Abnormal Intestinal Permeability in Primary Biliary Cirrhosis
- First Online:
- Cite this article as:
- Feld, J.J., Meddings, J. & Heathcote, E.J. Dig Dis Sci (2006) 51: 1607. doi:10.1007/s10620-006-9544-z
- 131 Downloads
Antimitochondrial antibodies (AMAs) found in patients with primary biliary cirrhosis (PBC) cross-react with bacterial proteins and hence molecular mimicry has been proposed as a mechanism for AMA development. Alterations in gastrointestinal permeability would provide a potential route for increased exposure of gut flora to the immune system. In this study we aimed to compare the measured gastrointestinal permeability in patients with PBC to that in patients with liver disease (hepatitis C) and healthy control populations. Subjects drank a mixture of sucrose, lactulose, and mannitol dissolved in water. Eight-hour urinary excretion of the sugars was measured to assess intestinal permeability. Antiendomysial antibody testing was performed to exclude subclinical celiac disease. Eighty-six patients with PBC were evaluated and compared to 69 hepatitis C patients and 155 healthy controls. The mean urinary excretion of sucrose in the PBC patients (133.89 ± 72.56 mg) was significantly higher than that in hepatitis C patients (101.07±63.35) or healthy controls (89.46±41.76) (P=0.0001), suggesting abnormal gastric or proximal small intestinal permeability. Sucrose excretion was not increased among patients with hepatitis C compared to healthy controls. The ratio of lactulose:mannitol excretion, reflecting small bowel permeability, was also elevated in the PBC group (0.017±0.012) compared to healthy controls (0.012±0.007) (P=0.0001) but was equal to that found among patients with hepatitis C (0.016±0.011) (P=NS). We conclude that the permeability of both the stomach and the small bowel is increased in patients with PBC, however, it is unclear if it is a cause, consequence, or manifestation of the disease.
KeywordsPrimary biliary cirrhosis Intestinal permeability Molecular mimicry Portal hypertension
Primary biliary cirrhosis (PBC) is a chronic, progressive, cholestatic liver disease that primarily affects middle-aged women. The precise etiology of PBC is unknown, however, there is substantial evidence that it has an autoimmune basis which may initially be triggered by bacterial infection. Support for this hypothesis comes from the demonstration that the antimitochondrial antibodies (AMAs) found in the sera of 95% of patients with PBC have been shown to recognize both human and bacterial polypeptide antigens [1, 2, 3]. The target antigens of AMAs have been identified as the E2 subunits of the 2-oxo acid dehydrogenase complexes located on the inner mitochondrial membrane, including pyruvate dehydrogenase (PDC), 2-oxoglutarate dehydrogenase (OGDC), and branched-chain 2-oxo dehydrogenase (BCOADC) [4, 5, 6, 7, 8, 9]. While these antigenic polypeptides were first identified in humans, a high degree of structural homology has been demonstrated in their bacterial counterparts . These bacterial epitopes also react with AMAs from the sera of PBC patients [4, 10]. Hence, it has been proposed that chronic bacterial exposure may initiate the development of antibodies which then cross-react with human antigens.
Both human and animal data lend further support to the bacterial hypothesis. Antibodies with the same specificity as AMAs found in PBC can be induced in rabbits inoculated with rough (R) mutants of gram-negative bacteria . PBC patients have higher rates of colonization with R forms of Escherichia coli, suggesting that there may be a link between infection and AMA . In addition, patients with recurrent bacteriuria, but no evidence of PBC, have been found to have low titers of AMAs and patients with PBC have an increased rate of urinary tract infections [13, 14].
PBC is associated with numerous other autoimmune diseases including Hashimoto’s thyroiditis, scleroderma, and sicca syndrome . A higher prevalence of celiac sprue has also been found in patients with PBC [16, 17]. Patients with celiac disease have been shown to have increased intestinal permeability. If patients with PBC also have increased intestinal permeability, this could provide one potential mechanism for increased exposure of the immune system to bacterial antigenic stimulation.
In a series of experiments, Smecuol et al. have shown that with ingestion of a solution containing sucrose, lactulose, and mannitol, one can identify areas of altered intestinal permeability in a site-specific manner . Sucrose is absorbed in the stomach and proximal small intestine. Increased urinary recovery of intact sucrose following ingestion implies increased proximal small intestinal permeability . Lactulose and mannitol are both absorbed throughout the small intestine. Due to its small size, mannitol is thought to be able to cross the gut epithelium via a transcellular route, which is uninfluenced by the integrity of the small bowel epithelium. In contrast, lactulose, a much larger molecule, must cross the gut epithelium between intestinal epithelial cells ie paracellularly. Intercellular tight junctions regulate paracellular permeability and consequently any breach of the epithelium will be reflected by increased recovery of lactulose in the urine. To control for such confounders as small bowel motility and surface area, a combination of lactulose and mannitol is used with the results expressed as a ratio. An increased lactulose to mannitol urinary excretion ratio implies increased small intestinal permeability . Using a similar protocol, we studied patients with PBC to identify any abnormalities in intestinal permeability that could potentially provide a route for bacterial pathogens to gain greater exposure to the immune system.
Materials and methods
Patients Studied Between November 1998 and February 2001, patients given a diagnosis of PBC based on liver histology, liver biochemistry, and/or AMA positivity, followed at The Toronto Western Hospital Liver Clinic, were offered participation in the study. Patients with chronic hepatitis C not currently receiving antiviral therapy were recruited to serve as liver disease controls and both current and historical healthy controls were also studied. The study was approved by the institutional review board (The Toronto Hospital Committee for Research on Human Subjects) and informed written consent was obtained from each patient.
Patients were excluded if they were known to have celiac disease, inflammatory bowel disease, other chronic intestinal illnesses, or diabetes mellitus requiring pharmacologic therapy or if they were regularly taking ASA, nonsteroidal anti-inflammatory medications (NSAIDs), or glucocorticoids. In addition, patients were instructed to abstain from alcohol consumption for 5 days prior to participation in the study and were not to have taken NSAIDs or ASA for a minimum of 2 weeks before study entry. All patients with PBC were on ursodeoxycholic acid therapy (15 mg/kg/day).
Laboratory Analysis The study subjects were given a mixture containing 100 g of sucrose, 5 g of lactulose, and 2 g of mannitol dissolved in water and were instructed to drink it following a 4- to 5-hr fast. They voided prior to ingestion of the study solution and then collected all of their urine for the next 8 hr. The urine was collected in preweighed containers with 5 mL of 10% thymol in isopropanol to preserve the sugars. Weight and volume of each sample were recorded and two 10-mL aliquots from each sample were frozen and transported on dry ice to the University of Calgary for analysis. Using high-performance liquid chromatography (HPLC), the quantity of each sugar in the solution was measured and the ratio of lactulose to mannitol excretion was calculated. The details of the urinary analysis have been described elsewhere .
To exclude celiac disease, the sera of the patients with PBC were tested for the presence of antiendomysial antibodies (EMAs) using the indirect immunofluoresence method previously described . AMAs were measured using M2 ELISA (Pharmacia Diagnostics) or immunofluoresence (Immco). A value of 5 IU/mL or a titer ≥1:40 was considered positive. Clinical records were reviewed for all patients, and laboratory, histology, and radiology results were reviewed. Cirrhosis was defined histologically or radiologically. Portal hypertension was defined as the presence of at least one of esophageal varices, portal hypertensive gastropathy (endoscopic appearance as defined by Northern Italian Endoscopy Club ), splenomegaly (>13 cm on ultrasound), and/or ascites.
Statistical Analysis The results from the PBC cohort were compared with values from both the healthy and the hepatitis C control groups. The mean values for both the urinary excretion of sucrose and the ratio of lactulose to mannitol excretion were compared against the historical control data using the Wilcoxon rank sum test. Patients with PBC with abnormal permeability were compared to those with normal permeability. Chi-square analysis was used for categorical data and the two-sided Student t test was used to compare numerical data.
One hundred seventeen of the 198 patients with PBC followed at The Toronto Western Hospital Liver Clinic were offered participation in the study. Participation was not offered to 81 patients, because of either an inadequate understanding of the English language, an inability to travel to the clinic due to distance/illness, or known use of a medication which excluded them from the study. Of those approached, 31 were not able to participate (refusal, illness, exclusion medication, inability to complete test or travel to clinic). A total of 86 patients with PBC completed the study and 21 performed the test a second time to ensure consistency of the results. Sixty-nine patients with hepatitis C infection completed the study. The healthy control population came from previous studies of intestinal permeability. These individuals consisted of 101 with values for the ratio of lactulose to mannitol excretion and 155 with values for the urinary excretion of sucrose. Fifteen new healthy controls were also recruited.
Baseline Data for PBC and Hepatitis C Cohorts
Hepatitis C (n=69)
54.9±10.9 (range, 28–78)
44.5±8.3 (range, 29–71)
Duration of PBC (yr)
5.0±3.6 (range, 1–16)
ALP (<110 IU/L)
Bilirubin (<22 μmol/L)
Albumin (38–50 g/L)
Antiendomysial antibody (>1:20)
Permeability Data of PBC and Hepatitis C Patients and Healthy Control Cohorts
Hepatitis C (n=69)
Healthy controls (n=101†/155*)
Mean sucrose excretion (mg)*
133.89±72.56‡ (mdn = 124.15, range 31–417)
101.07±63.35‡ (mdn = 81.55, range 13–326)
89.46±41.76‡ (mdn = 85.19, range 19–202)
Mean L:M ratio†
0.017±0.012 (mdn = 0.014,§ range 0.006–0.088)
0.016±0.011 (mdn = 0.015,§ range 0.006–0.035)
0.012±0.007 (mdn = 0.011,§ range 0.002–0.029)
Number of “abnormals”¶
Among patients with PBC, 21 (25.0%) had “abnormal” permeability, defined as >2 standard deviations above the mean values for the healthy control group. Fifteen (17.9%) had abnormal sucrose excretion (gastric/proximal small bowel permeability), 9 (10.7%) had increased L:M ratios (small bowel permeability), and 3 (3.6%) had increased permeability at both sites. Sixteen (23.2%) patients with hepatitis C had “abnormal” permeability as defined above (10 [14.5%] sucrose excretion, 10 [14.5%] L:M ratio, and 4 [4.5%] both sites). The PBC patients with abnormal permeability were of similar age and gender and had similar baseline laboratory values and duration of disease as those with permeability in the “normal” range. Repeat testing 6–12 months later in 21 of the patients with PBC documented that the test results were consistent over time. A greater percentage of patients with abnormal permeability had evidence of portal hypertension (33.3% “abnormals” with portal hypertension vs. 14.3% “normals” with portal hypertension; P < 0.05). However, overall, patients with portal hypertension did not have increased mean gastric or small bowel permeability compared to those without portal hypertension (L:M with portal hypertension, 0.018±0.007, vs. without portal hypertension, 0.016±0.012 [P= 0.33], and sucrose with portal hypertension, 164.8±92.7 mg, vs. without portal hypertension,127.5±66.2 mg [P= 0.15]) (see Table 3). Two patients with PBC had very high L:M excretion ratios (see Figure 2). The mean L:M ratio for the PBC cohort remained statistically significantly higher than that for the healthy control group even with the exclusion of these two outliers (mean L:M ratio: PBC cohort, 0.015±0.006, vs. healthy controls, 0.012±0.007; P=0.01).
Permeability in both the stomach and the small bowel was found to be increased in patients with PBC. The abnormality was most striking in the stomach and proximal small bowel, where patients with PBC had increased permeability compared to both patients with liver disease (hepatitis C) and healthy controls. The mean permeability in the rest of the small bowel was also increased among patients with PBC compared to healthy controls but was no different from that found in subjects with hepatitis C. While the observed alteration in intestinal permeability may be related to PBC in a causal role, it is also possible that it is consequence of the disease or its complications.
PBC Patients with Abnormal vs Normal Permeability*
Duration of PBC
ALP (<110 IU/L)
Bilirubin (<22 μmol/L)
Albumin (38–50 g/L)
5 (23.8%) 5.4±0.55
11 (17.5%) 5.3±0.65
The only factor that distinguished the group with “abnormal” permeability from the other patients with PBC was an increased prevalence of portal hypertension. Although clinically significant portal hypertension was more common among patients with abnormal permeability, the majority (66.6%) of patients found to have abnormal permeability did not have evidence of portal hypertension. In addition, as a group, the patients with portal hypertension did not have increased mean permeability compared to those with no documented portal hypertension. Finally, a similar proportion of patients with hepatitis C and PBC had portal hypertension, yet permeability was increased in patients with PBC compared to those with hepatitis C. Studies using similar methods have come to conflicting conclusions regarding the importance of portal pressure on intestinal permeability. Two studies have documented normal passive permeability in cirrhotics [22, 23]. In contrast, Taylor et al. reported increased small bowel permeability in children with cirrhotic and noncirrhotic portal hypertension and Giofre et al. found that patients with portal hypertensive gastropathy had increased gastric permeability . Our data are in agreement with the recent report of De Leo and colleagues , who also found increased intestinal permeability in patients with PBC. Notably, they observed that portal hypertension was more prevalent in their liver disease control group than in their PBC cohort. In both studies some patients with very early-stage PBC had increased permeability and others with advanced portal hypertension had normal permeability, suggesting that portal hypertension alone is unlikely to account entirely for the observed results.
The effect of cholestasis on permeability is another potential confounder. A study of patients with extrahepatic biliary obstruction found that intestinal permeability was increased and correlated with the degree of elevation of serum bilirubin, returning toward normal following biliary drainage . Although this could be potentially relevant to the results of our study of another cholestatic liver disease, the patient populations were in fact quite different. In this study by Welsh et al. , the patients with obstructive jaundice had a mean serum bilirubin of 356 (261–386) μmol/L, compared to 9 [7, 8, 9, 10, 11] μmol/L in their control group. With this marked difference in hyperbilirubinemia, they found only a small difference in intestinal permeability and some of the jaundiced patients had normal values. Although five (5.9%) of our patients with PBC were clinically icteric, the mean bilirubin for all patients was 15.9±12.7 μmol/L and only one of the jaundiced patients had abnormal permeability. This suggests that cholestasis alone does not explain the observed increase in intestinal permeability.
Patients with HCV infection had higher L:M ratios than healthy controls, reflecting increased distal small bowel permeability. Whether this is a consequence of liver disease in general or a specific viral factor remains to be seen. It is also possible that portal hypertension, which was seen equally in the PBC and HCV cohort, predominantly affects distal small bowel permeability and therefore accounts for the abnormalities seen in HCV but not in PBC.
External and non-liver-related confounders must also be considered. Our patients were not taking ASA, NSAIDs, or alcohol, all known to affect permeability, at the time the test was performed [28, 29]. Bile acids have been shown to reversibly increase jejunal and ileal permeability in rabbits, however, this has not been studied in humans . If UDCA had a major effect on human intestinal permeability, a greater proportion of the PBC cohort would likely have had abnormal results given that they were all on therapy. Smoking is associated with the development of PBC but has not been shown to affect intestinal permeability [31, 32]. Latent celiac disease was excluded by measuring EMAs in the PBC cohort and the two positive patients were excluded. Helicobacter pylori infection with associated gastritis was not evaluated, however, this is unlikely to account for the differences observed . A high rate of H. pylori infection has been documented in patients with hepatitis C , while studies of PBC have shown no increase in H. pylori prevalence [35, 36]. Therefore if H. pylori were a major putative factor contributing to abnormal permeability, one might expect higher permeability in the patients with hepatitis C than in those with PBC. Recent work has clearly demonstrated that interactions between luminal bacteria and epithelial cells can result in increased permeability by a number of mechanisms [37, 38]. It is possible that the intestinal flora differs between these groups of patients; unfortunately the study was not designed to address this issue. Finally, the healthy controls came from previous studies of intestinal permeability. They were not matched, and consequently, the patients with PBC were older and a higher percentage was female. Neither age nor gender has been shown to affect intestinal permeability [39, 40, 41].
In addition to greater passive movement across the mucosal surface under physiologic conditions, many patients with abnormal intestinal permeability display an exaggerated response to “permeability stress.” This has been demonstrated previously in patients with Crohn’s disease who were given a challenge with NSAIDs. Although they had abnormal intestinal permeability at baseline, the change observed after NSAID ingestion was out of proportion to the response seen in a control population . If abnormal permeability is a primary defect, such a situation may also exist in patients with PBC. Although many of the patients have “normal” permeability at baseline, perhaps they have increased sensitivity to environmental “permeability stressors” such as NSAIDs and alcohol. Exposure could lead to exaggerated increases in passive permeability and potential showering of the immune system with antigenic stimuli.
Increased intestinal permeability may be neither a complication nor a cause of PBC but rather another clinical manifestation. In 17% of patients with PBC, sialoadenitis or full-blown Sjogren’s syndrome coexists . Similarly to biliary epithelial cells in PBC, the salivary epithelial cells in such patients have been shown to express PDC-E2-like antigens which interact with AMAs [44, 45]. From this observation, it has been postulated that PBC is in fact a multisystem disease targeting epithelial surfaces . If small bowel and/or gastric epithelium also express PDC-E2-like antigens, they too may be targeted by the immune system, leading to damage and possibly increased intestinal permeability.
Patients with PBC have increased gastric and proximal small intestinal permeability compared to healthy and liver disease controls. In addition, patients with PBC have increased distal small bowel permeability compared to healthy controls. This finding lends support to the hypothesis that the AMAs found in PBC may develop as a consequence of abnormal intestinal permeability and subsequent increased exposure of bacterial antigens to the immune system.