Bacterial translocation (BT) in cirrhosis

Abstract

Gut flora and bacterial translocation (BT) play an important role in the pathogenesis of the complications of cirrhosis. Research on the pathogenesis of BT and its clinical significance transcends established boundaries between microbiology, cell biology, intestinal pathophysiology, and immunology. This review delineates multiple mechanisms involved in the process of BT, with an emphasis on alterations in intestinal flora and mucosal barrier function, particularly immunological defense mechanisms. Current knowledge on the innate and adaptive immune response that allows a “friendly” communication between bacteria and host is summarized, and alterations occurring in cirrhosis that may facilitate BT are discussed. In addition, definition of a “pathological” BT is proposed together with an analysis of the anatomical site and route of BT. Finally, therapeutic approaches for the prevention of BT in experimental and human cirrhosis are reviewed. Future research in the field of BT in cirrhosis will allow the development of new therapeutic targets in the prevention of infections and other complications of cirrhosis. (HEPATOLOGY 2005;41:422–433.)

Humans and intestinal bacteria have developed an adaptive commensal relationship, supported by the synergistic interplay of multiple intestinal defense mechanisms, including luminal factors, inhibition of mucosal attachment, prevention against penetration, and immunological clearance mechanisms (Fig. 1). In health, this relationship has important effects on immune function and nutrient processing. Bacterial translocation (BT), defined as the migration of bacteria or bacterial products from the intestinal lumen to mesenteric lymph nodes (MLNs) or other extraintestinal organs and sites,1 represents a disruption of the normal host/flora equilibrium that leads to a self-perpetuating inflammatory response and, ultimately, to infection. BT has been postulated as the main mechanism in the pathogenesis of spontaneous infections in cirrhosis as well as the hyperdynamic circulatory state, a key factor in the pathogenesis of portal hypertension and in the development of ascites and hepatorenal syndrome.2 Understanding the physiology of gut–bacteria interactions and the pathogenesis of BT could lead to new therapeutic targets in the prevention of infections and other complications of cirrhosis. Clinical studies of BT in cirrhosis are hampered by the lack of noninvasive and sensitive methods to detect its presence. Therefore, most investigations of BT in cirrhosis have been performed in experimental animals, in which BT is defined by the presence of positive bacteriological cultures of MLN.

Figure 1. Intestinal mucosal barrier and mechanisms of bacterial translocation (BT). Factors involved in the process of BT include changes in the intestinal microflora, alterations in mechanisms preventing attachment and penetration of bacteria, and, finally, local and systemic immune responses aiming to kill bacteria that have translocated.

BT, bacterial translocation; MLN, mesenteric lymph node; IgA, immunoglobulin A; TJ: tight junction; LPS, lipopolysaccharide; GALT, gut-associated lymphatic tissue; DC, dendritic cell; TLR, toll-like receptor; NFκB, nuclear factor κB; TNF, tumor necrosis factor; APC, antigen-presenting cell; CARD15/NOD2, caspase-activating and recruitment domain-15; NO, nitric oxide; iNOS, inducible nitric oxide synthase; PCR, polymerase chain reaction; bactDNA, bacterial DNA.

Microflora

Microbial flora and human health are inextricably linked. The gastrointestinal tract, the largest surface area of the body, is constantly exposed to microorganisms. Their peaceful coexistence implies the presence of clearly defined lines of communication. The intestinal microflora consists of a dynamic mixture of microbes with a different composition across the gastrointestinal tract and considerable quantitative and qualitative differences among individuals and particularly among species, a fact that should be recognized when translating results from animal studies to a clinical setting. The upper gastrointestinal tract is sparsely populated with bacteria; however, from the ileum on, a sharp increase occurs in microbial density, from 10⁵ colony-forming units/mL in the jejunum to 10⁸ in distal ileum and cecum, and up to 10¹² in the colon.3

Bacteria that translocate most readily are facultative intracellular pathogens (e.g.,Salmonella, Listeria) that resist phagocytic killing to a certain degree. In contrast, normal enteric species are easily killed after phagocytosis, surviving only when host defenses are impaired. Gram-negative bacteria (specifically Escherichia coli, Klebsiella pneumoniae, and other Enterobacteriaceae), enterococci, and other streptococci have been found to be the most adept at translocating to MLN.4 Interestingly, these species are those that most frequently cause infection in patients with cirrhosis.5 Special strains of E. coli have been shown to translocate more efficiently, probably as a result of a greater ability to adhere to the intestinal mucosa.6 Moreover, differences in virulence among strains may lead to greater resistance against host defense mechanisms, allowing for a more efficient survival and dissemination.7, 8

Intestinal anaerobic bacteria outnumber aerobic bacteria by 100:1 to 1,000:1, despite which anaerobes very rarely translocate.4 In contrast to aerobic gram-negative bacteria, which translocate easily and even across a histologically intact intestinal epithelium,9–12 anaerobic bacteria seem to translocate only in conditions associated with intestinal mechanical injury (e.g., athymic,13 lethally irradiated,14 or severely burned rodents10). Moreover, anaerobic bacteria limit the colonization and overgrowth of other potentially invasive microbes, thereby confining potentially pathogenic bacteria. In fact, selective elimination of anaerobic bacteria facilitates intestinal bacterial overgrowth and translocation of facultative bacteria.12 Bacterial overgrowth is one of the main factors that promote BT. A direct relationship between numbers of a specific bacterial strain populating a segment of the intestine and numbers of viable bacteria of this strain present in MLN has been demonstrated in mice,15 particularly when adherent bacteria are involved.16 Such bacterial overgrowth occurs in cirrhosis and has been linked to the development of BT, spontaneous bacterial peritonitis, and endotoxemia,17–19 and has been attributed, at least partly, to a decrease in small-bowel motility and intestinal transit time.20–24

Intestinal Mucosal Barrier

The intestinal mucosal barrier includes both secretory and physical preventive measures against the penetration of microbes. Mucins, secreted by epithelial goblet cells in large amounts (3 L/d), create a thick (400-500 nm) layer of membrane-anchored negatively charged glycoproteins that prevents direct contact of bacteria with the microvillus membrane.25 Mucosal secretions are rich in immunoglobulin A (IgA) antibodies that effectively bind and aggregate bacteria preventing mucosal adherence and colonization (immune exclusion).26 IgA accounts for more than 70% of total body immunoglobulin production (>5 g/d). This secretory mucosal IgA not only prevents microbial entry but also neutralizes toxins and infectious organisms and may actively transport IgA-bound antigens/microbes as a sort of “sump pump” from the lamina propria back to the lumen. As described later, IgA is secreted by plasma cells derived from B cells that “home” to the lamina propria. Bile inhibits bacterial overgrowth, has a trophic effect on the intestinal mucosa,27 decreases epithelial internalization of enteric bacteria,28 exerts detergent actions with anti-adherence effects, and binds and neutralizes endotoxins.29, 30 Therefore, the absence of bile in the intestine has been shown to facilitate BT31–34 and to enhance endotoxin-induced BT.35 Studies of BT in obstructive jaundice have been controversial,32, 36, 37 and although a study showed a higher incidence of BT in such patients,38 it did not correlate with the development of systemic infection. In cirrhosis, decreases in intestinal intraluminal concentrations of bile acids have been ascribed to decreased secretion and increased deconjugation by enteric bacteria.

The most critical barrier is the epithelium per se, through its lack of permeability and active production of antimicrobial peptides and proteins. Specialized cell-cell junctional complexes allow selective paracellular permeability pathways (tight junctions [TJs]), while maintaining intercellular adhesion (intermediate junctions and desmosomes) and permitting intercellular communication (gap junctions). At the apicolateral epithelial surface, TJs maintain a permeability seal restricting paracellular movement of even very small (2 kd) molecules, thereby preventing the transepithelial movement of not only bacteria, but also macromolecules such as lipopolysaccharide (LPS). In cirrhosis, structural changes of the intestinal mucosa including widening of intercellular spaces, vascular congestion, edema, fibromuscular proliferation, decreased villous/crypt ratio, thickened muscularis mucosae, and inflammation have been described.39–41 Ultrastructurally, dilated extracellular spaces between neighboring enterocytes and a reduced number of microvilli have been noted, although TJs in distal duodenum are unaltered in patients with advanced cirrhosis.42 More importantly, functional studies have demonstrated increased intestinal permeability in patients with cirrhosis,43–46 mainly in those with sepsis.43 Intestinal mucosal oxidative damage, evidenced by increased lipid peroxidation and altered enterocyte mitochondrial function,47 as well as endotoxemia and increased levels of pro-inflammatory cytokines and nitric oxide,48, 49 may play a role in mediating this increased permeability.

As part of the innate immunity, several antimicrobial molecules have been identified and function as natural antibiotics because of their ability to kill a variety of microbes under laboratory conditions.50 Paneth cells are a cluster of cells strategically located at the bottom of each intestinal crypt just below the stem cell zone, predominantly in jejunum and ileum. These cells secrete α-defensins in response to bacteria and LPS51, 52 as well as other larger proteins such as lysozyme53 and secretory phospholipase A254 that may be involved in local defense against commensal bacteria. Additionally, antimicrobial peptides such as human β-defensin 1 appear to be expressed by most epithelial cells of the small and large intestine.55 No studies have been published regarding the epithelial release of these antimicrobial agents in cirrhosis and particularly in BT.

Regarding the anatomical site of BT in experimental cirrhosis, histological changes are most marked in the cecum,56 where bacterial species known to translocate, such as E. coli and enterococci, are present in large numbers.3 Although this would suggest that the preferred site for BT might be cecum or colon, it has also been suggested that, because of its constant exposure to large numbers of bacteria, the colon is more efficient at eliminating translocating bacteria. Moreover, colonic epithelia has higher electrical resistance and lesser permeability to the passive movement of ions than the small bowel.57 Experimental studies inoculating equal concentrations of E. coli into small or large bowel show that bacteria translocate predominantly from the small bowel.58 This is supported by studies showing that proximal gut colonization is associated with increased BT and septic morbidity in surgical intensive care patients.59, 60 Moreover, in cisapride-treated animals with an increased intestinal transit, reduction in BT rates was observed to be associated with lower jejunal, but not cecal, bacterial counts.19

Communication Between Intestinal Bacteria and Host

Bacteria can normally be detected in underlying intestinal tissue without associated injury because organisms are usually efficiently removed by phagocytes.36 The intestinal tract is an active immune organ, containing essentially every type of leukocyte involved in immune response. The “gut” mucosal immune system consists of the gut-associated lymphoid tissue (GALT), the largest immunological organ of the body, which comprises four lymphoid compartments: Peyer's patches, lamina propria lymphocytes (including dendritic cells [DCs]), intraepithelial lymphocytes, and MLN. Microbial colonization of the gastrointestinal tract affects GALT composition. Exposure to luminal microbes leads to expansion of intraepithelial lymphocytes, appearance of germinal centers within follicles and lamina propria, and a rise in serum Ig levels. In contrast, germ-free animals have reduced numbers of lamina propria or intraepithelial T cells61 that increase after restitution of the normal microflora.62 Intestinal commensals interact with the gut epithelium and trigger both innate and adaptive immune responses.

Innate Immunity

This ubiquitous antigen nonspecific immune function is determined by recognition of bacteria/bacterial products and release of chemokines and cytokines, leading ultimately to bacterial killing by mononuclear cells. In the gut mucosa, monocytes and particularly DCs are in charge of providing broad nonadaptive (innate) protection against microorganisms. Activation of innate host defense depends on specific recognition of highly conserved microbial signature molecules called pathogen-associated molecular patterns, recognized by most mononuclear cells by pattern recognition receptors.63 These include the family of toll-like-receptors (TLRs), of which TLRs 2, 4, and 9 are the most relevant among its 11 members (Table 1). Stimulation of TLRs by PAMPs leads to activation of signaling pathways that result in activation of nuclear factor kappa B (NFκB)-, mitogen-activated protein (MAP)-kinases, IRF-3 etc. and transcription of inflammatory cytokines, chemokines, and antimicrobial genes. In response to bacterial components, gut epithelial cells release chemokines that recruit DCs to the mucosa, where they sample microbial antigens and return to lymphoid follicles deeper in the gut mucosa or in the MLN, where they present antigens to B and T cells of the adaptive immune system (see later discussion). The activation of DCs by microbial products through pattern recognition receptors ensures that DCs will not arrive empty-handed to the lymphoid follicles or MLN and provides a link between the innate and adaptive immune systems.63, 64

Table 1. Toll-like Innate Receptor (TLR) Repertoire

PAMP (ligand)	Origin/Pathogen	PRR (receptor)
LPS: lipopolysaccharide LTA: lipoteichoic acid
LPS	GNB	TLR 4. CD14. LBP
LTA	Gram ± ve bacteria	TLR 2/1
Peptidoglycan	Gram ± ve bacteria	TLR 2/6
Lipoprotein	Eubacteria	TLR 2/6 or TLR 2/1
HSP 60	Host	TLR 2/4
Unmethylated CpG-DNA	Most bacteria	TLR 9
Flagelin	Many bacteria	TLR 5
DsRNA	Viruses	TLR 3

Innate Immunity in Cirrhosis.

Studies performed in peripheral blood mononuclear cells of patients with cirrhosis show that TLR 2 expression is upregulated, whereas TLR 4 expression is unaltered or downregulated.65, 66 TLR 2 responds predominantly to cell wall components of gram-positive bacteria,63 and TLR 4 is the sole signal-transducing receptor for LPS in humans, suggesting an important stimulatory role for gram-positive but not for gram-negative bacteria.66 However, LPS administration in healthy volunteers increases TLR 2 expression on peripheral blood mononuclear cells both in vivo and ex vivo.67, 68 Moreover, gram-negative bacteria possess components other than LPS, such as bacterial DNA, lipoproteins and heat-shock protein 60, which stimulate different TLRs. All known TLR agonists have been shown to induce tumor necrosis factor (TNF) secretion by monocytes. Not surprisingly, activation of peripheral mononuclear cells with a marked increase in LPS-induced TNF expression has been reported in patients with cirrhosis and ascites.69–71 Moreover, increased TNF production by MLN has also been reported in end-stage cirrhosis and has been associated with an increased incidence of bacterial infections after liver transplantation.72 It is tempting to speculate that in cirrhosis, priming of mononuclear cells and associated release of pro-inflammatory cytokines is mediated by an upregulation of TLR expression.

Conversely, bacterial recognition by TLRs does not necessarily imply bacterial killing or neutralization. In fact, despite the stated activation of mononuclear cells in advanced cirrhosis, innate immune response is impaired in cirrhosis because of reduced phagocytic and killing capacity.73, 74 Decreased receptor-independent (intrinsic) phagocytosis, particularly for E. coli, and decreased opsonic activity have been postulated as mediators of this cellular dysfunction. Impaired tuftsin activity, known to modulate biological activities of phagocytic cells, is reduced in patients with cirrhosis and is associated with a higher incidence of bacterial infections.75 Reduced complement levels as well as defects in Fcy-receptors on mononuclear cells may contribute to decreased opsonic activity.76–78 In patients with cirrhosis, low ascites complement levels correlate with decreased opsonic activity, decreased bactericidal activity and increased risk for SBP.79, 80 Also, low serum C3 levels independently predict infection and correlate with a poor liver synthetic function.81 Interestingly, peripheral blood mononuclear cells but not resident spleen mononuclear cells exhibit a normal Fcy-receptor function and expression in vitro,82 emphasizing the importance of distinguishing between systemic/circulating and local/resident cellular mechanisms of innate immunity. One could speculate that, in cirrhosis, alterations in TLR expression and phagocytic capacity at the local GALT level may lead to increased BT.

Adaptive (Acquired) Immunity.

Luminal bacterial antigens are presented to DCs by two mechanisms: indirectly by M cells, which are specific cells in the epithelial layer that reside over lymphoid follicles and endocytose luminal antigens transporting them to local DCs and macrophages (antigen-presenting cells [APCs]). Or directly, by local APCs that take up luminal antigens by extending pseudopods between epithelial cells. APCs present antigenic peptides from captured microbes to B and T lymphocytes either locally (in the intestine) or in the MLN (to which they migrate via afferent lymph vessels). APCs determine the class of immune response by triggering naïve Th-cells to develop into either effector Th1-, Th2-cells or a mixed phenotype.83 These “primed” T cells are released into the bloodstream and “home” back through the gut vascular endothelium to populate the lamina propria.

Presentation of microbial antigens to B cells triggers production of a commensal-specific mucosa-protective IgA (or IgG) response. It also appears that activation of intestinal T cells is critical in maintaining humoral immunity against the translocation of enteric bacteria.84 Mechanisms by which T cells help maintain the mucosal barrier function include T-cell–dependent B-cell activation and production of chemokines/cytokines by which macrophages and neutrophils are recruited and activated. T cells exposed to antigen in Peyer's patches also migrate into the lamina propria and the epithelium, where they mature into cytotoxic T cells, providing another mechanism for containing microbial assaults. Despite increased BT in IgA-deficient mice,85 commensal-related sepsis is not observed in IgA-deficient animals or humans. In patients with cirrhosis, decreased fecal IgA concentrations have been described, implying decreased secretion of mucosal IgA.86 However, its relationship to BT and development of infections in cirrhosis requires further investigation.

Commensal bacteria transported to MLN by DCs produce a local immune response and are normally killed in situ by mononuclear cells, without inducing systemic immunity or intestinal inflammation.85, 87 Surgical removal of MLN leads to invasion of the spleen by commensal bacteria, triggering a systemic immune response.87 Similarly, immunosuppression allows translocated bacteria to spread systemically, ultimately leading to sepsis.88 Therefore, BT can only become clinically significant if a failure of local or systemic immune defense mechanisms occurs. Isolation of bacteria from MLN can actually represent a “normal” process,37, 149 and inability of MLN to contain bacterial infection would represent a “pathological” type of translocation, evidenced by the isolation of bacteria from sites other than the MLN. Spontaneous BT of bacteria, such as E. coli, to spleen and liver has been noted in athymic (nu/nu) mice, whereas no translocation was noted in heterozygous (nu/+) or nude mice grafted with thymus,13 indicating the important role of T cells. Moreover, T-cell depletion causes accumulation of bacteria in MLN and spread of bacteria to extraintestinal sites.89 T-cell depletion has been noted in patients with advanced cirrhosis.69

Efficient killing of commensals penetrating the mucosa normally prevents an inflammatory response. However, if mucosal LPS load becomes overwhelming, DCs and other monocytes and neutrophils are recruited, perpetuating BT90, 91 (Fig. 2). Epithelial permeability increases with migration of neutrophils across epithelial surfaces associated with cytokine release, particularly of TNF and interferon gamma.92 In experimental cirrhosis, BT is associated with mononuclear cell infiltrate in the lamina propria as well as severe submucosal and mesenteric inflammation, particularly at the cecal level.41, 56, 93 Although it is uncertain whether inflammation is the cause or the result of BT, in animal models of severe intestinal inflammation viable bacteria appear in the portal circulation even before they appear in efferent intestinal lymph,94 suggesting that inflammation may influence the route of translocation by damaging intercellular associations and allowing access to submucosal capillaries.

Figure 2. Physiological DC-mediated transport of bacteria by the GALT and alterations in BT and inflammation. Commensal bacteria in the intestinal lumen are continuously sampled by cells of the innate immune system, e.g., DCs and M cells via various routes: (1) DCs may open TJ between epithelial cells, sending processes into the lumen that directly phagocytose microbes; this occurs without compromising the epithelial barrier function because of expression of major TJ proteins by DCs; (2) indirectly, through uptake of exosomes shed from the epithelium; (3) through interaction with antigenic material in underlying tissue that occurs particularly when epithelial integrity is compromised; or (4) through sampling by M cells that deliver luminal antigens to underlying lymphoid tissue. Intracellular microbes/microbial products are then transported to and enter the draining MLN through the subcapsular sinus, ending up in the T-cell area of the paracortical region, where they are thought to activate antigen-specific T cells. DCs also present antigenic microbial peptides to B and T lymphocytes in the submucosa. Presentation of antigens to B cells triggers production of a commensal-specific IgA response. DCs and probably other GALT monocytes are primed by microbial products and migrate through lymphatics and MLN before eventually entering the systemic circulation. In case of increased (“pathological”) BT, recruitment of other DCs and mononuclear cells occurs with resultant release of pro-inflammatory cytokines and inflammation, leading to a vicious cycle that perpetuates the process of BT. Modified from Krahenbuhl and Corbett.148

In this context, genetic polymorphisms related to nucleotide-binding and leucine-rich repeat proteins, involved in intracellular recognition of microbes and their products, namely, the caspase-activating and recruitment domain-15 (CARD 15/NOD2) gene, have been implicated in the pathogenesis of mucosal inflammation in Crohn's disease95, 96 and in gastrointestinal graft-versus-host disease,97 conditions associated with increased BT. Bacterial (or bacterial products such as peptidoglycan) sensing by NOD2 triggers the activation of pro-inflammatory pathways such as NFκB.98, 99 Thus, NOD2-associated mutations can lead to insufficient activation of NFκB and have been associated with “autoinflammatory” disorders.99 Insufficient activation of NFκB could result in deficient elimination of translocated bacteria as well as further enhancement of BT via multiple secondary effects (Fig. 3). The presence of BT susceptibility genes in cirrhosis is as yet unknown.

Figure 3. Model of NOD2-dependent changes in epithelial and mononuclear defense mechanisms. NOD 2 is expressed in epithelial cells as well as underlying antigen-presenting cells (e.g., immature DCs) and other mononuclear cells. “Sensing” of bacteria through this intracellular receptor leads to activation of NFκB, or in the case of NOD2 mutations, to a dysregulation of NFκB resulting in deficient elimination of translocated bacteria or increased “pathological” inflammation. Bacteria or bacterial products stimulate extracellular TLRs (innate immune response) and, hence, maturation of DCs, triggering the release of cytokines. These in turn upregulate NOD2 and TLR expression, stimulate the production of β-defensins, and activate intestinal T and B cells (adaptive immune response). Modified from Chamaillard et al.99

Role of Nitric Oxide

Nitric oxide (NO) not only plays an important role in the pathogenesis of portal hypertension and the hyperdynamic circulatory state of cirrhosis100 but also has been implicated in bacterial elimination and gut integrity. However, its role in BT and cirrhosis is complex and may depend on factors such as the stage of liver disease, site of NO production, and concomitant hemodynamic alterations. NO modulates macrophage function, cytokine release and bactericidal killing capacity.101, 102 Antimicrobial activity is severely impaired in inducible nitric oxide synthase (iNOS)/phagocyte oxidase double knock-out mice, leading to massive abdominal abscesses and sepsis.103 In experimental cirrhosis, iNOS expression by peritoneal macrophages is increased, and its inhibition results in peritoneal infection.104 Peritoneal macrophages isolated from patients with unresolved or resolved spontaneous bacterial peritonitis (SBP) produce NO and express iNOS, suggesting that NO may contribute to the control of SBP.105 However, low ascites NO levels, although predictive of a poor prognosis, are unable to predict SBP in patients with cirrhosis.106 NO also participates in maintaining gut barrier function.107 It increases gastrointestinal mucus secretion, modulates epithelial chloride transport and associated fluid secretion, maintains blood flow, inhibits intestinal muscular motor activity, reduces mast cell reactivity and mediator release, suppresses neutrophil aggregation and sequestration, and scavenges reactive oxygen metabolites.102 Additionally, NO influences DC maturation and differentiation and, hence, may have an impact on bacterial sampling and initiation of adaptive immune responses.108 NO has been shown to modulate intestinal permeability and BT in different ways and directions. Nonspecific inhibition of NO synthesis results in increased ileal epithelial permeability,109 and administration of an NO donor ameliorates gut mucosal hyperpermeability induced by ischemia–reperfusion or endotoxin.110 Conversely, NOS inhibition has been found to aggravate barrier dysfunction induced by ischemia–reperfusion, LPS, or other noxious stimuli.107, 109, 111 Thus, endogenous NO appears to regulate normal mucosal barrier, and its removal allows injury to occur at a greater degree. On the other hand, vast overproduction of NO has been shown to impair the integrity of the intestinal epithelium. At high concentrations, NO induces gastric mucosal damage, decreases the viability of rat colonic epithelial cells,112, 113 dilates TJs, disrupts the actin cytoskeleton, inhibits adenosine triphosphate formation, and, hence, increases intestinal permeability.114, 115 Endotoxemia, hemorrhagic shock, ischemia–reperfusion, or thermal injury appear to lead to BT through iNOS-derived NO production,116, 117 in support of which iNOS knock-out mice exposed to LPS lack BT.118 However, BT was not abolished in these mice in conditions of intestinal bacterial overgrowth, pointing at the complexity of the relationship between NO and BT.

Identifying Bacterial Translocation

In experimental studies, BT is defined as any positive MLN bacteriological culture. In these studies, the whole chain of MLN is dissected, homogenized, and cultured. Studies of BT in humans are limited because of the need for surgery and the removal of only one MLN in conditions that may alter the results (e.g., perioperative antibiotics). Although the rate of positive MLN cultures appears to be higher in patients with Child C cirrhosis, BT has not been predictive of the development of postoperative infections.119 Alternative approaches to diagnosing BT in humans have been postulated. TNF levels in MLN seem to have a better correlation with Child status and with the development of bacterial infections72; however, this method still requires surgery. Over the last couple of years, polymerase chain reaction (PCR)-based detection of bacterial DNA (bactDNA) has been proposed as a surrogate marker for BT because it has been detected in blood and ascites of approximately one third of patients with cirrhosis and culture-negative ascites.120 Lending validity to this test, sequential testing shows that bactDNA from subsequent samples was identical to the one detected in a first sample,121 and although the presence of bactDNA did not correlate with more severe liver disease or more altered hemodynamics, it was recently found to correlate with a higher synthesis of NO by peritoneal macrophages and higher cytokine production.122 However, methodological concerns remain, particularly because only E. coli DNA was identified, and bactDNA was undetectable in controls, remarkable in light of the detection of a variety of bacterial species even in healthy volunteers,123, 124 a result inherent to a technique that amplifies eubacterial PCR bactDNA contamination in the pre-PCR workup or at the time of DNA extraction.125 Further studies are needed to validate this and other modalities, such as green-fluorescence–marked bacteria, for detecting BT.

Manipulation of Gut Flora and Its Effect on BT in Cirrhosis

Most of the currently tested therapies geared at preventing BT in cirrhosis are aimed at decreasing bacterial overgrowth or changing the composition of gut flora, elements that are more easily modifiable than other factors such as increased intestinal permeability and decreased immunity.

Selective Intestinal Decontamination.

Antibiotics that selectively eliminate intestinal gram-negative bacteria such as norfloxacin have been shown to produce a marked reduction in fecal gram-negative bacteria of patients with cirrhosis, without significant effects on gram-positive cocci or anaerobic bacteria.126 Experimental studies examining the effect of oral norfloxacin or trimethoprim/sulfamethoxasole on BT are controversial; although some show a decrease in BT (particularly by gram-negative bacteria),127–129 others show no effect.130, 131 Nevertheless, selective intestinal decontamination has been shown to be effective in preventing bacterial infections in patients with gastrointestinal hemorrhage132, 133 and low ascites protein,134, 135 as well as preventing SBP recurrence.126 Interestingly, preliminary studies have shown that selective intestinal decontamination appears to ameliorate the hyperdynamic circulatory state of cirrhosis.129, 136–138 Unfortunately, long-term antibiotic prophylaxis has been associated with the development of antibiotic-resistant infections,135, 139 necessitating the search for non-antibiotic methods to eliminate bacterial overgrowth and BT.

Prokinetics.

Bacterial overgrowth in cirrhosis is at least partly secondary to decreased intestinal motility probably because of increased adrenergic activity. Propranolol, a β-adrenergic blocker, shortens intestinal transit time in rats with cirrhosis and has been shown to decrease bacterial overgrowth and BT.140 However, no differences in the rate of ascites infection were observed between treatment groups. Cisapride, an intestinal prokinetic, also has been shown to shorten intestinal transit time, eliminate bacterial overgrowth, and decrease BT in experimental cirrhosis.19, 141 In patients with cirrhosis, a 6-month course of cisapride led to a decrease in orocecal transit time and elimination of bacterial overgrowth,19, 142 with a tendency for a lower incidence of infections.142 These results are encouraging and should stimulate further studies using prokinetics.

Probiotics and Others.

Probiotics have been proposed as a means of re-equilibrating gut flora in favor of protective anaerobic bacteria. Bacteriotherapy with Lactobacillus has been reported to correct bacterial overgrowth, stabilize mucosal barrier function, enhance host defenses, and hence, to decrease BT in rat models of acute liver injury and failure.143 However, experimental studies using different lactobacilli-fermented diets in animals with portal hypertension and cirrhosis failed to reduce the rate of BT.130, 144 Patients with liver disease treated with a combination of probiotics (Lactobacillus plantarum) and prebiotics (fiber) had a lower rate of postoperative bacterial infections than those treated with selective intestinal decontamination.145 In animal models, the combination of probiotics and antioxidants suppressed BT to the MLN, reduced ileal and cecal counts of Enterobacteria and enterococci as well as intestinal malondialdehyde levels (used as an index of intestinal oxidative damage).146 Conversely, administration of a synbiotic (pre- and probiotic) gram-positive gut flora regimen to patients with cirrhosis led to further increases in peripheral mononuclear cell expression of TLR2 and circulating TNF levels, the significance of which remains to be determined.66 Considering the stated effects of bile acids, another approach is to administer conjugated bile acids orally, such as cholylsarcosine and cholylglycine. In rats with cirrhosis, such compounds eliminated bacterial overgrowth and reduced in the rates of BT and endotoxemia, as well as improved survival.147 However, there is a clear need for randomized controlled trials evaluating measures to prevent BT in humans.

Conclusions and Future Research

BT plays an important role in the genesis of infections and hemodynamic complications of cirrhosis; however, it is not a universally “pathological” phenomenon. With advancing knowledge of the physiology of gut microbial/host interactions, it becomes apparent that, in normal conditions, BT to MLN is a physiological event with important immunological functions. A “pathological” BT would be one associated with local/systemic inflammatory response or spreading of bacteria beyond MLN. Studies of local intestinal immunity, a critical element in determining whether BT remains normal or becomes pathological, are mostly lacking in cirrhosis, and it is our impression that alterations in innate and adaptive immunity play a key role in this process. The local mucosal immune system in cirrhosis may be incapable of limiting bacteria from translocating to the underlying tissue or to MLN and, conversely, it may release increased amounts of pro-inflammatory cytokines, leading to inflammation and perpetuating BT. Studies on TLR expression, migration, and chemotaxis as well as phagocytic and killing capacity of GALT-associated mononuclear cells in cirrhosis are lacking and are encouraged. Moreover, factors that may greatly increase BT across the mucosa include relative deficiencies in antimicrobial peptides as well as gene mutations that may increase the susceptibility to BT and hence should be investigated in cirrhosis. Importantly, information summarized in this review is almost exclusively based on animal studies and ex vivo investigations. Data in humans are scarce and limited by the lack of a method that will detect the presence and extent of BT. Therefore, improvements and standardization of microbiological techniques are needed to enable a valid definition of “pathological” BT in humans against which therapeutic strategies could be directed.

Acknowledgements

The authors thank Dr. G. Rogler (Department of Internal Medicine I, University of Regensburg, Germany) for valuable discussions on mucosal immunity. We also gratefully acknowledge Dr. U. Reischl for his help in preparing the section on diagnosis of the manuscript.