn disease with small-bowel strictures.2–9 Studies in the pediatric population have documented SIBO in 37.5% of children living in an urban slum (ie, environmental enteropathy), 56% of patients with pancreatic insufficiency due to cystic fibrosis, 61% of infants receiving parenteral nutrition because of short bowel syndrome, and commonly following many other types of bowel surgery in the neonatal period.10–13 In the elderly, SIBO is a common but underrecognized cause of malnutrition, especially among disabled older adults, among whom, 26% suffer from SIBO.14–16
Several methods have been used to detect SIBO, but each has its disadvantages. The gold standard for diagnosis is a small intestinal aspirate culture showing growth of at least 105 colony-forming units of bacteria per milliliter (CFU/mL) of duodenal or jejunal fluid.17 Cultures of small-bowel mucosal biopsies can be substituted when there are inadequate luminal secretions.18 Noninvasive breath tests may be used to make a diagnosis of SIBO, but these are labor-intensive and have shown wide-ranging sensitivity and specificity.1,17,19,20 Another concern is the (minor) radiation exposure in breath testing using 14C-xylose or 14C-glycocholic acid.21
The role of duodenal morphology—if any—in evaluating for SIBO is not entirely clear. Scanning electron microscopy has shown derangements of villous architecture and injury to enterocytes and their microvilli.22 Higher counts of immunoglobulin (Ig) A–producing plasma cells in the lamina propria and CD8+ lymphocytes in the villous epithelium exist in some cases of SIBO, but demonstration requires immunohistochemistry and mucosal morphometry.23 Only a few studies of routine light microscopic changes in SIBO have been published, and these are complicated by conflicting results, different methods of diagnosis, and small numbers of patients.6,12,24–28
Given the limited and conflicting data in the literature on the light microscopic appearance of the small intestinal mucosa in the setting of bacterial overgrowth, we sought to delineate the histopathologic features of SIBO detectable in a routine pathology practice by study of a large group of patients with clearly defined disease. Identification of specific histopathologic features—or the absence thereof—in this condition may be helpful to direct the diagnostic workup and therapy in patients with malabsorption or diarrhea.
MATERIALS AND METHODS
We conducted a retrospective search of the microbiology database at the Mayo Clinic, Rochester, Minnesota, for patients who had both duodenal or jejunal aspirate cultures and corresponding duodenal biopsies during 2006–2007. All of the patients underwent upper gastrointestinal endoscopic examination because of gastrointestinal symptoms. Cultures were considered positive when there was growth of at least 105 CFU/mL of gram-negative bacteria, aerobic or anaerobic, and negative when there was less than 105 CFU/mL. Patients were excluded from the study if duodenal biopsy slides were unavailable for review, if there was a history of upper gastrointestinal Crohn disease or recent infection with enteric pathogens or if there was a clinical diagnosis of celiac disease or celiac serologic markers were found to be unequivocally positive. The final study population comprised 122 consecutive patients meeting the above criteria: 67 patients (54.9%) with positive cultures and 55 control patients (45.1%) with negative cultures. A total of 18 patients were excluded from the study. Approval for the study was obtained from the Mayo Clinic Institutional Review Board.
The computerized medical records were searched for the following parameters: (1) age and sex; (2) presenting symptoms; (3) serologic tests for gluten sensitivity (antigliadin IgA and IgG, antiendomysial, and tissue transglutaminase IgA and IgG antibodies); (4) documented vitamin or mineral deficiencies; (5) albumin level; (6) presence of ascites; (7) history of upper gastrointestinal surgery; (8) concomitant gastrointestinal or colonic diseases; (9) endoscopic appearance of the upper gastrointestinal tract, including evaluation for any structural abnormalities; and (10) any other medical conditions that might predispose the patient to SIBO, such as portal hypertension or chronic renal failure.
All culture samples were obtained from the small intestine endoscopically and were placed into an anaerobic sample container for aerobic and anaerobic cultures. For aerobic culture, the number of aerobic gram-negative bacilli was reported as “aerobic gram-negative bacilli <105 CFU/mL” or “aerobic gram-negative bacilli ≥105 CFU/mL.” Yeast was reported when pure or predominant growth was noted. When present with other flora, yeast was reported when growth was greater than 1000 CFU/mL. Aerobes other than aerobic gram-negative bacilli and yeast were not reported. Both aerobic sheep blood agar and anaerobic plates were examined, and the single plate with the highest colony count was selected; the total flora was reported as greater or less than 105 CFU/mL.
All duodenal biopsies had been fixed in formalin and processed routinely for hematoxylin-eosin staining of 4-µm sections. Two of us (P.J.L. and T.-T.W.) were blinded to the culture results and clinical information and jointly assessed the biopsies for the following features: (1) villous to crypt (v∶c) ratio, (2) number of surface and crypt intraepithelial lymphocytes, (3) number of crypt apoptoses, (4) basal plasmacytosis, (5) neutrophilic cryptitis or villitis, (6) peptic duodenitis, (7) accompanying erosions or ulcers, (8) increased mucosal eosinophils, and (9) abnormal absence of goblet cells and/or Paneth cells. Villous length to crypt depth ratio was considered abnormal if it was less than 3∶1 in well-oriented biopsy fragments. Intraepithelial lymphocytosis was evaluated by counting the number of lymphocytes within the villous epithelium in the area of highest lymphocyte density after scanning the biopsy fragments at low power. Cases were categorized as normal (≤20 lymphocytes per 100 epithelial cells) or increased (21–40, 41–60, 61–80, and >80 lymphocytes per 100 epithelial cells). Crypt intraepithelial lymphocytosis was evaluated in a similar fashion. For crypt apoptosis, any number greater than 5 apoptotic bodies per 10 crypts was considered abnormal. Basal plasmacytosis was characterized by lymphoplasmacytic inflammation located between the muscularis mucosae and the base of the crypts with or without a qualitative increase in mononuclear cells in the intercryptal spaces and villi. We defined peptic change histologically as gastric foveolar metaplasia of the small-bowel epithelium and increased intramucosal Brunner glands in postbulbar biopsies. Because there is no well-defined reference range for eosinophil counts in small-bowel lamina propria, the determination of increased mucosal eosinophils was subjective; we also considered anything more than rare intraepithelial eosinophils to be abnormal.
Fisher exact test and t test were used to compare the SIBO and control groups for categoric and noncategoric data, respectively. Two-tailed P values <.05 were considered significant.
The demographic data are summarized in Table 1. Patients with SIBO were on average older than controls (mean [SD], 60 (14) y vs 52 (19) y; P = .02). The SIBO and control groups had similar presenting symptoms, with diarrhea (55%; 37 of 67), weight loss (27%; n = 18), abdominal pain (27%; n = 18), and abdominal bloating (18%; n = 12) being the most common symptoms in patients with SIBO, and with diarrhea (64%; 35 of 55), abdominal pain (29%; n = 16), nausea (22%; n = 12), and weight loss (18%; n = 10) reported most commonly in controls. A slight female predominance was present in both groups (SIBO, 53%, 36 of 67; and control, 58%, 32 of 55, respectively; P = .71).
As expected, conditions that are known to predispose to bacterial overgrowth were more prevalent in the SIBO group (Table 2): 66% of patients with SIBO (44 of 67) had predisposing conditions as compared with only 36% (20 of 55) in the control group (P = .002). The most common condition in both groups was the use of acid-lowering medications. The presence of a blind loop was significantly more frequent in the SIBO group than in controls (12% [8 of 67] vs 2% [1 of 55]; P = .04). Two patients (3%) with SIBO had scleroderma, 2 (3%) had prior resection of the ileocecal valve, and 2 (3%) had chronic pancreatitis; none of the patients in the control group had these conditions.
Overall, 18% of patients with SIBO (12 of 67) had one or more of the other conditions, most commonly corticosteroids or other immunosuppressants (in 13%; 9 of 67). This did not differ significantly from the control group, 16% (9 of 55) of whom had at least one of the above conditions, including immunosuppressive medications in 15% (8 of 55; P >.99). More detailed results are given in Table 3. Two patients (3%) in the SIBO group had minimally elevated levels of tissue transglutaminase antibodies (one IgA, the other IgG), and two patients (3.6%) in the control group also had minimal elevations in tissue transglutaminase IgA antibodies (range, 22.3–27.6 U; reference range, <20 U). None of these 4 patients was considered clinically to have celiac sprue. One patient in the SIBO group (1.5%), and none in the control group had Helicobacter pylori infection found by histology or other diagnostic means.
Small Intestinal Aspirate Cultures
Specific data regarding bacterial growth in anaerobic and aerobic cultures, and growth of yeast in the SIBO and control groups, are given in Table 4. Notably, all 67 patients with SIBO (100%) had significant (≥105 CFU/mL) growth of bacteria in anaerobic conditions; 9 (13%) of these also grew ≥105 CFU/mL gram-negative bacilli in aerobic cultures, and 12 (18%) grew yeast, including 6 (9%) with 103 to 104 CFU/mL, 4 (6%) with 104 to 105 CFU/mL, and 2 (3%) with >105 CFU/mL. The Bacteroides fragilis group was detected in one (1.5%) case. Among the 55 controls, 16 aspirates (29%) showed no growth at all in anaerobic or aerobic conditions. The other 39 aspirates (71%) grew low numbers (<105 CFU/mL) of anaerobes and/or gram-negative aerobes; 4 (7%) had 103 to 104 CFU/mL of yeast.
The histologic findings are summarized in Table 5. Patients with SIBO were more likely to have architecturally abnormal duodenal biopsies, with a v∶c ratio of less than 3∶1 in 16 (24%) of cases with SIBO but in only 4 controls (7%; P = .02; Figures 1 and 2, A). In contrast, there were no significant differences between the 2 groups in any of the other histopathologic features, including villous and crypt intraepithelial lymphocytosis, crypt apoptoses, basal plasmacytosis, neutrophilic cryptitis/villitis, erosion/ulcers, increased mucosal eosinophils, presence of goblet and Paneth cells, or peptic duodenitis (Figures 2, B, and 3). Overall, 35 of the duodenal biopsies from patients with SIBO (52%) were judged to be within reference range, whereas the other 48% (32 of 67) demonstrated one or more histologic abnormalities. In the control group, there were slightly higher numbers of biopsies considered normal (64%; 35 of 55), but that difference was not statistically significant (P = .27).
To evaluate whether there was any significance to growth in low numbers of organisms from aspirate cultures, we separately compared each histopathologic feature in 3 patient groups: (1) 16 controls (29%) with no anaerobic or aerobic growth; (2) 39 controls (71%) with growth of <105 CFU/mL of anaerobes, gram-negative aerobes, or yeast; and (3) 67 cases of SIBO (100%). With the exception of a decreased v∶c ratio, there were again no significant differences between these groups in any specific histologic finding. In particular, small-bowel biopsies from the 16 controls with no anaerobic or aerobic growth at all were no more likely to be within histologic reference range than those from the other 2 groups (P = .25).
Small intestinal bacterial overgrowth is a well-recognized cause of chronic diarrhea and malabsorption in patients with anatomic abnormalities of the small bowel—such as surgical blind loops—and in patients with motility disorders that promote colonization of the small bowel with gram-negative flora from the colon. In the present study, we sought, by studying a large group of patients who mirrored the general population undergoing upper endoscopy, to identify whether there are histologic abnormalities observable by routine light microscopy that might suggest a diagnosis of SIBO. All 122 patients in our study had upper gastrointestinal symptomatology, and all were diagnosed as having—or not having—SIBO by duodenal or jejunal aspirate cultures (considered to be the diagnostic gold standard for bacterial overgrowth). As expected, there were some clinical differences between SIBO cases and controls. Patients with SIBO were on average older than patients without SIBO (60 years vs 52 years; P = .02), underscoring prior studies that have shown a higher prevalence of SIBO in elderly patients who are more likely to suffer from altered gut motility, poor nutrition, and/or medical conditions that predispose them to bacterial overgrowth.29–33 In our study, patients with SIBO were also more likely to have one or more risk factors for bacterial overgrowth (66% versus 36%; P = .002).
Numerous conditions have been associated with SIBO, including blind loop, surgical adhesions, scleroderma, chronic renal failure, resection of the ileocecal valve, chronic pancreatitis, and intestinal dysmotility.2–16 We included use of acid-lowering medications as a possible risk factor, although it is likely not a strong one. These medications were common in both SIBO (34%; 23 of 67) and controls (24%; 13 of 55), but even if this factor is disregarded, the results remain statistically significant (P = .02). Aside from age and background risk factors, the 2 groups were similar. In particular, there was a marked overlap in presenting symptoms, with diarrhea and abdominal pain being the most common indications for small bowel biopsy in both SIBO cases and control cases.
Several recent studies suggest that SIBO is much more common in the general population than previously thought and that it is frequently underdiagnosed. For example, Teo et al1 prospectively studied 87 consecutive patients with chronic diarrhea, and SIBO was by far the most frequent etiology, accounting for symptoms in 48% of cases (n = 42), whereas irritable bowel syndrome accounted for only 13% (n = 11). There is also increasing evidence that SIBO may be the underlying etiology in some cases of irritable bowel syndrome, which is currently the most common diagnosis rendered in gastroenterology practice.34–37 Pimental et al37 used lactulose hydrogen breath testing to evaluate for SIBO in 202 patients with a clinical diagnosis of irritable bowel syndrome and obtained positive results in 78% (n = 158). Although breath tests suffer from lack of specificity, in the Pimentel et al37 study, eradication of SIBO after antibiotic therapy was significantly associated with improvement in symptoms of diarrhea and abdominal pain, to the extent that 48% of subjects who achieved eradication (n = 97) no longer met clinical criteria for irritable bowel syndrome.37 Recently, it has been shown that treatment with rifamixin—a nonabsorbable oral antibiotic—is both effective and safe for bacterial eradication and results in significant symptom improvement in patients with SIBO.38
Given the evidence that SIBO is a prevalent and treatable—but underrecognized—condition, it is important to ask whether the histologic features of small intestinal biopsies can provide useful input into the diagnosis of SIBO. Histologically, we found that modest architectural distortion was the most common abnormality in SIBO. Villous blunting was present in 24% of SIBO biopsies (16 of 67) but only 7% of controls (4 of 55; P = .01). In most cases, blunting was mild or moderate in degree, with only one SIBO biopsy (1%; and none of the controls) that demonstrated mucosal flattening. Lymphocytosis within villous epithelium was also slightly more frequent in SIBO (22%; 15 of 67) than in controls (15%; 8 of 55), but this did not reach statistical significance. As with villous blunting, the degree of intraepithelial lymphocytosis was typically modest. None of the patients with SIBO had more than 60 lymphocytes per 100 enterocytes, and most (13 of 15; 87%) of the biopsies that were classified as abnormal had only 21 to 40 lymphocytes per 100 enterocytes, a range that would be considered borderline in clinical practice. This suggests that SIBO would be an unlikely etiology for a small bowel biopsy that showed severe architectural distortion or severe lymphocytosis and that such biopsies should be investigated for other possible etiologies, such as celiac disease. Overall, 52% of duodenal biopsies from patients with SIBO (35 of 67) and 63% of biopsies from nonceliac, non-SIBO (but symptomatic) controls (35 of 55) in our study population were within histologic reference range. The high rate of histologically “normal” biopsies does not negate the pathogenic role of bacterial overgrowth in symptomatic patients. It has been hypothesized that excess colonic flora within the small intestine can impede fat digestion by deconjugating bile acids and hindering the formation of micelles, that these bacteria can induce vitamin B12 deficiency, and that they can result in osmotic or secretory diarrhea through the production of organic acids, free bile acids, fatty acids, and other organic acids.1 All of these proposed mechanisms can coexist with microscopically healthy small bowel mucosa.
Injury to enterocytes/enterocytes microvilli and abnormal villous architecture have been reported in a study using scanning electron microscopy,22 but there is limited published data on the light microscopic appearance of the small bowel mucosa in the setting of SIBO.23,24,26–28,39–40 Investigators have described an absence of mucosal abnormalities, patchy villous blunting, villous blunting, and villous blunting with increased intraepithelial lymphocytes in SIBO.23,24,26,28,40 Overall, the published findings related to the light microscopic findings in SIBO are based on studies of small numbers of patients, use varying diagnostic tests to define SIBO, and have included control groups that range from truly healthy patients to patients with diarrhea-predominant irritable bowel syndrome. The results of these studies have been conflicting.
Ament et al24 found mild to moderate villous blunting and increased cellularity in the lamina propria in 3 of 3 patients (100%) studied but emphasized the patchy nature of these abnormalities in the numerous biopsies taken from these patients In contrast, Riordan et al23 reported no differences in villous height, crypt depth, or v∶c ratio between 26 patients with SIBO and 26 controls. In that study,23 and in another by Riordan et al,28 using immunohistochemistry, patients with SIBO had significantly increased IgA+ plasma cells in the lamina propria and increased CD8+ intraepithelial lymphocytes. Computer-assisted morphometric analysis of the small bowel mucosa on well-oriented biopsies has also shown light microscopic changes in patients with SIBO. Mucosal abnormalities were identified in 16 elderly patients with SIBO, including decreased villous height, decreased crypt depth, and decreased total mucosal height, compared with 23 controls.25 These abnormalities resolved in patients treated with antibiotic therapy. Another study in infants with diarrhea due to enteropathogenic Escherichia coli strains showed similar mucosal abnormalities by morphometric analysis.39 Although morphometric studies confirm that there are changes in villous architecture in SIBO and in infection, the practice of orienting small bowel biopsies and applying computer-assisted morphometry would not be applicable to everyday practice. Even the routine use of immunohistochemistry for IgA and CD8 is likely to be financially prohibitive.
Many conditions other than bacterial overgrowth can potentially lead to alteration of small bowel mucosal architecture, including use of corticosteroids or other immunosuppressants, recent antibiotic use, zinc deficiency, alcoholism, hypogammaglobulinemia, infection with enteric pathogens, and upper gastrointestinal Crohn disease. As previously described, patients with Crohn disease or documented infection with enteric pathogens were not included in this study. The presence of mild villous blunting shown in the present study, as well as other published literature,23,24,26,28,40 is a nonspecific finding, but in the right clinical setting, such as chronic diarrhea, should also flag SIBO as a potential etiology in the differential diagnosis.
In summary, our results show that only approximately half of small intestinal biopsies from symptomatic patients with a gold standard diagnosis of SIBO show any histologic abnormality on routine hematoxylin-eosin examination. When abnormalities are present—most commonly villous blunting or increased intraepithelial lymphocytes—they are typically mild and relatively nonspecific. Therefore, when trying to determine the etiology for a patient's upper gastrointestinal symptoms, both pathologists and gastroenterologists should be aware of the limitations of histology in diagnosing, suggesting, or ruling out SIBO. Small-bowel biopsies remain extremely important in the evaluation for other gastrointestinal diseases (eg, celiac disease and peptic duodenitis), but if SIBO is a diagnostic consideration then specific clinical or microbiologic testing for bacterial overgrowth should also be performed.
The authors have no relevant financial interest in the products or companies described in this article.
From the Departments of Laboratory Medicine and Pathology (Drs Lappinga, Patel, and Wu, and Ms Vetter) and Gastroenterology (Dr Murray), Mayo Clinic, Rochester, Minnesota; and the Department of Pathology, M. D. Anderson Cancer Center, Houston, Texas (Dr Abraham).