The most common causes of cirrhosis are chronic liver diseases related to alcohol consumption [alcoholic liver disease (ALD)], hepatitis virus infection and obesity and/or other components of the metabolic syndrome [non-alcoholic fatty liver disease (NAFLD)]. Increasing evidence from animal studies indicates that the translocation of bacteria and bacterial products (e.g. endotoxin) from the intestinal lumen into the systemic circulation, resulting in endotoxemia, is a contributing factor in the pathogenesis of several chronic liver diseases by inducing inflammatory changes in the liver [3, 4]. In cirrhosis, bacterial translocation is considered to play an important role in the pathogenesis of its complications . In addition to alterations in the intestinal microbiota and the immune system, dysfunction of the intestinal epithelial barrier is an important mechanism contributing to enhanced bacterial translocation . Barrier dysfunction may be affected by putative aetiological factors, such as alcohol and obesity, as well as by characteristics of cirrhosis itself. Better understanding of the role of the intestinal epithelial barrier function in human chronic liver diseases and cirrhosis may provide further insight into the pathogenesis of cirrhosis and the development of its complications. The role of bacterial translocation, intestinal microbiota and the immune system [e.g. toll-like receptors (TLRs)] in chronic liver diseases and cirrhosis have recently been reviewed elsewhere [6-9].
The intestinal epithelial barrier
The intestinal epithelial barrier consists of multiple defence mechanisms, and can mainly be subdivided into an epithelial (i.e. the mucus layer and epithelial cells) and an immunological barrier (i.e. the epithelial secretions and immune cells). In this review, we focus on the epithelial barrier, in particular on the single layer of epithelial cells, which are connected to each other by junctional complexes. The epithelium on the one hand facilitates absorption of luminal nutrients, water and electrolytes, whereas on the other hand serves as a barrier to prevent translocation of potential harmful substances via transcellular and paracellular transport . Paracellular transport is regulated by the apical junctional complex, consisting of the tight junction (TJ) and the subjacent adherens junction (AJ) .
The TJs seal the paracellular space and form a selective barrier that allows transport via at least two pathways: a high capacity, charge selective pore pathway for small ions and uncharged molecules and a low-capacity leak pathway for larger molecules, regardless of charge . TJs are considered to be highly dynamic and open and close continuously in response to a variety of stimuli . They consist of several transmembrane proteins, such as occludin, members of the claudin family and junction adhesion molecules, as well as cytoplasmic plaque proteins, such as the zonula occludens proteins (i.e. ZO-1, ZO-2 and ZO-3), which connect the transmembrane proteins with the perijunctional actomyosin ring [11, 14]. Contraction of this ring is important in regulating paracellular permeability and is mainly mediated by activation of myosin light chain kinase (MLCK), which phosphorylates myosin II regulatory light chain (MLC) [11, 14].
The AJs primarily consist of the interaction between the transmembrane protein E-cadherin and the cytoplasmic protein β-catenin. Similar to the TJs, the AJs are connected with the perijunctional actomyosin ring and thereby may also play a role in regulating paracellular permeability .
Methods to assess the intestinal epithelial barrier
The intestinal epithelial barrier integrity can be assessed by evaluating the structure and function of the TJs. The morphological structure of the TJs can be studied by freeze fracture electron microscopy using intestinal biopsies. Hereby, the TJs are identified as a network of anastomosing linear fibrils or chains of protein particles, termed TJ strands, which seal adjacent cell membranes . The number of TJ strands was found to correlate positively with the integrity of the epithelium . The TJ structure can also be evaluated in thin sections of tissue or epithelial cells using transmission electron microscopy, where membrane fusions and increased electron density between adjacent cells can be observed . Using the aforementioned techniques, the intercellular space was found to be occluded and therefore the epithelial TJs were considered to be static. However, more advanced techniques using fluorescently labelled TJ proteins, such as fluorescence recovery after photobleaching (FRAP), have demonstrated that, for example, occludin and ZO-1 are highly mobile at the TJ [19, 20].
Furthermore, the expression of TJs and its associated proteins in tissue samples and cell cultures can be assessed on the protein and gene level by means of immunohistochemistry, Western blotting and quantitative polymerase chain reaction (qPCR) [21, 22].
In addition to structural data, functional analyses are also often used to evaluate intestinal epithelial barrier integrity. In vivo culture of intestinal epithelial cell monolayers or ex vivo mounting of live excised intestinal mucosa tissue in Ussing chambers with subsequent measurement of transepithelial resistance (TEER) and/or permeation of specific markers (e.g. FITC-dextran or sucralose) across the epithelium are widely accepted [23, 24]. TEER measures the overall epithelial ion fluxes, whereas permeation of most markers points to the paracellular leak pathway [12, 25]. Recently, three-dimensional culture of Caco-2 and T84 cells in synthetic extracellular matrix proteins have also shown to be suitable for the assessment of intestinal epithelial barrier function [26-28].
In vivo, the most widely accepted method to evaluate TJ function is analysing the urinary recovery of orally administered inert test markers. Combining two test markers that differ in size and transport (i.e. paracellular or transcellular) enables correction for differences in, for example, gastric emptying, intestinal transit, dilution by secretions, tissue distribution and renal function . Examples of these test markers include sugars (monosaccharides and disaccharides, i.e. rhamnose and lactulose respectively), polyethylene glycols (PEGs: 400, 1000, 4000) and radiolabeled chelates (51Cr-ethylenediaminetetraacetic acid (51Cr-EDTA), 99mTc-diethylenetriaminepentaacetic acid (99mTc-DTPA)). Because of their size, most of these markers reflect the leak pathway . TJ function can also be evaluated by measuring the presence of intraluminal substances, such as endotoxin and D-lactate, in the systemic circulation .
In addition to loss of TJ integrity, intestinal epithelial barrier dysfunction can result from damage to intestinal epithelial cells. Intestinal cell damage can be assessed non-invasively by measuring endogenous cytosolic enterocyte proteins in plasma or urine, such as intestinal fatty acid binding protein (I-FABP) . Intestinal biopsies provide the possibility for histological evaluation of intestinal epithelial cell damage and apoptosis.
Intestinal epithelial barrier dysfunction in chronic liver diseases and cirrhosis
Several studies have been published assessing intestinal permeability in patients with chronic liver diseases that may eventually progress towards cirrhosis as well as in patients with cirrhosis in comparison with healthy subjects by measuring the urinary excretion of orally administered test markers.
In Table 1, studies in patients with chronic liver diseases are presented, grouped by aetiology. Studies performed in chronic alcoholics, of which at least a subgroup had evidence of liver disease, showed a significant increased gastroduodenal and/or whole intestinal permeability compared with healthy subjects [31-33](Table 1).
Table 1. Studies comparing intestinal permeability in patients with chronic liver diseases vs. healthy controls, either as a primary or secondary outcome
|Reference||N of pt with CLD analysed|| Aetiology /degree of CLD||N of HC analysed||IP results (P-value)||Duration of urine collection|
| Alcoholic liver disease |
|Bjarnason et al.  1984|| ||Alcoholics without cirrhosis (degree np.)||34|| |
51Cr-EDTA (P < 0.001)
51Cr-EDTA (P < 0.02)
|Parlesak et al.  2000|| || |
Alcoholics with moderate LD
PEG 400 ns. (np), PEG 1500 + 4000 (P < 0.01)
PEG 400 ns. (np), PEG 1500 + 4000 (P < 0.01)
|Farhadi et al. 2010||69||Alcoholics with/without LD||49|| sucrose (P = 0.004)||5–12 h|
| Non-alcoholic fatty liver disease |
|Wigg et al.  2001||18||NASH||16||L/R ratio ns. (P = 0.37)||5 h|
|Farhadi et al.  2008|| || ||12|| |
L/M ratio ns. (np), sucralose ns. (np)
( sucralose after aspirin in NASH (P = 0.002)
L/M ratio ns. (np), sucralose ns. (np)
|Miele et al.  2009||35||NAFLDf||24||51Cr-EDTA (P < 0.001)||24 h|
|Volynets et al.  2012||20||NAFLD||10 (NAFLD-free)|| L/M ratio (P < 0.05)||6 h|
| Other chronic liver diseases |
|Resnick et al.  1990||33||Hepatobiliary disordersa||11||99mTc-DTPA ns. (P > 0.05)||24 h|
|Keshavarzian et al.  1999||10||HCV + PBCb||20||sucrose ns. (np), L/M ratio ns. (np)||5 h|
|Feld et al.  2006|| || ||101-155 15 new|| |
sucrose (P = 0.0001), L/M ratio (P < 0.0001)
Sucrose ns. (np), L/M ratio (P < 0.0001)
|Cariello et al.  2010||43||Chronic hepatitis (HCV/ALD/NASH)||134|| L/M ratio (P < 0.01)||5 h|
In patients with NAFLD, results are conflicting (Table 1). Two studies showed a significant increased small and whole intestinal permeability comparing 20 and 35 NAFLD patients, respectively, with healthy subjects [34, 35], whereas two smaller studies could not find significant differences [36, 37]. However, the latter study by Farhadi et al.  did observe a significantly increased 0–24 h and 6–24 h urinary excretion of sucralose (indicating whole intestinal and large intestinal permeability respectively) in non-alcoholic steatohepatitis (NASH) patients after an aspirin challenge.
In addition, several studies have assessed intestinal permeability in patients with chronic liver diseases as a result of hepatitis C virus infection (HCV) , primary biliary cirrhosis (PBC)  or mixed aetiologies [39-41]. Results were not always consistent and no definite conclusion can be drawn from these single studies. Furthermore, in most of them also patients with cirrhosis were included [38-40] (Table 1).
Studies assessing intestinal permeability in patients with cirrhosis are given in Table 2. Most studies found a significantly increased small intestinal [40-54], gastroduodenal [40, 54] and whole intestinal [32, 55-58] permeability in patients with cirrhosis when compared with healthy subjects, despite different methods used (e.g. test markers) and patients included (e.g. aetiology and severity) (Table 2). Studies investigating specifically the large intestine in cirrhosis are scarce; one recent study reported a significantly increased 5–26 h urinary excretion of sucralose in cirrhotic patients vs. healthy subjects .
As mentioned above, in most studies with cirrhotic patients, mixed groups of patients were included with regard to aetiology, severity of cirrhosis and the presence of complications. Most of the studies assessing intestinal permeability in relation to the severity of cirrhosis could not find significant differences among patients with Child A, B and C cirrhosis [41-44, 47, 53-56, 59, 60]. However, several of them observed a non-significant increase in permeability as the severity changed from Child A to B and C cirrhosis [41, 43, 44, 47, 56] and three others did find a significantly higher permeability in patients with Child C vs. those with Child A and B cirrhosis [50, 57, 61]. Overall, these observations indicate that an increased intestinal permeability may be more pronounced in the most severe degree of cirrhosis.
Studies have also compared intestinal permeability in cirrhotic patients with vs. those without complications, either as primary or secondary outcome (Table 3). Most studies did not observe a significant difference in cirrhotic patients with ascites vs. those without ascites [32, 41, 43, 54, 59, 60], but several of them performed these analyses as secondary outcomes only. Four well-designed studies (based on primary outcome and/or large numbers) did find a significantly higher intestinal permeability [47, 50, 57, 61]. This is further supported by studies showing a significantly increased intestinal permeability in patients with ascites, but not in patients without ascites when compared with healthy controls [47, 50, 56]. These findings therefore suggest that especially the presence of ascites is associated with changes in intestinal permeability in cirrhosis. Results on the association between intestinal permeability and spontaneous bacterial peritonitis (SBP) are difficult to interpret because of methodological differences, such as the time of assessing intestinal permeability, and the small number of patients with SBP [43, 46-48, 53, 55, 57, 60-62]. Contrasting results have been reported on the association between intestinal permeability and hepatic encephalopathy (HE) [50, 53, 59, 61].
Overall, there are indications for an increased intestinal permeability in patients with chronic liver diseases, especially in patients with ALD and possibly also in patients with NAFLD, but the number of studies is very limited, a variety of test markers is used and in some of them also patients with cirrhosis are included. The majority of studies in cirrhotic patients point to the presence of an increased intestinal permeability. Changes seem to be most pronounced in patients with advanced cirrhosis and/or complications, especially in those with ascites.
It is important to bear in mind that the evaluation of the intestinal epithelial barrier function by the urinary excretion of orally administered test markers has several pitfalls. The possibility of a delayed gastrointestinal transit, differences in volume distribution (for example as a result of the presence of ascites) and renal dysfunction in cirrhotic patients, for example, can affect intestinal permeability results when using a single test marker. The urinary excretion ratio of two test markers that differ in size and transport pathway will therefore more accurately reflect intestinal permeability in these patients. Furthermore, medication often used by cirrhotic patients such as NSAIDs or lactulose may influence test results.
Findings of increased endotoxin levels in patients with chronic liver diseases [32, 34, 37] and cirrhosis [52, 63, 64], and also the presence of bacterial DNA in blood and ascites in 30–40% of cirrhotics [65-67] support the relevance of the increased intestinal permeability observed in these patients. Direct measures of bacterial translocation in humans are too invasive and therefore scarce. One study has been published reporting a significantly increased prevalence of bacterial translocation, defined as the isolation of enteric organisms from mesenteric lymph nodes (MLNs) in patients with Child C cirrhosis compared with Child A and B cirrhosis .
The above-mentioned findings of an increased intestinal permeability suggest an impairment of TJs. However, studies assessing TJ structure in patients with chronic liver diseases and cirrhosis are limited. A small study by Tang et al.  showed that patients with ALD had a significantly reduced protein expression of ZO-1 in sigmoid biopsies compared with healthy subjects, which was accompanied by a significantly higher miR-212 expression in ALD patients. The mRNA expression of ZO-1, however, was not significantly different between both groups. Protein expression of ZO-1 was also found to be significantly lower in duodenal biopsies of NAFLD patients than in healthy subjects . In addition, in this latter study intestinal permeability was increased.
Cirrhotic patients with and without ascites were studied by Assimakopoulos et al. . They showed a significantly reduced expression of the TJ proteins occludin and claudin-1 in duodenal biopsies of the total patient group compared with healthy controls and this correlated inversely with the increased plasma endotoxin levels. In addition, the cirrhotic patients with ascites showed a significantly reduced expression of occludin and claudin-1 compared with those without ascites. Recently, increased duodenal protein levels of claudin-2, and a decreased TEER and increased permeability was detected in duodenal biopsies mounted in Ussing chambers from patients with decompensated cirrhosis . However, no differences in mRNA and proteins levels of occludin, claudin-1, ZO-1 and connexion-43 were found between the cirrhotic patients and controls. Application of electron microscopy in this study, further showed an intact epithelial barrier in patients with decompensated cirrhosis, suggesting that the epithelial barrier in cirrhosis is functionally altered, but structurally normal . In contrast with the latter finding, Such et al.  found distended intercellular spaces with morphologically intact TJs in duodenal biopsies of six cirrhotic patients with complications.
An increased intestinal permeability may also be related to epithelial cell damage. Histological changes in the intestinal mucosa have been reported in ALD patients, such as villous atrophy, increase in lamina propria infiltrate and intraepithelial lymphocytes, as well as changes in the cellular functions (i.e. brush border, membrane and cellular enzymes) . Changes were also found in the intestinal mucosa of cirrhotic patients with portal hypertension and included shortened and wider microvilli, a decreased villus/crypt ratio, oedema of lamina propria, fibromuscular hyperplasia, thickened muscularis mucosae, increased apoptosis and capillary abnormalities [71, 73-75].
Although differences in TJ structure and expression supporting functional permeability analyses have been demonstrated in cirrhotic patients, overall conclusions on specific TJ proteins or subgroups of patients cannot be drawn as a result of methodological differences and the relative small number of studies.