Effect of rifaximin on gut microbiota composition in advanced liver disease and its complications

Abstract

Liver cirrhosis is a paradigm of intestinal dysbiosis. The qualitative and quantitative derangement of intestinal microbial community reported in cirrhotic patients seems to be strictly related with the impairment of liver function. A kind of gut microbial “fingerprint”, characterized by the reduced ratio of “good” to “potentially pathogenic” bacteria has recently been outlined, and is associated with the increase in Model for End-Stage Liver Disease and Child Pugh scores. Moreover, in patients presenting with cirrhosis complications such as spontaneous bacterial peritonitis (SBP), hepatic encephalopathy (HE), and, portal hypertension intestinal microbiota modifications or the isolation of bacteria deriving from the gut are commonly reported. Rifaximin is a non-absorbable antibiotic used in the management of several gastrointestinal diseases. Beyond bactericidal/bacteriostatic, immune-modulating and anti-inflammatory activity, a little is known about its interaction with gut microbial environment. Rifaximin has been demonstrated to exert beneficial effects on cognitive function in patients with HE, and also to prevent the development of SBP, to reduce endotoxemia and to improve hemodynamics in cirrhotics. These results are linked to a shift in gut microbes functionality, triggering the production of favorable metabolites. The low incidence of drug-related adverse events due to the small amount of circulating drug makes rifaximin a relatively safe antibiotic for the modulation of gut microbiota in advanced liver disease.

Key Words: Liver cirrhosis, Gut microbiota, Rifaximin, Hepatic encephalopathy, Spontaneous bacterial peritonitis, Ascites, Endotoxemia, Thrombocytopenia

Core tip: Advanced liver disease is characterized by intestinal dysbiosis, which has been involved in the pathogenesis of complications. Rifaximin is able to improve cognitive tests and practical abilities, to reduce the risk of hepatic encephalopathy (HE) recurrence and the number of HE-related hospitalizations. Rifaximin efficacy seems not associated with major changes in gut bacteria composition but rather with a shift in the microbiome functionality. Rifaximin is useful in the prevention of spontaneous bacterial peritonitis in patients with ascites. Rifaximin reduces endotoxemia and has beneficial effects on cirrhotic patients hemodynamics, reducing the incidence of complications related to portal hypertension.

INTRODUCTION

Rifaximin is a non-systemic antibiotic approved for the treatment of traveler’s diarrhea, irritable bowel syndrome (IBS) with diarrhea and overt hepatic encephalopathy (HE)[1]. It has in vitro bactericidal and bacteriostatic activity against aerobic and anaerobic Gram-positive and Gram-negative species, being also able to reduce bacterial virulence and translocation, and to inhibit bacterial adherence to gut mucosa[2-7].

Due to the low systemic absorption (only 0.4% of the oral administered dose), rifaximin has an optimal tolerability profile, and side effects as well as the induction of bacterial resistance are nearly lacking[1,8,9].

Beyond that, rifaximin has particular features which are not typical of a common antibiotic molecule. In vitro and in vivo models and preliminary experiences in humans[10-14] have demonstrated that rifaximin does not change the overall composition of the gut microbiota while it is able to provide minimal changes, such as promoting the growth of bacteria beneficial to the gut. Nevertheless, rifaximin modulates the release of inflammatory cytokines[15,16] and increases NF-κB expression, exerting anti-inflammatory effects that could counteract the pro-inflammatory response observed in conditions of gut microbiota derangement[17].

Based on these evidences, rifaximin use has been extended to the management of pathologies associated with gut microbiota deregulation such as irritable bowel syndrome[11,18-21], inflammatory bowel diseases[10,13,22-30], diverticular disease[31-36] and liver cirrhosis and its complications.

Liver cirrhosis is a paradigm of intestinal dysbiosis. Indeed, the physiological partitioning of the gastrointestinal tract is deranged in cirrhotic patients, due to the decreased secretion of gastric acid (often favored by medications[37]), to the reduced gastrointestinal motility, to the impaired systemic and mucosal immune response and to the low concentration of bile acids in the colon[38]. The epiphenomenon of this chronic dysfunction is a profound alteration of the gut microbiota composition, which is both quantitative (Small Intestinal Bacterial Overgrowth, SIBO) and more pronounced in the advanced stages of the disease and in case of decompensation (Figure 1).

Figure 1 Effects of Rifaximin on gut-liver axis. Rifaximin decreases endotoxemia and inflammation both directly and indirectly, by reducing bacterial translocation, counteracting bacterial overgrowth and modulating gut microbiota composition and function. Due to these peculiar effects, rifaximin is used for the treatment of advanced liver disease complications. HE: Hepatic encephalopathy; SIBO: Small intestinal bacterial overgrowth; SBP: Spontaneous bacterial peritonitis; HRS: Hepatorenal syndrome.

This is the rationale for gut microbiota modulation in patients with liver cirrhosis, especially in those with severe impairment of liver function presenting with complications.

THE “FINGERPRINT” OF GUT MICROBIOTA IN LIVER CIRRHOSIS

The introduction of metagenomic techniques such as 16S rRNA-based pyrosequencing has recently allowed to identify which modifications of the gut microenvironment are the most frequently observed in liver cirrhotic patients[39]. The human gut hosts a bacterial core community involved in maintaining gastrointestinal health and mainly composed of the phyla Bacteroidetes and Firmicutes, which include the genera Bacteroides, Clostridium clusters XIVa and IVa, Eubacterium, Faecalibacterium, Lactobacillus, and Roseburia. In patients affected by liver cirrhosis, at the phylum level, Bacteroidetes are decreased in favor of Fusobacteria and Proteobacteria, such as Enterobacteriaceae and Pasteurellaceae[40-42]. Looking at family, genus and species division, the increase in Enterobacteriaceae, Streptococcaceae and Veillonellaceae abundance has been reported in cirrhotic patients compared with healthy controls, whereas Lachnospiraceae, Ruminococcaceae, Clostridium clusters XI and XIVab, lactic acid bacteria, Bifidobacteria and Faecalibacterium prausnitzii seem to be reduced[40-46]. Notably, Enterobacteriaceae family includes Escherichia coli and Klebsiella spp., key bacteria in the pathogenesis of spontaneous bacterial peritonitis (SBP). In addition to the unbalance between potentially pathogenic and beneficial bacteria, the major part of the metagenomic species enriched in cirrhotics’ fecal samples belong to Veillonella or Streptococcus taxa, which usually derive from the mouth or the small intestine[47]. Although this may apparently confirm the subversion of the gastrointestinal physiology occurring during the course of liver disease, when cirrhotics’ salivary microbiota is specifically analyzed and compared with the fecal one, they seem significantly different rather than similar[46]. More in detail, Streptococcaceae are prevalent in the saliva, whereas stools are characterized by a reduction in the autochthonous taxa Lachnospiraceae, Ruminococcaceae, and Clostridiales XIV. However, about half of samples analyzed in this study belonged to patients who have had previous episodes of HE and were on lactulose, with the addition of rifaximin in two cases. Further analyses to discriminate conditions predisposing to the “mixing-up” of bacteria from different sites of the gastrointestinal tract are needed to quell this debate.

Interestingly, the alteration of gut microbiota composition seems to have a prognostic significance, or at least to follow the evolution of liver disease. Generally speaking, Qin et al[47] demonstrated that metagenomic species enriched in cirrhotic patients correlate with the severity of the disease, in a proportion dependent on bacterial load. In other studies, the reduction in Clostridiaceae as well as in Veillonellaceae and in Porphyromonadaceae has been associated with inflammation and with the progression of liver disease and Streptococcaceae have been reported to correlate positively with Child Pugh score in contrast to Lachnospiraceae which correlated negatively[41,43,48].

Taking together these findings, cirrhotic patients’ microbiota is characterized by a higher proportion of potentially pathogenic bacteria, lacking of those species recognized as beneficial to intestinal health and homeostasis. Notwithstanding, the reduction in the ratio between “good” (e.g., Lachnospiraceae, Ruminococcaceae and Clostridia cluster XIV) and potentially “bad” bacteria (e.g., Staphylococcaeae, Enterobacteriaeceae and Enterococcaceae) - namely “cirrhosis dysbiosis ratio” or CDR - is characteristic of the individuals with a more severe disease, such as cirrhotic outpatients and inpatients[48].

Given the evidence that the progression of liver disease is associated with a change in the gut microenvironment, liver cirrhosis complications consequently grow in the soil of intestinal dysbiosis.

RIFAXIMIN AND GUT MICROBIOTA MODULATION IN ADVANCED LIVER DISEASE AND ITS COMPLICATIONS

Rifaximin and gut microbiota modulation in HE

Several differences have been reported in the gut microbiota of cirrhotic patients with or without HE. In patients with minimal HE, Streptococcaceae represent the prevalent bacterial family, and the abundance of Streptococcus salivarius, which is involved the production of ammonia, is increased[45]. Alcaligeneceae, Porphyromonadaceae and Enterobacteriaceae have also been associated with HE in cirrhotics; in particular, Alcaligeneceae and Porphyromonadaceae are significantly linked with poor cognitive performance, and Enterobacteriaceae with a worse MELD score[49]. In addition, a decreased CDR has been reported in cirrhotic patients with HE[48]. Similar results have been obtained by the analysis of mucosal microbiome from sigmoid biopsies: Enterococcus, Veillonella, Megasphaera, Bifidobacterium, and Burkholderia were predominant in patients with HE, whereas cirrhotics without HE presented an increased abundance of the “good” genus Roseburia, and the healthy controls an increased abundance of Dorea, Subdoligranulum, Incertae Sedis XIV, Blautia, Roseburia, Faecalibacterium and a few pathogenic genera[50]. Since the intestinal microenvironment of cirrhotics without HE has been demonstrated to be closer to healthy people's one[46], it is not surprising that, in patients with HE, the more the mucosal microbiota resembled that of controls, the better was the cognitive performance and the lower were the serum markers of inflammation[50].

Studies focused on clinical outcomes reported a high efficacy of rifaximin in cirrhotics with HE and a mild/moderate stage of disease. A randomized, double-blind, placebo-controlled trial including 299 patients has proved that rifaximin with or without lactulose is able to reduce the risk of HE recurrence and the rate of HE-related hospitalization, especially in patients with MELD score < 18[51]. Similar results were also obtained in other studies including patients in different stages of liver disease, receiving various treatment schedules (Table 1)[52-58].

Table 1 Major studies describing the changes in gut microbiota composition and the effects of rifaximin treatment in patients with advanced liver disease and hepatic encephalopathy.

Study	Study design	No. patients	Disease severity	HE type	Treatment schedule	Results	Safety
Mas et al[56] 2003	Prospective randomized, double-blind, double-dummy, controlled trial	103	Not reported	Overt HE	50 pts rifaximin 1200 mg/d for 5-10 d	Improved neuropsychiatric and psychometric parameters in both groups	Abdominal pain: 4% rifaximin
					53 pts lactitol 60 g/d for 5-10 d	Reduced blood ammonia levels in both groups	Mild diarrhea: 2% lactitol
						No significant differences in efficacy (resolution/improvement 81.6% rifaximin vs 80.4% lactitol; unchanged/failure 18.4% rifaximin vs 19.6% lactitol)	Vomiting: 2% lactitol
						HE complete resolution: 53.1% rifaximin vs 37.2% lactitol
Paik et al[57] 2005	Prospective randomized		CTP:	Overt HE	32 pts rifaximin 400 mg TID for 7 d	Reduction in blood ammonia levels similar in both groups	Abdominal pain: 3% rifaximin
			rifaximin A: 0%, B: 50%, C: 50%		22 pts lactulose 90 mL/d for 7 d	Improvement in HE grade and index similar in both groups	Severe diarrhea: 4.5% lactulose
			lactulose A: 0%, B: 64%, C: 36%			Improvement in HE grade similar in both groups
Leevy et al[58] 2007	Retrospective	145	Not reported	Overt HE	Lactulose 30 cc BID for ≥ 6 mo followed by rifaximin 400 mg TID for ≥ 6 mo	HE grade III or IV: 6% after rifaximin 25% after lactulose (P < 0.001)	Hospitalizations (mean number): 0.5 rifaximin period vs 1.6 lactulose period (P = 0.001)
						Asterixis: 63% after rifaximin vs 93% after lactulose (P < 0.001)	Hospitalizations days (mean): 2.5 rifaximin period vs 7.3 lactulose period (P = 0.001)
							Diarrhea: 89% during lactulose, 99% during rifaximin
							Flatulence: 100% during lactulose, 100% during rifaximin
							Abdominal pain: 100% during lactulose, 100% during rifaximin
							Headache: 100% during lactulose, 99% during rifaximin
							However, severe adverse events were more common in the lactulose period (P < 0.001)
Bass et al[51] 2010	Prospective, randomized, double-blind, placebo-controlled	299	MELD score (%):	Overt HE	140 pts 550 mg BID for 6 mo	Rifaximin is more effective than placebo in maintaining HE remission (P < 0.001)	Incidence of adverse events was similar in the two groups; most frequently reported: nausea diarrhea, fatigue
			rifaximin		159 pts placebo	Breakthrough episodes rate: 22.1% rifaximin vs 45.9% placebo	Bacterial peritonitis: 1.4% rifaximin vs 2.5% placebo
			≤ 10: 24.3%		90% of pts also received lactulose	Risk of HE-related hospitalization: 13.6% rifaximin vs 22.6% placebo (P = 0.01)	Bacteremia: 0.7% rifaximin vs 1.3% placebo
			11-18: 67.1%				C. difficile infection: 1.4% rifaximin vs 0% placebo
			19-24: 8.6%				Sepsis: 0% rifaximin vs 1.3% placebo
			placebo:
			≤ 10: 30.2%
			11-18: 60.4%
			19-24: 8.8%
Bajaj et al[59] 2011	Prospective, randomized, double-blind, placebo-controlled	42	MELD score (mean)	Minimal HE	21 pts rifaximin 550 mg BID	Total driving errors improvement: 76% rifaximin vs 31% placebo (P = 0.013), with a significant reduction of speeding tickets (P = 0.005) and illegal turns on navigation (P = 0.01)	Infections rate: 0%
			rifaximin: 9		21 pts placebo for 8-wk		Hospitalization rate: 0%
			placebo: 9				Nausea: 14% rifaximin vs 14% placebo
						Cognitive performance improvement: 91% rifaximin vs 61% placebo (P = 0.01)	Self-limited vomiting: 5% rifaximin vs 5% placebo
						Improved psycho-social dimension (quality of life assessment by Sickness Impact Profile questionnaire) in the rifaximin group compared with the placebo group (P = 0.04)	Abdominal pain: 24% rifaximin vs 24% placebo
							Flatulence: 19% rifaximin vs 43% placebo
							Headache: 19% rifaximin vs 33% placebo
							Flu-like symptoms: 5% rifaximin
							Constipation: 5% rifaximin
							Self-limited diarrhea: 5% rifaximin vs 5% placebo
							Hitching: 5% placebo
							Anorexia and dry mouth: 5% placebo
Neff et al[52] 2012	Retrospective	203	MELD score (mean, range):	Overt HE	149 pts rifaximin monotherapy (400-1600 mg/d)	1-yr HE remission rate: 81% rifaximin vs 67% rifaximin + lactulose	Incidence of gastrointestinal bleeding, infection, hospitalization for dehydration/overt HE similar in both groups
			rifaximin 12 (8-27)		54 pts rifaximin (600-1200 mg/d) + lactulose (90 mL/d) dual therapy	Lower incidence of overt HE episodes in pts with mean MELD score ≤ 20
			rifaximin + lactulose 13 (11-26)
Bajaj et al[62] 2013	Prospective	20	MELD score (mean ± SD): 9.8 ± 3.3	Minimal HE	550 mg BID for 8 wk	Significant improvement in cognitive performance on all tests apart from the block design test	Not reported
						Significant improvement in serum bilirubin but not the other MELD score components
						No significant microbial change (modest reduction in Veillonellaceae and increase in Eubacteriaceae)
						Significant increase in serum saturated (myristic, caprylic, palmitic, palmitoleic, oleic and eicosanoic) and unsaturated (linoleic, linolenic, gamma-linolenic and arachnidonic) fatty acids, serum fructose, succinic acid and citramalic acid
						Change in correlation networks involving several bacteria (Enterobacteriaceae, Bacteroidaceae, Veillonellaceae, Porphyromonadaceae and Rikenellaceae) reflecting a functional shift in the gut microbiome
Sharma et al[53] 2013	Prospective, randomized, double-blind, placebo-controlled	120	CTP score (mean ± SD):	Overt HE	group A (63 pts): lactulose + rifaximin 1200 mg/d	HE remission rate: 76% group A vs 50.8% group B (P < 0.004)	Diarrhea: 13% group A vs 10% group B (P > 0.05)
			group A 9.9 ± 2.8		group B (57 pts): lactulose + placebo	Mortality: 23.8% group A vs 49.1% group B (P < 0.05). Death was mainly due to sepsis	Abdominal pain: 6% group A vs 7% group B (P > 0.05)
			group B 9.4 ± 2.5			Hospital stay (mean ± SD): 5.8 ± 3.4 in group A vs 8.2 ± 4.6 group B (P = 0.001)
			MELD score (mean ± SD):
			group A 24.9 ± 6.6
			group B 23.8 ± 5.18
Maharshi et al[54] 2014	Prospective, randomized, controlled	120 pts with acute variceal bleeding and no HE	CTP and MELD scores comparable between groups but not reported	Overt HE	60 pts lactulose 30 mL QID	Incidence of HE: 15% rifaximin vs 17% lactulose (P = 1)	Rifaximn group: 5% abdominal pain and nausea
					60 pts rifaximin 400 mg TID	Mortality: 17% rifaximin vs 13% lactulose (P = 1)	Lactulose group: 26.6% diarrhea, 15% abdominal bloating
					for 5 d	Hospital stay (mean ± SD): 10.6 ± 3.1 d rifaximin vs 12.4 ± 3.5 lactulose (pts with HE, P = 0.35); 6.3 ± 1.6 rifaximin vs 6.9 ± 1.9 lactulose (pts without HE, P = 0.18)
Sharma et al[55] 2014	Prospective, randomized, controlled	124	CTP	Minimal HE	31 pts LOLA 3 g TID for 2 mo	LOLA, rifaximin, and probiotics are superior to placebo in improving critical flicker frequency score	Not reported
			LOLA		31 pts rifaximin 400 mg TID for 2 mo	LOLA, rifaximin, and probiotics are superior to placebo in improving neuropsychometric tests
			A: 22.5%, B: 42%, C: 35.5%		32 pts probiotics BID for 2 mo
			rifaximn		30 pts placebo
			A: 39%, B: 32%, C: 29%
			probiotics
			A: 19%, B: 66%, C: 16%
			placebo
			A: 33%, B: 27%, C: 40%

MELD: Model for end stage liver disease; HE: Hepatic encephalopathy; CTP: Child turcotte pugh; LOLA: L-ornithine L-aspartate.

In addition to the roughly evident benefits on overt HE, rifaximin has also been reported to improve operational abilities and input integration capacity in patients with minimal HE, as demonstrated by the amelioration of driving simulator performance[59]. This positive shift in cognitive tests and practical abilities is undoubtedly accompanied by a significant improvement in health-related quality of life[60,61].

At the microscopic level, rifaximin does not seem to change stool microbiota composition in patients with minimal HE, and only a reduction in Veillonellaceae and an increase in Eubacteraceae have been observed[62]. Reasonably, the improvement in cognitive function and the reduced endotoxemia associated with rifaximin treatment derive from a beneficial modulation of gut microbiota metabolic profile rather than from a major rearrangement of the intestinal microbial community. Indeed, the Authors reported an increase in saturated and unsaturated fatty acids and in serum fructose, succinic acid and citramalic acid production after rifaximin treatment, but the most relevant finding was the modification of correlation networks involving several bacteria (Enterobacteriaceae, Bacteroidaceae, Veillonellaceae, Porphyromonadaceae and Rikenellaceae), metabolites and clinical outcomes, suggesting a functional change in the gut microbiome. Although only patients with minimal HE have been included and some selection biases could be identified, the study by Bajaj et al[62] is to date the only published experience reporting the metagenomic and metabolomic changes produced by rifaximin treatment in cirrhotics with minimal HE. Nevertheless, despite the good results in terms of efficacy, rifaximin role in the treatment of cirrhotics at high risk of developing HE, such as patients with high MELD scores or with transjugular intrahepatic portosystemic shunts or surgical portosystemic shunts or those with a recent episode of acute variceal bleeding, needs to be further investigated[63-65].

Rifaximin and gut microbiota modulation in SBP

Ascites and SBP are typical manifestations of decompensated liver disease. SIBO and bacterial translocation are the mainstay of SBP. Indeed, SIBO prevalence among cirrhotics is high, ranging between 30% and 70%[38] and it has been associated with the development of SBP due to the translocation of intestinal bacteria to the systemic circulation and the ascitic fluid[66]. Gram-negative bacteria such as Escherichia coli and Klebsiella spp. as well as Pneumococci, Streptococci and other Gram-positive and Gram-negative bacteria have been identified in 50% of cases by culture-based analysis of ascitic fluid[67]. However, bacterial DNA can be recognized in the ascites of half of cirrhotics even in absence of SBP and with negative cultures[44], and several studies identified microbes usually present within the gut[41,43,68]. Ascites microbial composition is linked with the stage of liver disease; indeed, Child-Pugh score is correlated with ascitic bacteria similarity and ascitic neutrophil count, further strengthening the connection between gut microbiota and liver cirrhosis progression[68].

Therefore, it has been hypothesized that rifaximin, being effective on SIBO, could be useful in preventing SBP. In the retrospective study by Hanouneh et al[66] a 72% reduction in SBP occurrence and a transplant free survival of 72% were observed in the 49 cirrhotic patients with ascites who received rifaximin (Table 2).

Table 2 Major studies describing the efficacy of rifaximin in preventing episodes of spontaneous bacterial peritonitis in patients with advanced liver disease.

Study	Study design	No. patients	Disease severity	Disease complication	Treatment schedule	Results	Safety
Hanouneh et al[66] 2012	Retrospective	404	MELD score (mean ± SD):	SBP	49 pts received rifaximin	SBP incidence: 11% in pts on rifaximin vs 32% in controls (P = 0.002)	Not reported
			rifaximin: 17.6 ± 7.7		400 mg TID mainly for HE
			no rifaximin 17.7 ± 7.5		(recurrent HE or intolerance to lactulose)	72% SBP reduction rate in rifaximin group after adjusting for MELD score, CTP score, serum sodium, and ascitic fluid total proteins (P = 0.007)
			CTP score
			rifaximin B: 6.1%, C: 93.9%
			no rifaximin B: 33%, C: 67%
						72% transplant-free survival for pts on rifaximin vs 57% for controls (P = 0.045)
Lutz et al[69] 2014	Prospective, observational	152	CTP score:	SBP	Group 1 (108 pts): no prophylaxis	SBP occurrence rate: 32/152 (21%) overall, 22.2% group 1, 29.6% group 2 and 0% group 3 (P = 0.02 group 2 vs group 3 and P = 0.04 group 1 vs group 3)	Data available for SBP pts only
			no prophylaxis:				Nosocomial infections: 38% rifaximin vs 54% no rifaximin (P = 0.690)
			A: 1%, B: 57%, C: 43%		Group 2 (27 pts): rifaximin 400 mg TID
			rifaximin:				Isolation of bacteria resistant to III generation cephalosporin: 25% rifaximin vs 46% no rifaximin
			A: 0%, B: 33%, C: 67%		Group 3 (17 pts): systemically absorbed antibiotic prophylaxis
			systemically absorbed antibiotics:
			A:12%, B: 47%, C: 41%				Isolation of multidrug resistant bacteria: 25% rifaximin vs 9% no rifaximin

SBP: Spontaneous bacterial peritonitis; CTP: Child turcotte pugh.

Another prospective observational study reported that different bacterial species could be identified in the ascitic fluid of patients receiving rifaximin compared to those who did not receive SBP prophylaxis[69]. Indeed, Enterococci and Escherichia coli were isolated from the ascites of patients without prophylaxis and Klebsiella spp. were isolated in those on rifaximin. However, this finding had no predictive value, since the incidence of SBP was similar between the two groups.

Rifaximin, gut microbiota modulation and liver hemodynamics

Intestinal decontamination improves hemodynamics in animal models of cirrhosis by reducing endotoxemia related to bacterial translocation[70]. Similar results have also been obtained in humans[71], and have been associated with a lower incidence of complications (Table 3).

Table 3 Available studies describing the effects of rifaximin on endotoxemia in patients with advanced liver disease.

Study	Study design	No. patients	Disease severity	Treatment schedule	Results	Safety
Vlachogiannakos et al[71] 2009	Prospective	30	welve patients (40%) were Child-Pugh B and 18 (60%) Child-Pugh C	Rifaximin 1200 mg/d for 28 d	Median (range) plasma endotoxin levels decreased significantly after rifaximin administration both in systemic [1.45 (0-3.1) vs 0.7 (0-2.7), P < 0.0001] and splanchnic circulation [1.8 (0-3.4) vs 0.8 (0-2.1), P < 0.0001]. Meanwhile, the difference seen in endotoxin levels between the splanchnic and systemic circulation at day 0 (P = 0.001) was not noted at day 29 (P = 0.137)	Abdominal pain: 3%
			CTP score: A: 0%, B: 40%, C: 60%			Self-limited diarrhea: 3%
			MELD score (mean, range): 17 (11-27)		Reduction in endotoxin levels in both systemic and splanchnic circulation compared to baseline (P < 0.0001)
			B: 40%, C: 60%		Reduction in HVPG compared to baseline (P < 0.0001)
					Reduction in HVPG correlated with hepatic vein endotoxin values (P = 0.023)
Kalambokis et al[73] 2012	Prospective	9	CTP score: B: 56%, C: 44%	8-wk course of rifaximin (1200 mg/d)	Rifaximin significantly reduced plasma endotoxin levels	Not reported
				Rifaximin 1200 mg/d for 8 wk	Reduction in plasma endotoxin levels compared to baseline (P < 0.01)
Vlachogiannakos et al[72] 2012	Prospective	69	welve patients (40%) were Child-Pugh B and 18 (60%) Child-Pugh C	23 pts who achieved a decrease in HVPG after 28-d rifaximin treatment[71]	Median (range) plasma endotoxin levels decreased significantly after rifaximin administration both in systemic [1.45 (0-3.1) vs 0.7 (0-2.7), P < 0.0001] and splanchnic circulation [1.8 (0-3.4) vs 0.8 (0-2.1), P < 0.0001]. Meanwhile, the difference seen in endotoxin levels between the splanchnic and systemic circulation at day 0 (P = 0.001) was not noted at day 29 (P = 0.137)	Nausea: 9%
						Self-limited rash in the extremities: 4%
						Persistent diarrhea: 4%
			CTP score	46 cirrhotic controls
			rifaximin		Reduction in plasma endotoxin levels in both systemic and splanchnic circulation compared to baseline (P < 0.0001)
			A: 0%, B: 48%, C: 52%
			controls:		Risk of developing variceal bleeding: 35% rifaximin vs 59.5% controls (P = 0.011)
			A: 0%, B: 48%, C: 52%		Incidence of HE: 31.5% rifaximin vs 47% controls (P = 0.034)
			MELD score (mean ± SD)		Incidence of SBP: 4.5% rifaximin vs 46% controls (P = 0.027)
			rifaximin: 17.2 ± 3.6		Incidence of HRS: 4.5% rifaximin vs 51% controls (P = 0.037)
			controls: 16.6 ± 3.5
Kalambokis et al[75] 2012	Prospective, randomized, placebo-controlled	23	CTP score	13 pts: rifaximin 1200 mg/d for 4 wk	Reduction in endotoxin levels compared to control group (P = 0.005)	Not reported
			rifaximin:		Increase in mean platelets count in rifaximin group compared to controls (P = 0.006)
			A: 0%, B: 46%, C: 54%
			placebo:	10 cirrhotic pts: placebo
			A: 0%, B: 40%, C: 60%
Bajaj et al[62] 2013	Prospective	20	MELD score (mean ± SD): 9.8 ± 3.3	Rifaximin 550 mg BID for 8 wk	Reduction in plasma endotoxin levels compared to baseline (P = 0.02)	Not reported

MELD: Model for end stage liver disease; CTP: Child turcotte pugh; HVPG: Hepatic venous pressure gradient.

Twenty-three patients with decompensated alcoholic cirrhosis who achieved a reduction of hepatic venous pressure gradient (HVPG) after 28 d of rifaximin treatment were then followed-up for 5 years[72]. Compared to matched controls, rifaximin group showed a lower incidence of complications related to portal hypertension, such as variceal bleeding, HE, SBP and hepatorenal syndrome, and a better survival compared to controls. Other studies confirmed a reduction in endotoxemia, serum bilirubin, Child-Pugh and MELD scores, together with an increase in serum albumin levels after rifaximin treatment[62,73].

Rifaximin has also been demonstrated to have beneficial effects in the treatment of thrombocytopenia, the pathogenesis of which has not been completely clarified yet in cirrhotics. Endotoxemia has been advocated to contribute, together with portal hypertension, in the development of thrombocytopenia in these patients[74]; indeed, a small prelimiary study demonstrated an increase in platelets count and a decrease in endotoxin levels in 13 patients with alcoholic cirrhosis receiving rifaximin for a 4-wk course, compared to 10 controls[75]. Even if these results may encourage the use of rifaximin to minimize the complications of endotoxemia due to portal hypertension, larger, randomized, controlled studies extended also to non alcoholic liver disease are required to confirm any clinical efficacy.

RIFAXIMIN SAFETY IN ADVANCED LIVER DISEASE

Rifaximin benefits are generally paralleled by a good safety profile, since the reported rate of adverse events between treated cirrhotics and those who did not receive the drug is similar, and toxicity mainly involves the gastrointestinal tract (e.g., abdominal pain or diarrhea) (Tables 1-3). In particular, nor increase in the rate of infections neither development of antibiotic resistance are common in cirrhotics treated with rifaximin[76,77]. Although some cases of Clostridium difficile infection have been reported[51,78], the incidence is comparable to that observed in patients with advanced liver disease and is affected by confounding factors, such as age, repeated hospitalizations, ongoing therapy with proton pump inhibitors and previous courses of antibiotics[78]. Candida albicans has also been isolated in fecal samples of about 20% of cirrhotics during rifaximin treatment[65]. Probably, this finding should not be considered unequivocally harmful, since Candida organisms commonly colonize the human gastrointestinal tract as a component of the resident mycobiota[79].

Even if the limited incidence of adverse events has to be attributed to the small amount of rifaximin reaching the systemic circulation, a special consideration regarding its absorption in patients with advanced liver disease is mandatory. Indeed, due to the increased intestinal permeability, higher systemic rifaximin concentrations have been observed in cirrhotics compared to healthy subjects[80]. For this reason, although it may not represent a major problem in the short-term drug administration, the effects of a possible increase in systemic absorption should be cautiously taken into account in cases of prolonged rifaximin administration.