The gut microbiome and liver cancer: mechanisms and clinical translation

2.1. ALD

ALD contributes to about half of all cirrhosis cases⁴¹ and is a cofactor in liver disease induced by HBV, HCV and NASH. Although ALD might have a lower relative risk of causing HCC than other types of CLD³⁹, the sheer number of patients with alcoholic cirrhosis means that the absolute number of HCCs caused by ALD is high. Moreover, subgroups of patients, such as those with cirrhosis, men, patients >55 years of age, individuals positive for antibodies against hepatitis B core protein (anti-HBc) as well as patients with high cumulative consumption of alcohol, might have extremely high risk development for HCC development, risk that may be >40% in a 10-year period⁴².

The key contribution of the gut microbiota to early stages of ALD has been firmly established in the past two decades. Even a single binge of alcohol is sufficient to increase bacterial translocation, as evidenced by an increase of LPS in portal blood rats from undetectable levels to 30–80 pg/ml after ethanol administration ¹⁶. Likewise, serum LPS levels are increased in patients with chronic alcohol abuse¹⁶. The ability of ethanol and its metabolite acetaldehyde to disrupt tight junctions contributes to the high levels of bacterial translocation in ALD⁴³. Moreover, mice receiving intragastric alcohol feeding show perturbations of the intestinal microbiota, with reduced synthesis of long-chain fatty acids⁴⁴. A number of functional studies have shown a key contribution of the gut-microbiota–TLR4 axis to ALD⁴⁵: Global TLR4 deficiency in mice as well as gut sterilization with nonabsorbable antibiotics in rats reduces hepatic steatosis, oxidative stress and inflammation^46–48.

Owing to difficulties in modelling advanced stages of ALD in rodents, the functional contribution of the intestinal-microbiota–TLR4 axis in advanced liver disease, such as in the development of cirrhosis and HCC, is not well known. In one study, ethanol-fed transgenic mice with global TLR4 deficiency, which additionally expressed the NS5A HCV protein, were protected from HCC development, suggesting that TLR4 signalling synergizes with HCV to promote HCC⁴⁹. This finding fits well with the well-established clinical observation that alcohol abuse is an important cofactor in promoting liver disease development and HCC in patients with chronic HCV infection ⁵⁰ and suggests a potential role for the LPS–TLR4 axis in the synergy between alcohol and HCV.

2.2. NAFLD

Although recognized as a disease only about two decades ago, NAFLD represents the most prevalent liver disease, and is projected to become the leading contributor to CLD and the development of HCC⁵¹. In comparison to other CLDs, NAFLD carries a low relative individual risk for HCC development, but makes a big population-wide contribution to HCC development owing to its high prevalence⁵¹. Studies in germ-free and gnotobiotic mice have revealed a key contribution of the gut microbiota to metabolism and energy harvest. As such, germ-free mice display decreased body weight despite increased food intake⁵². Metagenomic and microbiota transplantation studies have shown that the gut microbiota from obese individuals is more efficient at energy extraction and thereby contributes to obesity^53,54. Hence, treatment with antibiotics ameliorates high-fat-diet-induced NAFLD in mice⁵⁵. Moreover, patients with NAFLD display dysbiosis. However, bacterial abundance patterns were not consistent between studies, with levels of Bacteroidetes increased in some studies^36,56 and decreased in other studies^57,58, and a substantial overlap with healthy individuals⁵⁹. Interestingly, dysbiotic microbiota from mice fed a high-fat diet metabolize and convert dietary choline into methylamines, resulting in low circulating levels of plasma phosphatidylcholine⁶⁰. These low levels of phosphatidylcholine impaired secretion of VLDL, thereby reducing hepatic lipid export and contributing to fatty liver⁶¹. Thus, alterations in choline metabolism might link dysbiosis to the development of NAFLD.

The contribution of the gut microbiota to NASH is not as well documented as its role in earlier disease stages. High-fat diet increases intestinal permeability in mice with a two-to-three-fold increase in systemic LPS levels⁶². Likewise, intestinal permeability is increased in patients with NAFLD⁶³. In a model of NASH triggered by high-fat and high-cholesterol diet given to ApoE-deficient mice, TLR4 deficiency reduced hepatic inflammation and injury⁶⁴. To date, the functional role of pathways in NASH development has often been studied in mouse models such as the methionine-choline-deficient (MCD) diet — this diet results in a NASH phenotype strongly resembling human NASH but lacks other essential features of NASH, such as adiposity and insulin resistance, and therefore lacks clinical relevance. In the MCD diet model, the microbiota has a key role in NASH exacerbation as demonstrated by experiments, in which co-housing transmitted NASH risk and antibiotics reduced NASH risk⁶⁵. Conversely, fecal microbiota transplantation from healthy mice attenuated steatohepatitis in high fat diet treated mice ⁶⁶. The gut microbiota also has an important part in promoting HCC development in a mouse model in which HCC is driven by the carcinogen 7,12-dimethylbenz(a)anthracene (DMBA) and subsequent high-fat diet¹⁵. However, this model does not incorporate key features of NASH such as liver fibrosis and insulin resistance. Hence, further studies are needed to determine the functional role of dysbiosis in the progression of NASH and HCC in mouse models that incorporate adiposity and insulin resistance.

2.3. Chronic viral hepatitis

In contrast to ALD and NAFLD, there is little information on the role of the gut microbiota in chronic viral hepatitis. Current data suggest that dysbiosis and alterations of the gut–liver axis in patients with end-stage viral hepatitis and cirrhosis are similar to alterations in patients with cirrhosis from other causes⁶⁷. However, it is not known whether the gut microbiota contributes to the pathophysiology of chronic viral hepatitis and its progression to more advanced stages. A recent study demonstrated that the gut microbiota controls immune responses and tolerance to HBV in adult mice, with 6 weeks of antibiotic treatment preventing the clearance of HBV⁶⁸. Whether the impaired response to HBV is mediated by specific bacteria or the result of broad suppression of the bacterial microbiota remains an important unanswered question. Notably, HBV titres in patients positively correlate with risk for disease progression and HCC development⁶⁹. Hence, the gut microbiota might control antiviral responses that affect disease progression and HCC development.

2.4. Liver fibrosis

Liver fibrosis is part of the hepatic wound healing response and common to all types of advanced CLD. Notably, there is a strong correlation between hepatic fibrosis and HCC development with 80–90% of HCCs developing in fibrotic or cirrhotic livers. Thus, fibrosis represents a risk factor for HCC development⁷⁰. There is strong evidence for an important contribution of the microbiota–TLR4 axis to liver fibrosis. Studies from the past six decades have shown that antibiotics prevent hepatic injury and fibrosis induced by CCl₄ treatment, bile duct ligation or a choline-deficient diet, and that endotoxin enhances hepatic fibrosis induced by a choline-deficient diet^71–73. Studies in knockout mice have highlighted a key role for TLR4 and other important mediators in the TLR4 signalling pathway, such as CD14 and lipopolysaccharide-binding protein (LBP), in experimental models of toxic and cholestatic liver fibrosis^73,74. However, recent studies have demonstrated an increase in liver fibrosis in germ-free mice^75,76, which seemingly contradicts the decrease of liver fibrosis seen in gut-sterilized mice. It has become apparent that the endogenous commensal microbiota provides hepatoprotective signals and that complete absence of the gut microbiota results in increased liver injury — probably owing to an absence of TLR4-mediated activation of anti-apoptotic NF-κB signalling — and a subsequent increase in liver fibrosis, as demonstrated in several models^13,75,76. Nonetheless, the bacterial microbiota has an important role in promoting HCC in the setting of liver fibrosis, as demonstrated by reduced HCC formation in TLR4-deficient, germ-free and antibiotic-treated mice in a diethylnitrosamine (DEN) plus CCl₄ model of HCC¹³. Consistent with previous studies^71–73, treatment with nonabsorbable antibiotics resulted in a strong reduction of fibrosis despite increased liver injury¹³. However, further studies are required to investigate how the intestinal microbiota affects HCC development promoted by chronic inflammation, injury and fibrosis, without preceding carcinogen exposure.

3.1. HCC promotion via a leaky gut and the MAMP–TLR axis

High circulating LPS levels in mice and patients with CLD as well as in HCC^14,24,77,78 demonstrate the presence of a leaky gut during multiple stages of CLD and hepatocarcinogenesis (FIG. 1). Functional experiments in germ-free, gut-sterilized, TLR-deficient and LPS-treated mice have provided evidence that the leaky gut, via LPS and its receptor TLR4, makes essential contributions to hepatocarcinogenesis. As such, HCC development induced by the combination of DEN and CCl₄ was attenuated in gut-sterilized and germ-free mice compared with their specific pathogen-free counterparts¹³. In addition to causing characteristic infectious complications in end-stage liver disease, increased bacterial translocation also generates a chronic inflammatory state in the liver. The inflammatory responses in the liver are mediated by interaction between MAMPs and host PRRs, specifically the TLRs⁷⁹. Accordingly, chronic infusion of low-dose LPS via osmotic pumps promotes HCC development in mice¹³. Likewise, disruption of the gut barrier by administration of dextran sulfate sodium not only results in increased systemic LPS levels and increased liver fibrosis, but also promotes HCC formation in mice^80,81. Conversely, inhibition of TLR4 signalling suppresses liver inflammation, fibrosis and HCC formation in mice and rats^13,14,73. The majority of tumour-promoting signals from the leaky gut occur in late stages of DEN+CCl₄-induced hepatocarcinogenesis, as demonstrated by strong inhibitory effects of gut sterilization on HCC formation in late stages but only mild effects in early stages¹³. However, the relative contribution of the leaky gut at early versus late stages of hepatocarcinogenesis has not yet been tested in other models.

TLR4 is present in multiple hepatic cell types, including Kupffer cells, hepatic stellate cells (HSCs), endothelial cells and hepatocytes. Experiments in bone-marrow-chimeric mice demonstrated that TLR4 expressed on liver-resident cells (which include hepatocytes, HSCs and Kupffer cells) is responsible for promotion of fibrogenesis and hepatocarcinogenesis¹³. LPS from the leaky gut seems to promote hepatocarcinogenesis via multiple cellular targets, including HSCs, the hepatocyte–tumour compartment as well as liver-resident Kupffer cells. In HSCs, TLR4 activation leads to an NF-κB-mediated upregulation of the hepatomitogen epiregulin¹³. Epiregulin is an epidermal growth factor family member with a potent mitogenic effect on hepatocytes⁸². Accordingly, epiregulin-deficient mice displayed reduced hepatocarcinogenesis when treated with DEN+CCl₄¹³. Another key mechanism by which the LPS–TLR4 axis promotes HCC formation is via NF-κB-mediated prevention of hepatocyte apoptosis. Accordingly, expression of the apoptosis marker cleaved caspase 3 in TLR4-deficient and gut-sterilized mice is inversely correlated with the formation of tumours¹³. However, due to the lack of studies in mice with conditional TLR4 ablation, it remains unclear whether this survival pathway is directly activated in the hepatocyte–tumour cell compartment, or whether it might involve paracrine signals from neighboring TLR4-expressing cells such as HSC or Kupffer cells. Moreover, it has been demonstrated that activation of the LPS–TLR4 signalling pathway in Kupffer cells leads to TNF- and IL6-dependent compensatory hepatocyte proliferation as well as reduced oxidative stress and apoptosis¹⁴. In addition, TLR4 activation in HCC cell lines by LPS enhances their invasive potential and induces the epithelial–mesenchymal transition⁸³. In order to delineate the contribution of TLR4 on specific cell types in the liver, further experiments in mice with conditional TLR4 ablation are required.

Together, these data clearly show that the leaky gut, via MAMP–TLR-mediated signals, contributes to hepatocarcinogenesis. Dysbiosis (discussed below) and the leaky gut are probably intimately linked; it is likely that intestinal dysbiosis contributes to a leaky gut by multiple mechanisms, such as dysbiosis-induced alterations of the intestinal barrier as well as a shift to bacterial species with increased propensity to translocate.

3.2. HCC promotion via dysbiosis, bacterial metabolites and immunosuppression

Increasing evidence supports a key role for dysbiosis in the development of CLD and HCC (FIG. 1). Metagenomic studies have revealed substantial alterations in the composition of the gut microbiota in a range of CLD as well as in patients with cirrhosis^12,32. The gut microbiomes of patients with advanced liver disease and cirrhosis are characterized by an increase in potentially pathogenic bacteria, along with reduced numbers of bacteria with beneficial properties^32,84,85. Studies conducted so far on the gut microbiota in liver cirrhosis have pooled patients with different underlying liver diseases³², indicating that at least some of the microbial alterations in cirrhosis are common to different aetiologies, and suggesting that alterations are driven by characteristic features of end-stage liver disease, such as reduced bile output and changes to the intestinal secretion of antimicrobial peptides and IgA . Key changes in the composition of the intestinal microbiota in cirrhosis include enrichment of Veillonella or Streptococcus as well as decreased bacteria from the order Clostridiales³². Of note, the majority of the patient-enriched species were of buccal origin, suggesting an invasion of the gut from the mouth in liver cirrhosis³². The finding that the intestinal microbiota of patients with compensated cirrhosis differs from that of patients with decompensated cirrhosis³⁴ suggests that cirrhosis stage, rather than the underlying liver disease, drives gut microbiota changes. However, a recent small-scale study described differences in the gut microbiota between different types of underlying liver disease⁸⁶. Therefore, sufficiently powered studies in large cohorts are needed determine disease-specific alterations of the gut microbiota in liver cirrhosis. In addition to alterations in bacterial composition, there is evidence for bacterial overgrowth in the upper gastrointestinal tract, which in turn is associated with increased circulating LPS levels ⁸⁷. Bacterial translocation in the upper gastrointestinal tract is relevant for the development of liver disease owing to the anatomic connection of the small intestine to the liver. Recent studies have demonstrated differences in the duodenal and salivary microbiota between healthy controls and patients with cirrhosis^86,88, suggesting that there are also qualitative and quantitative changes in the upper gastrointestinal tract that might be linked to changes in the more distal microbiota and contribute to the pathophysiology of CLD as well as the development of HCC.

Functional studies utilizing co-housing and faecal transplantation have provided evidence that dysbiosis is a transmissible driver of liver disease development and progression ^65,89. In one study, high-fat diet feeding in mice resulted in dysbiosis, with increased abundance of Gram-negative bacteria and a reduced ratio of Bacteroidetes to Firmicutes . Transplantation of these dysbiotic microbiota into control-diet-fed mice that had undergone bile duct ligation increased liver damage and fibrosis in the recipient⁸⁹. Similarly, dysbiosis represented a transmissible risk factor in a genetic NASH model in which NASH was triggered by inflammasome deficiency; co-housing of dysbiotic inflammasome-deficient mice with control mice resulted in the development of NASH in control mice⁶⁵. Although studies demonstrating a transmissible HCC risk by dysbiotic microbiota are still missing, several functional studies point towards a contribution of dysbiosis. As such, perturbation of the gut eubiosis by penicillin increased HCC formation in rats⁷⁷, which could be suppressed by probiotics⁷⁷.

Recent evidence suggests that the effects of dysbiosis on the development of liver disease and HCC are mediated by bacterial metabolites, possibly in a disease-specific manner. In a mouse model of NASH-induced HCC, triggered by the combination of DMBA and high-fat diet, there was a strong increase in Gram-positive bacterial strains, in particular of specific Clostridium clusters¹⁵. At the same time, this treatment led to increased serum levels of deoxycholic acid (DCA), a secondary bile acid whose production depends on 7α-dehydroxylation of primary bile acids by the bacterial microbiota, notably Clostridium clusters. The key role of DCA in hepatocarcinogenesis was further demonstrated in experiments that showed increased HCC development in mice after supplementing diets with DCA, and decreased HCC formation after inhibition of 7α-dehydroxylation¹⁵. In concert with TLR2 agonist lipoteichoic acid, DCA promoted a senescence-associated secretory phenotype in hepatic stellate cells, which in turn suppressed anti-tumor immunity through a prostaglandin E2-dependent mechanisms⁹⁰. Together, these studies link bacterial dysbiosis to altered immune responses via bacterial metabolites and MAMPs. Further studies are required to determine whether the procarcinogenic effects of dysbiotic microbiota may be mediated by additional pathways. The gut microbiota exerts a key role in a number of other metabolic pathways, including overall energy extraction from the diet as well as the generation of a wide range of important metabolites with beneficial effects for the host⁹¹. One example is the production of short-chain fatty acids (SCFAs), which are a primary energy source for intestinal epithelial cells⁹¹, and might provide a link between dysbiosis and alterations of the intestinal barrier that lead to a leaky gut and increased risk for HCC development, as discussed above.

4.1. Antibiotics

Because antibiotics target several pathways through which the gut microbiota promote HCC development (Table 1), they could represent one of the most efficient strategies to interrupt the tumor-promoting gut liver axis in CLD: Decreasing the overall number of bacteria in the gut and eliminating bacteria that have a high ability to translocate will reduce bacterial translocation and thereby inhibit proinflammatory signals coming from leaky gut. At the same time, selective antibiotics might also block the production of HCC-promoting bacterial metabolites, such as DCA¹⁵, by reducing the number of bacteria that produce specific metabolites. Continuous gut sterilization by a cocktail of oral antibiotics, consisting of ampicillin, neomycin, metronidazole and vancomycin, effectively reduced the number and size of HCC induced by DEN+CCl₄ or DMBA+HFD in mice^13,15. Moreover, this antibiotic cocktail also reduced liver fibrosis ⁷³, which often precedes HCC and represents a risk factor for HCC development in CLD⁷⁰. Of note, administration of antibiotics at late stages of carcinogenesis, when microscopic tumours already existed, was more efficient at reducing HCC in mice than administration at earlier stages¹³. These data support the concept that HCC prevention by antibiotic treatment could be applied even at late stages, i.e. in patients with advanced cirrhosis and high risk for HCC development. However, findings from mice cannot be translated directly to patients as long-term administration of the employed antibiotic cocktail would be deleterious due to the depletion of almost all detectable commensal microbiota (>99.5%)^73,104 and the inclusion of nephrotoxic drugs, such as neomycin. Moreover, HCC prevention with antibiotics would require long-term, possibly life-long administration. Therefore, the use of single antibiotics with a high safety profile in patients with CLD represents the only clinically feasible approach.

Currently, two antibiotics, norfloxacin and rifaximin, have shown beneficial effects in patients with CLD or murine HCC models and fulfill these criteria. Vancomycin, another antibiotic that has shown effectiveness as monotherapy in the prevention of HCC in the combined DMBA and high-fat diet mouse model¹⁵, is rarely used for long-term therapy in patients and may cause a number of potentially severe adverse effects. Gram-negative bacteria have been found to be the most adept at translocating to the mesenteric lymph nodes and are the most frequent cause of spontaneous bacterial infections in patients with cirrhosis^105–107. Norfloxacin, a poorly absorbed quinolone, is currently one of the drugs of choice for the primary or secondary prophylaxis of spontaneous bacterial peritonitis and infections in high-risk patients with cirrhosis ¹⁰⁸. Clinical trials in patients with advanced cirrhosis have shown that long-term use of orally administered norfloxacin is safe, produces a marked reduction of gram-negative bacteria in the faecal microbiota¹⁰⁹, reduces the 1-year probability of developing spontaneous bacterial peritonitis and hepatorenal syndrome and improves 3-month survival⁹². Although these data show that norfloxacin can effectively reduce small intestinal bacteria overgrowth and bacterial translocation in patients with advanced cirrhosis, the effects of norfloxacin on HCC development in patients with liver cirrhosis are not known. A major problem with the use of norfloxacin is the development of antibiotic resistance^110–112, suggesting that it might be suitable for treatment lasting weeks to months but not for long-term or life-long application in patients with cirrhosis. Rifaximin is a nonabsorbable antibiotic with broad-spectrum antimicrobial activity and an excellent safety profile¹¹³ that was initially approved for the treatment of traveller’s diarrhoea but is increasingly used for the prevention of hepatic encephalopathy⁹⁹. Moreover, rifaximin appears to reduce the development of spontaneous bacterial peritonitis and may improve portal hypertension, suggesting that it effectively targets the gut–liver axis in advanced liver disease^93,114,115. Similar to norfloxacin, rifaximin has been noted to increased survival in patients with advanced liver cirrhosis in several small-scale trials^92–95. Of note, rifaximin reduces HCC development in the DEN–CCl₄ model of HCC, albeit less efficiently than the quadruple antibiotics cocktail described above¹³. Despite the large number of patients receiving rifaximin for the prevention of hepatic encephalopathy, the effects of rifaximin on HCC development remain unknown. Therefore, studies that determine the effects of long-term rifaximin treatment on HCC development are urgently needed. In contrast to norfloxacin, clinically relevant development of resistance to rifaximin has not been reported, suggesting that it is well-suited for long-term or even life-long treatment. As data from murine studies show that non-absorbable antibiotics and norfloxacin improve ALD^48,116 and insulin resistance in NAFLD¹¹⁷, they might be particularly attractive for HCC prevention in these patient groups.

4.2. Probiotics

Probiotics have been proposed as a means of re-equilibrating the gut microbiota in CLD by restoring beneficial bacteria. Although a large number of studies have demonstrated the effectiveness of probiotics in treating liver diseases both in animal models and in patients (reviewed elsewhere^118,119), substantial controversy remains on the basis of: the inability of most probiotics to permanently colonize the gut; largely unknown mechanisms of action, in particular given the lack of permanent colonization; the large number of different combinations of bacteria within different probiotics that have not been systemically evaluated and compared for their efficacy in CLD; and in view of a lack of large-scale studies, potential publication bias towards studies reporting positive results . So far, probiotics have only been investigated in murine HCC models and data in patients are lacking (Table 1). In a rat model of DEN-induced hepatocarcinogenesis, administration of VSL#3 (containing Streptococcus thermophiles, Bifidobacterium breve, Bifidobacterium longum, Bifidobacterium infantis, Lactobacillus acidophilus, Lactobacillus plantarum, Lactobacillus paracasei, Lactobacillus delbrueckii subsp. Bulgaricus) mitigated enteric dysbiosis, ameliorated intestinal inflammation and decreased liver tumour growth and multiplicity⁷⁷. In a subcutaneous transplant mouse model, the probiotic mixture Prohep (comprising Lactobacillus rhamnosus GG, Escherichia coli Nissle 1917 and heat inactivated VSL#3) reduced tumour size and weight¹²⁰. The authors suggested that a shift of the gut microbiota composition towards beneficial bacteria such as Prevotella and Oscillibacter, and production of anti-inflammatory mediators by these bacteria, decreased Th17 cell levels within tumors and thereby limited tumour growth. However, the effects of probiotics on endogenously arising tumours are not known. To date, there are several clinical trials on probiotics in patients with CLD but none in patients with HCC. A double-blind trial showed that daily intake of VSL#3 reduced the risk of hospitalization for hepatic encephalopathy, as well as Child–Turcotte–Pugh and model for end-stage liver disease (MELD) scores, in patients with cirrhosis¹²¹. Another randomized trial showed that 4-month supplementation with VSL#3 improves NAFLD in children¹²². The possible mechanisms of action include modulation of the host microbiota¹²³, improvement of gut barrier function and modulation of the immune system. However, further studies are required to confirm these data, extend human studies and investigate mechanisms of action.

4.3. FMT

FMT has successfully been used in patients with C. difficile infection, resulting in restoration of eubiosis and clinical improvements that were superior to standard antibiotic therapy¹⁸. Currently, FMT is being evaluated in clinical trials for a number of additional diseases including NASH¹²⁴ and cirrhosis¹²⁵. A randomized controlled trial demonstrated amelioration of hepatic and peripheral insulin resistance in patients with metabolic syndrome who had received microbiota from lean donors¹²⁶. However, one needs to keep in mind that patients receiving FMT for recurrent C. difficile infection have usually undergone multiple courses of antibiotic treatment, and present with a marked reduction of microbial diversity ¹⁸. Thus, these patients not only represent an ideal ‘breeding ground’ for transplanted microbiota but also suffer from a disease that is clearly linked to reduced bacterial diversity and overgrowth of single and measurable pathogenic strain. Nonetheless, it is conceivable that FMT might also restore eubiosis in patients with CLD, similar to effect seen in the trial of Vrieze et al.¹²⁶, and that FMT might reduce or delay the development of HCC (Table 1). However, there are currently no data supporting this premise and a number of hurdles have to be overcome. Most importantly, it is not clear whether the severe alterations of the gastrointestinal ecology in cirrhosis would allow permanent restoration of the microbiota by FMT. It is possible that effects will be transient and the microbiota will ultimately revert to the pre-FMT state. Moreover, there is substantial concern that viral infections and other pathogens might be transmitted via FMT, which would be particularly harmful to patients with advanced liver disease owing to their immunosuppression. In the future, faeces might be substituted in favour of defined mixtures of cultured bacteria that resemble the human microbiota transplanted via FMT and confer the same beneficial effects. This approach will not only alleviate concerns regarding the inadvertent transmission of disease-causing pathogens through FMT, but also make intestinal microbiota therapy more acceptable to patients and physicians¹²⁷. Once this goal has been achieved, patients with advanced liver disease should be considered as potential candidates to study effects on disease progression and HCC development.

4.4. TLR antagonists

Several studies have shown a key role for the TLR4 pathway as a mediator of the disease-promoting effects of the gut–liver axis in CLD and hepatocarcinogenesis^13,14,73. On the basis of these findings, blocking the TLR4 pathway might represent another avenue for HCC prevention (Table 1). With detailed knowledge about mechanisms by which LPS activates TLR4, a variety of TLR4 antagonist have been developed, which can be clustered into several groups: compounds binding and sequestering LPS, such as polymyxin B; compounds antagonizing LBP and CD14–LPS interactions¹²⁸; compounds targeting LPS–MD-2 or LPS–MD-2–TLR4 interactions, such as E5531 and eritoran (E5564); compounds directly targeting TLR4, such as resatorvid (TAK-242); and molecules inhibiting TLR4 activity such as thalidomide (reviewed elsewhere¹²⁹). Eritoran¹³⁰ and resatorvid¹³¹ improved survival in animal models of sepsis, but did not reduce mortality in patients with severe sepsis^132,133. So far, none of these agents have been tested in clinical trials in patients with CLD or HCC. Although TLR antagonists represent an exciting opportunity, long-term inhibition of TLR4 could result in immunosuppression, which might be deleterious due to the severely immunocompromised state of CLD patients. Therefore, the safety profile of TLR4 antagonists needs to be carefully evaluated before long-term studies for prevention of HCC and other complications of CLD can be considered.

4.5. Targeting the gut barrier

On the basis that the leaky gut is a major driver of liver disease progression and HCC development (FIG. 1), targeting the gut barrier seems an attractive therapeutic approach (Table 1) that might avoid some of the complications of targeting the microbiota (such as development of resistance and/or decreased microbial diversity) or receptors that mediate the disease-promoting effects of a leaky gut (such as immunosuppression resulting from TLR4 antagonism. Moreover, therapies that target the gut barrier could potentially be combined with other approaches that directly target the gut microbiota or liver. With improved understanding of the gut barrier and mechanisms that disrupt the gut barrier in cirrhosis, targeting the gut barrier via specific pharmacologic approaches seems to be realistic.

Bile acids are an important regulator of the gut barrier. Decreased bile secretion in rodents by either ligation of the common bile duct or induction of cirrhosis contributes to bacterial translocation, which is not only caused by intestinal bacterial overgrowth but also by increased gut permeability^20,28,134. Notably, these effects are attenuated after oral administration of bile acids in different experimental cirrhotic models^20,134,135. FXR is a receptor for bile acids that mediates their effects on the intestinal epithelial barrier as well as multiple effects on the liver, such as suppression of bile acid synthesis, inhibition of liver inflammation, promotion of liver regeneration and tumour suppression (reviewed elsewhere¹³⁶). Many of the hepatic effects of FXR activation are mediated by intestinal FXR receptors, resulting in the release of FGF19, which then acts on targets in the liver^136–139. Fxr-deficient mice exhibit compromised intestinal integrity, with further deterioration after bile duct ligation²⁰, and a high incidence of HCC¹⁴⁰. Accordingly, FXR activation by agonists GW4064 or obeticholic acid (OCA) attenuates mucosal injury, ileal barrier permeability, bacterial overgrowth and bacterial translocation in mice and rats ^{20,21,141,142}. Moreover, OCA improves portal hypertension (which might contribute to bacterial translocation in cirrhosis) in thioacetamide- or bile duct ligation-treated rats¹⁴³. OCA has a high safety profile in patients with NASH as demonstrated in the FLINT trial, with the major adverse effects being pruritus and alterations of serum lipid profiles¹⁴⁴. Thus, OCA seems to be a promising candidate for HCC prevention therapies, and could be particularly effective by correcting multiple abnormalities in the gut–liver axis that promote the development of chronic inflammation and HCC in patients with cirrhosis.

Increased production of TNF by monocytes in mesenteric lymph nodes constitutes one of the main factors increasing tight junction permeability^145,146. TNF increases tight junction permeability by decreasing expression of tight junction proteins as well as by activating myosin light chain kinase (MLCK)¹⁴⁷. Treatment with an anti-TNF monoclonal antibody decreases the incidence of bacterial translocation in experimental cirrhosis in rats¹⁴⁸. However, translating these findings to patients might be difficult because of strong immunosuppressive effects of TNF inhibitors and increased rates of severe infection. Owing to these adverse effects, long-term anti-TNF therapy might confer more harm than benefit, and further efforts need to be made to develop therapies that act locally to improve gut barrier function without negatively affecting systemic immune responses.

4.6. Prokinetics

Another factor that contributes to intestinal bacteria overgrowth in liver cirrhosis is gut dysmotility¹⁴⁹. The prokinetic drug cisapride not only decreases intestinal transit time but also inhibits intestinal bacterial overgrowth and bacterial translocation, both in animal models^150,151 and in patients with cirrhosis^150,152. However, the long-term benefits of prokinetics such as cisapride have yet to be determined in patients with CLD (Table 1). One of the purported mechanisms for altered motility in cirrhosis is increased adrenergic activity. Accordingly, nonselective β-adrenergic blockers decrease intestinal transit time and reduce intestinal bacterial overgrowth, intestinal permeability and bacterial translocation in experimental models of cirrhosis as well as in patients^153–157. Interestingly, a retrospective long-term observational study suggests that propranolol treatment might decrease HCC occurrence in patients with HCV cirrhosis¹⁵⁸, suggesting a potential role for HCC prevention.

Malnutrition is common in patients with CLD and is associated with increased morbidity and mortality (reviewed elsewhere¹⁴⁹). Of note, malnutrition increases intestinal permeability and facilitates bacterial translocation¹⁵⁹. Therefore, nutritional support represents an important aspect to correct dysbiosis in patients with CLD.