New Developments in Microbiome in Alcohol-associated and Non-alcoholic Fatty Liver Disease

Abstract

Alcohol-associated liver disease (ALD) and non-alcoholic fatty liver disease (NAFLD) are important causes of morbidity and mortality worldwide. The intestinal microbiota is involved in the development and progression of both ALD and NAFLD. Here we describe associated changes in the intestinal microbiota, and we detail randomized clinical trials in ALD and NAFLD which evaluate treatments modulating the intestinal microbiome including fecal microbiota transplantation (FMT), probiotics, prebiotics, synbiotics, and antibiotics. Finally, we discuss precision medicine approaches targeting the intestinal microbiome to ameliorate ALD and NAFLD.

1. Introduction

Alcohol-associated liver disease (ALD) and non-alcoholic fatty liver disease (NAFLD) are important causes of morbidity and mortality. ALD is responsible for 0.9% of all global deaths and 47.9% of all liver cirrhosis-attributable deaths [1]. The global prevalence of NAFLD is 25% [2], and non-alcoholic steatohepatitis (NASH) as a severe form of NAFLD is now the second most common indication for liver transplantation in the US after alcohol [2–4].

The intestinal microbiome harbors a multitude of bacteria, archaea, fungi, and viruses. There are as many bacteria in the gut as there are human cells [5,6], although the genome of the human microbiome is much more complex with 3 million genes than the human genome with ~23,000 genes [7]. The two most predominant major bacterial phyla are Firmicutes and Bacteroidetes with almost 90% of all bacteria; other major phyla include Actinobacteria, Fusobacteria, Proteobacteria and Verrucomicrobia [8]. It is well known that the intestinal microbiome is involved in the development and progression of various liver diseases including ALD and NAFLD [9–11]. Intestinal dysbiosis can be observed in ALD and NAFLD, which refers to a microbial imbalance with decreased concentrations of beneficial microbes and increased levels of deleterious microbes with its associated negative effects on the colonized host [9,10].

2. Microbiota changes in ALD and NAFLD

A. ALD

Cross-sectional human studies demonstrated that alcohol consumption results in intestinal dysbiosis [9,12]. Patients with ALD have lower levels of Bifidobacterium, and Lactobacillus spp. [13,14] as well as lower concentrations of Faecalibacterium prausnitzii, Ruminoccoccus spp. and Bacteroidaceae, and increased Lachnospiraceae [14,15]. Additionally, patients with alcohol associated liver cirrhosis have lower proportions of Bacteroidaceae and Prevotellaceae [15,16]. Furthermore, patients with alcohol use disorder oftentimes show small intestinal bacterial overgrowth (SIBO) [9]. The gold standard for diagnosis of SIBO is culturing of bacteria in jejunal aspirates to demonstrate at least 10⁵ colony forming units (CFUs) per mL; however, in clinical practice, hydrogen breath tests are most commonly used [9]. Testing for relative abundances of the various bacteria in SIBO would be helpful to tailor antibiotic [17] or possibly probiotic treatment [18] more accurately, as bacterial products related to SIBO contribute to liver disease in preclinical models [19,20]. However, this is difficult in clinical practice on a routine basis at the moment, as jejunal aspirates would require endoscopy. On the other hand, in the research setting, quantitative changes have been largely ignored when analyzing microbiota sequencing data, which focus on relative abundance levels.

Various factors result in the changes of the intestinal microbiome observed in ALD. They include small intestinal dysmotility [21], changes in gastric acid secretion [22], and alterations to the intestinal innate immune response [23,24]. Furthermore, multiple antimicrobial molecules are secreted by enterocytes or intestinal Paneth cells, and the antimicrobial molecules regenerating islet-derived (Reg-)3b and Reg3g were found to be decreased in the small intestine in ALD [23,24]. Conversely, experimental overexpression of Reg3g in intestinal epithelial cells in mice restricts bacterial colonization of mucosal surfaces, reduces bacterial translocation, and protects mice from ethanol-induced steatohepatitis [25]. Increased intestinal permeability is associated with ALD [23,26]. However, approximately half of precirrhotic patients with alcohol use disorder do not have a measurable gut barrier dysfunction [14]. Increased gut permeability leads to translocation of bacteria or bacterial products, e.g. lipopolysaccharide (LPS), into the systemic circulation and to the liver [26,27]. Besides alcohol itself, LPS – among other bacterial products – causes liver inflammation (via Kupffer cells), liver cell apoptosis (via hepatocytes), and liver fibrosis (via hepatic stellate cells) in ALD [9]. Prevention of bacterial translocation by intestinal decontamination with antibiotics [28,29] or interruption of inflammatory pathways in the liver such as in mice with non-functional Toll-like receptor 4 (TLR4) [30], protects against experimental ethanol-induced liver disease.

The intestinal microbiome produces ethanol itself [31,32], although it is currently unknown whether this occurs in patients with ALD. Ethanol is partially metabolized into acetaldehyde in the intestine, but this occurs mostly in the liver [29,33,34]. Acetaldehyde can disrupt tight junctions [35]. The reduced content of intestinal Lactobacillus spp. observed in ALD [13,23,24] can contribute to increased gut permeability; administration of Lactobacillus improves tight junction expression and the intestinal barrier [36,37]. Subclinical intestinal inflammation is an additional contributor to increased intestinal permeability; a mast cell membrane stabilizer prevents gut barrier dysfunction in experimental ALD [29].

Metagenomic and metabolomic studies found low luminal levels of saturated long-chain fatty acids (LCFAs) with down-regulated LCFA synthesis genes of the intestinal microbiota in experimental ethanol-induced liver disease in mice [37]. Saturated LCFAs are metabolized by commensal Lactobacillus spp. and promote their growth; the lower abundance of Lactobacillus could hence be (at least partly) due to lower levels of LCFAs, and supplementation with saturated LCFAs increases the abundance of Lactobacillus and alleviates experimental ethanol-induced liver disease [37]. Likewise, repletion of decreased luminal levels of the short-chain fatty acid (SCFA) butyrate via supplementation with glyceryl tributyrate improves intestinal permeability and experimental ethanol-induced liver disease [38,39]. We recently showed that patients with alcoholic hepatitis (AH) have elevated proportions of Enterococcus faecalis in their stool in relation to non-alcoholic individuals or patients with alcohol-use disorder [40]. In particular one characteristic of this bacterium has prognostic value: Significantly higher severity of liver disease and mortality was seen in patients with AH who had cytolysin-positive (cytolytic) E. faecalis. However, the prognostic value of presence of cytolytic E. faecalis for clinical outcome might be specific only for AH since presence of cytolytic E. faecalis in NAFLD was low and does not correlate with severity of liver disease in that entity [41].

B. NAFLD

Diet is an important environmental factor that changes the intestinal microbiome. The microbiota composition changed within 24 hours of initiating a high-fat/low-fiber or low-fat/high-fiber diet in humans [42]. It is hence not surprising that patients with NAFLD harbor a significantly different intestinal microbiome relative to healthy controls. Patients with NAFLD have significantly higher concentrations of Lactobacillus [43–45], Lactobacillaceae [43,45], Escherichia [44], as well as reduced levels of Alistipes [44,45], Oscillibacter [43,44], Blautia [45], Faecalibacterium [45], Ruminococcus [45] and Ruminococcaceae [43] compared with healthy controls. The major phylum Bacteroidetes is increased and the phylum Firmicutes is decreased in NAFLD [45]. These changes are associated with increased inflammatory cytokines (tumor necrosis factor-alpha [TNF-α], interleukin 6 [IL-6] and interferon gamma [IFN-γ]) in the intestinal mucosa in patients with NAFLD [44]. Similarly, patients with NASH exhibit elevated proportions of Bacteroidetes [31,46] and Escherichia [31] and lower levels of Firmicutes [31,47] as well as lower proportions of Alistipes, Blautia, and Ruminococcaceae [31] in comparison with healthy controls. Pediatric patients with NASH have higher concentrations of Lactobacillus and lower proportions of Akkermansia when compared with NAFLD patients without NASH [48]. The same findings were observed when comparing children with moderate to severe fibrosis versus patients with no to mild fibrosis [48]. Not only NAFLD in general [45] but also advanced liver fibrosis in particular is characterized by less abundant Ruminococcus and Faecalibacterium prausnitzii [49]. Interestingly, several well-powered studies on NAFLD with and without fibrosis or NASH found variable and sometimes contradictory results [31,46–50]. This can be due to differences in the study groups with regard to sample size, age, gender, ethnicity, medication use, documentation of liver disease, and methodology. Well-characterized larger cohorts, ideally followed longitudinally, could provide more representative and reproducible results.

Similar to ALD, NAFLD is associated with SIBO [51,52]. SIBO and gut permeability correlate with degree of hepatic steatosis but not with liver fibrosis or inflammation [51]. One study showed that SIBO as determined by breath test was present in half of all NASH patients, while it was present in 22% of healthy controls [53]. Increased intestinal permeability was found in 39.1% of NAFLD patients in a meta-analysis, compared with 6.8% of healthy controls [54]. Patients with NASH are more likely to have increased intestinal permeability compared with healthy controls [54]. Thus, only a subset of NAFLD patients have gut barrier dysfunction. A contribution to liver disease by translocating bacterial LPS, or endotoxin, can be considered, in particular in patients with higher intestinal permeability, as endotoxemia worsens with deteriorating NAFLD in humans, and it is highest in NASH [55–57]. However, inhibition of TLR4 (part of the LPS-related inflammatory pathway) with JKB-121 in a phase 2 randomized, placebo-controlled, double-blind trial (NCT02442687) did not improve NASH. Nucleotide-binding domain and leucine-rich repeat containing proteins (NLRPs) are inflammasomes which sense microbe-derived pathogen-associated molecular patterns (PAMPs) or danger-associated molecular patterns (DAMPs) generated by the host cell, which have been implicated in the development of NASH [58].

Several intestinal bacteria are known to produce ethanol [31,59]. Obese mice have higher blood alcohol levels than controls, which is likely due to increased ethanol production by the intestinal microbiome [32]. Absorbed alcohol can result in liver inflammation and fatty liver [9,12]. Interestingly, obese children with NAFLD were not found to have elevated serum alcohol levels but children with NASH were [31]. Further, fecal microbiota transplant (FMT) into mice using microbiota containing a high-alcohol-producing Klebsiella pneumoniae strain isolated from an individual with auto-brewery syndrome and progressive NASH, induced NAFLD in those mice [59]. Selective elimination of this strain prior to FMT prevented NAFLD in the recipient mice [59]. This suggests that at least in some cases of NAFLD/NASH an alteration in the gut microbiome drives the condition due to excess endogenous ethanol production. Auto-brewery syndrome (ABS), is due to endogenous ethanol synthesis either by fermenting yeasts such as Saccharomyces cerevisiae, S. boulardii, and various strains of Candida or by ethanol-producing bacteria [59–61]. Patients with auto-brewery syndrome can exhibit symptoms related to alcohol intoxication, which is exacerbated with a diet rich in carbohydrates [60]. Increased endogenous ethanol production can contribute to NAFLD in these patients. Antimicrobial treatment coupled with a low-carb diet improves symptoms of auto-brewery syndrome in some cases. Auto-brewery syndrome hence warrants further investigation. Another metabolite, choline, has been implicated in the pathogenesis of NAFLD based on animal studies [11,62], however this could not be convincingly shown in human studies [63,64].

3. Precision approaches and new developments on how to target the intestinal microbiome

A. Randomized controlled trials

Fecal microbiota transplantation (FMT)

FMT has been shown to transfer the liver phenotype [65,66]. FMT from alcohol-resistant donor mice to alcohol-sensitive receiver mice conferred resistance to experimental ALD to the latter [65]. Germ-free mice transplanted with stool from patients with severe AH, developed severe liver inflammation after ethanol feeding [66]. Additionally, ethanol-fed mice that were transplanted microbiota from patients with severe AH and were subsequently transplanted microbiota from alcoholic patients without AH had significantly less liver injury than mice that did not receive the second FMT [66]. In a pilot study in humans (Table 1), FMT from a healthy donor via a nasojejunal tube daily for 8 days to patients with steroid-ineligible severe AH (n=8) resulted in significantly better survival at 12 months relative to controls (87.5% versus 33.3%) [67]. Other associated changes included improved ascites, hepatic encephalopathy (HE), total bilirubin (average 20.5 mg/dL to 2.86 mg/dL), Child-Turcotte-Pugh (14.5 to 7.7), Model for End Stage Liver Disease (MELD, 31.0 to 12.3), and MELD-sodium scores (33.6 to 13.7) [67]. These changes were associated with reduced proportions of certain pathogenic species such as Klebsiella pneumonia (from 10% to <1% at 1 year), increased contributions of non-pathogenic species including Enterococcus villorum (from 9% to 23% at 6 months), Bifidobacterium longum (from 6% to 50% at 6 months), and Megasphaera elsdenii (from 10% to 60% at 1 year); as well as markedly modulated metabolic pathways: methane metabolism and bacterial invasion of epithelial cells were down-regulated post-FMT at 1 year, and bile secretion, carotenoid biosynthesis, and pantothenate biosynthesis pathways were increased to near normal levels following FMT [67]. Daily FMT via nasoduodenal infusion for 7 days has been found promising in another study in 51 patients with severe AH (8 treated with corticosteroids, 17 with nutritional support only, 10 with pentoxifylline, 16 receiving FMT). FMT was associated with a survival of 75% at 3 months, which was significantly higher than with steroids, nutrition, and pentoxifylline with 38%, 29%, and 30% 3-month survival, respectively [68]. A 90mL enema FMT (frozen-then-thawed) retained for 30 minutes was evaluated in a randomized controlled trial with cirrhotic patients with recurrent HE, and it was shown that FMT from a rationally selected donor reduced hospitalizations secondary to HE over the observation period of 150 days and improved cognition in these patients 15 days after FMT [69]. Of note, the FMT group was pretreated with broad-spectrum antibiotics (metronidazole, ciprofloxacin, and amoxicillin) for 5 days, whereas the placebo group was not [69]. The Shannon diversity index of the intestinal microbiome was significantly decreased after 5 days of antibiotics but returned back to the recipient’s baseline 7–15 days after FMT. FMT did not change the MELD score, aspartate aminotransferase (AST), alanine aminotransferase (ALT), or albumin levels 30 days after transplantation [69]. In a retrospective study of 10 cirrhotics with recurrent overt HE who received a one-time FMT via colonoscopy from vigorously screened patient-identified donors, 6 patients had sustained clinical response at post-treatment week 20 [70]. The FMT was performed 48 hours after completion of 5 days of broad-spectrum antibiotics; and at time of procedure, the patients were also started on 5 days of rifaximin and lactulose [70]. This was associated with significantly lower ammonia concentration, Child-Turcotte-Pugh score (9.5 to 8), and MELD score (18 to 15) at 20 weeks post FMT; however, one patient died within 2 months after the procedure [70]. Another method of FMT administration via oral capsules was trialed [71]. A one-time dose of 15 FMT capsules from a donor enriched in Lachnospiraceae and Ruminococcaceae was well-tolerated in 10 patients with cirrhosis and recurrent HE, and resulted in improved duodenal mucosal diversity, dysbiosis, and antimicrobial peptide expression, reduced LPS-binding protein (LBP), and improved cognitive performance. In a second randomized controlled trial with a one-time dose of 15 oral FMT capsules in 10 cirrhotics with HE [72], FMT was associated with reduced inflammation (as per serum IL-6 and LBP) and improved cognition both of which correlated with the abundance of Ruminococcaceae, Verrucomicrobiaceae, and Lachnospiraceae. FMT-assigned participants demonstrated higher deconjugation and secondary BA formation in feces and serum compared to baseline [72].

Table 1.

Clinical fecal microbiota transplantation (FMT) trials targeting the intestinal microbiome in alcoholic and non-alcoholic liver disease and related conditions (selected articles)

Type of Intervention	Cohort	Intervention Details	Outcome	Reference
Fecal microbiota transplantation (FMT) in severe alcoholic hepatitis (AH)	8 male patients with severe AH received FMT with matched historical controls	Daily nasoduodenal infusion of 30g of donor stool homogenized in 100mL of normal saline for 7 days	Survival 87.5% at 12 months vs 33.3% of controls; also improved ascites, hepatic encephalopathy (HE), total bilirubin, Child-Turcotte-Pugh, MELD, and MELD-sodium scores.	Philips, 2017
	Total of 51 male patients with severe AH (8 treated with corticosteroids, 17 with nutritional support only, 10 with pentoxifylline, 16 receiving FMT)	Daily nasoduodenal infusion of 30g of donor stool homogenized in 100mL of normal saline for 7 days in FMT group	Survival at 3 months was 75% for FMT, 38% for steroid, 29% for nutrition only, and 20% for pentoxifylline group.	Philips, 2018
FMT in liver cirrhosis with HE	10 male patients with liver cirrhosis and recurrent HE received FMT versus 10 matched patients who received standard of care	An enema of 90mL (frozen-then-thawed) FMT from a healthy volunteer from a universal stool bank was instilled and retained for 30min after 5 days of broad-spectrum antibiotics (metronidazole, ciprofloxacin, and amoxicillin). Placebo group was not pretreated with antibiotics	FMT reduced hospitalizations secondary to HE over the observation period of 150 days and improved cognition in these patients 15 days after FMT. It did not change the MELD score, AST, ALT, or albumin levels 35 days after transplantation.	Bajaj, 2017
	Retrospective study 10 patients with liver cirrhosis and recurrent overt HE received FMT	FMT via colonoscopy 48 hours after completion of 5 days of broad-spectrum antibiotics. At time of procedure, the patients were also started on 5 days of rifaximin and lactulose. Patients also received 1 dose of loperamide pre- and 1 dose post-procedure to retain stool	6 patients had sustained clinical response at 20 weeks post treatment. This was associated significantly lower ammonia concentration, Child-Turcotte-Pugh score (9.5 to 8), and MELD score (18 to 15) at 20 weeks post FMT; one patient died within 2 months after the procedure	Mehta, 2018
	10 patients with liver cirrhosis and recurrent HE with MELD <17 received FMT versus 10 matched patients who received standard of care	15 oral FMT capsules (4.125g) versus placebo from a single donor enriched in Lachnospiraceae and Ruminococcaceae	FMT resulted in improved duodenal mucosal diversity, dysbiosis, and antimicrobial peptide expression, reduced LPS-binding protein (LBP), and improved cognitive performance. No marked change in laboratory markers or MELD	Bajaj, 2019 (Hepatology)
	10 patients with liver cirrhosis and recurrent HE with MELD <17 received FMT versus 10 matched patients who received standard of care	15 oral FMT capsules (4.125g) versus placebo from a single donor enriched in Lachnospiraceae and Ruminococcaceae	FMT resulted in reduced serum IL-6 and LBP, and improved cognition at 4 week after FMT. FMT-assigned participants demonstrated higher deconjugation and secondary BA formation in feces and serum compared with baseline	Bajaj, 2019 (JCI Insight)
FMT in obesity/metabolic syndrome	9 male obese patients with metabolic syndrome who received FMT from lean donors (allgenic FMT) vs 9 male obese patients with metabolic syndrome who received FMT from own feces (autologous FMT)	FMT of stool sample homogenized in 500mL normal saline, administered via nasoduodenal tube after 5 hours of cleanout with polyethylene glycol solution	Significantly improved peripheral insulin sensitivity 6 weeks after FMT	Vrieze, 2012
	26 male obese patients with metabolic syndrome who received FMT from lean donors (allgenic FMT) vs 12 male obese patients with metabolic syndrome who received FMT from own feces (autologous FMT)	FMT of stool sample homogenized in 500mL normal saline, administered via nasoduodenal tube after 5 hours of cleanout with polyethylene glycol solution	Significantly improved peripheral insulin sensitivity and hemoglobin A1c at 6 weeks after FMT reverted back to baseline at 18 weeks; the improved insulin sensitivity at 6 weeks depended on decreased fecal microbial diversity at baseline	Kootte, 2017
	12 male obese patients with metabolic syndrome who received FMT from post-Roux-en-Y gastric bypass donors vs 10 male obese patients with metabolic syndrome who received FMT from metabolic syndrome donors	FMT of 200–300g stool sample homogenized in 500mL normal saline, administered via nasoduodenal tube over 30min 2 hours after completion of a cleanout with polyethylene glycol solution over 3 hours	FMT from post-Roux-en-Y gastric bypass donors resulted in better insulin sensitivity, energy expenditure and intestinal transit time in patients with metabolic syndrome 2 weeks after FMT compared with recipients of FMT from metabolic syndrome donors without gastric bypass	de Groot, 2020
	24 obese patients with mild to moderate insulin resistance who received either healthy lean donor FMT (12 patients) or placebo capsules (12 patients)	Weekly oral FMT capsules vs placebo for 6 weeks (15 capsules per week after initial 15 capsules per day on 2 consecutive days at start of trial)	Oral administration of FMT capsules in obesity resulted in gut microbiota engraftment in most recipients for at least 12 weeks; however, despite engraftment, no clinically significant metabolic effects were observed during the study	Yu, 2020

A study in patients with metabolic syndrome demonstrated that FMT from lean donors increases insulin sensitivity [73]. The authors refer the better insulin sensitivity to a regulating role of various butyrate-producing bacteria in the feces (e.g. Roseburia intestinalis) and small intestine (e.g. Eubacterium hallii) [73]. A follow-up study in patients with metabolic syndrome showed that the associated improved peripheral insulin sensitivity and hemoglobin A1c at 6 weeks after FMT from lean donors is only transient and reverts back to baseline at 18 weeks; equally, duodenal and fecal microbiota composition at 18 weeks after FMT was similar to baseline [74]. Interestingly, the altered microbiota composition with the observed changes in plasma metabolites such as γ-aminobutyric acid and improved insulin sensitivity 6 weeks after FMT depends on decreased fecal microbial diversity at baseline [74]. The same group later showed that FMT from post-Roux-en-Y gastric bypass donors to patients with metabolic syndrome resulted in better insulin sensitivity, energy expenditure and intestinal transit time in these patients compared with recipients of FMT from metabolic syndrome donors without gastric bypass [75]. Weekly oral administration of FMT capsules in adults with obesity over 6 weeks was shown to result in gut microbiota engraftment in most recipients for at least 12 weeks; however, despite engraftment, no clinically significant metabolic effects were observed during the study period [76].

Probiotics

Probiotics are living non-pathogenic microorganisms that exert beneficial effects on the host by changing the gut microbiota profile that promote anti-inflammatory effects, improve gut barrier integrity, and prevent bacterial translocation [9]. Various rodent studies showed that probiotics including Lactobacillus GG [77–80] or VSL #3 (containing the 3 genera Lactobacillus, Bifidobacterium, and Streptococcus) [81], and other bacteria such as Akkermansia muciniphila [82] reduced intestinal permeability and ameliorated liver injury in experimental ethanol-induced liver disease. Kirpich et al. [13] conducted a randomized, prospective study in ALD, and found that patients with alcohol use disorder, admitted at a Russian hospital with alcohol-related psychosis (n=66), had a lower abundance of Lactobacilli and Enterococci, and higher concentrations of E. coli (Table 2). After 5 days of probiotic treatment, the decreased levels of Bifidobacteria, Lactobacilli, and Enterococci normalized as did their liver function tests. Additionally, in a subgroup of patients with mild alcoholic hepatitis (defined as AST and ALT greater than 30 U/L with AST to ALT ratio greater than 1.0), probiotic therapy resulted in a significant ALT reduction compared with controls [13]. Loguercio et al. [83] treated alcoholic cirrhotics with VSL #3 over 3 months which lead to significantly reduced plasma levels of oxidative stress parameters, reduced inflammatory cytokine levels, and improved liver function. Similarly, a double-blind placebo-controlled study in 130 patients with liver cirrhosis who had recovered from an episode of HE (66 received probiotics, 64 received placebo) found that treatment with VSL #3 over 6 months resulted in a significantly reduced risk of hospitalization for HE, and lower Child-Turcotte-Pugh and MELD scores [84]. A recent meta-analysis of 21 trials with 1420 patients with liver cirrhosis and HE found probiotic therapy results in lower ammonia blood levels, rate of overt HE, rate of no recovery and higher quality of life but similar all-cause mortality in relation to placebo or no treatment; no significant differences were noted with regard to clinical outcomes when compared with lactulose treatment [85]. Finally, 2 recent meta-analyses of 28 [86] and 22 randomized trials [87] involving up to 1555 biopsy-proven NAFLD/NASH patients evidenced that probiotics can significantly improve body mass index (BMI), ALT, AST, alkaline phosphatase, gamma-glutamyl transferase (GGT), total cholesterol, insulin/homeostasis model assessment-insulin resistance (HOMA-IR), and leptin levels [86,87]. Tang et al. [87] also found that probiotics decrease glucose, triglyceride and TNF-α levels in NAFLD/NASH, whereas Xiao et al. [86] did not establish a significant difference in these serum parameters.

Table 2.

Clinical intervention trials targeting the intestinal microbiome in alcoholic and non-alcoholic liver disease and related conditions (selected articles)

Type of Intervention	Cohort	Intervention Details	Outcome	Reference
Probiotics	66 alcoholic patients (32 received probiotics with standard therapy, 34 received standard therapy only), 24 control patients	Probiotic therapy consisted of 0.9 × 108 colony forming units (CFUs) Bifidobacterium bifidum and 0.9 × 109 CFUs Lactobacillus plantarum 8PA3 daily for 5 consecutive days	Alcoholic patients had significantly lower AST and ALT levels after probiotic treatment than alcoholics with standard therapy only; probiotics also resulted in significantly lower ALT levels in patients with mild AH vs standard therapy	Kirpich, 2008
	22 non-alcoholic fatty liver disease (NAFLD) and 20 alcoholic liver cirrhosis (AC) patients were treated with probiotics, and compared with 36 HCV-positive patients with chronic hepatitis (20 without, 16 with cirrhosis)	Probiotic treatment over 3 months with VSL #3 with 900 billion CFUs twice daily (Streptococcus thermophilus, Bifidobacterium breve, Bifidobacterium longum, Bifidobacterium infantis, Lactobacillus acidophilus, Lactobacillus plantarum, Lactobacillus casei, Lactobacillus bulgaricus), followed by 1 month of washout	VSL #3 resulted in significantly lower plasma levels of oxidative stress parameters (MDA and 4-HNE), improved cytokine levels (TNF-α, IL-6, and IL-10), and improved liver function in patients with AC. NAFLD patients were found to have improved MDA, 4-HNE, and liver function after VSL #3	Loguercio, 2005
	130 patients with liver cirrhosis who had recovered from an episode of HE (66 received probiotics, 64 received placebo)	Probiotic treatment over 6 months with VSL #3 with 900 billion CFUs once daily	VSL #3 resulted in a significantly reduced risk of hospitalization for HE, lower Child-Turcotte-Pugh and MELD scores	Dhiman, 2014
	Meta-analysis of 21 trials with 1420 patients with liver cirrhosis and HE treated with probiotics	14 trials compared a probiotic with placebo or no treatment, and 7 trials compared a probiotic with lactulose	Probiotic therapy resulted in lower ammonia blood levels, rate of overt HE, rate of no recovery and higher quality of life but similar all-cause mortality in relation to placebo or no treatment; no significant differences were noted with regard to clinical outcomes when compared with lactulose treatment	Dalal, 2017
	Meta-analysis of 28 trials with 1555 patients with NAFLD treated with probiotics	Compared with placebo	Probiotics had beneficial effects on BMI, ALT, AST, GGT, insulin, HOMA-IR, total cholesterol but not on fasting blood sugar, lipid profiles, or TNF-α	Xiao, 2019
	Meta-analysis of 22 trials with 1356 patients with NAFLD treated with probiotics	At least majority of studies compared with placebo	Probiotics improved BMI, ALT, AST, alkaline phosphatase, GGT, total cholesterol, LDL, triglycerides, plasma glucose, insulin, TNF-α, leptin, and the degree of fat infiltration	Tang, 2019
Prebiotics	20 patients with NAFLD received probiotics, 19 prebiotics, 17 synbiotics, and 19 received a placebo	Probiotics consisted of 2 × 107 CFUs per day of Bifidobacterium longum and Lactobacillus acidophilus; prebiotic treatment was 10g of inulin daily; synbiotics were a combination of the aforementioned pro- and prebiotics. All were given for 3 months	Pro-, pre-, and synbiotic treatment resulted in significant amelioration of AST and ALT levels compared with placebo. Fatty liver grade per ultrasound also improved after pro- and synbiotic treatment	Javadi, 2017
	7 patients with NASH received prebiotics or placebo in a randomized double-blind crossover design	Prebiotics were 16g oligofructose daily for 8 weeks	Prebiotics resulted in AST improvement after 8 weeks of treatment	Daubioul, 2005.
	30 patients with NAFLD received probiotics, 29 prebiotics, and 30 placebo	Probiotics consisted of 2 × 107 CFUs per day of mixture of Lactobacillus casei, Lactobacillus rhamnosus, Lactobacillus acidophilus, Bifidobacterium longum, and Bifidobacterium breve; prebiotics were 16g oligofructose daily; treatment was given for 12 weeks	Pro- and prebiotics resulted in improved leptin, adiponectin, insulin, HOMA-IR, and quantitative insulin-sensitivity check index (QUICKI); prebiotic treatment also improved fasting blood sugar. Leptin, insulin, HOMA-IR, and QUICKI improved in both groups significantly when compared to controls	Behrouz, 2017
	14 patients with NASH (8 received prebiotics, 6 received placebo)	Prebiotics consisted of oligofructose 8g/day for 12 weeks followed by 16g/day for 24 weeks	Prebiotic supplementation improved liver steatosis relative to placebo and improved overall non-alcoholic fatty liver activity score (NAS)	Bomhof, 2019
	Meta-analysis of 9 trials with 349 patients with liver cirrhosis and minimal HE treated with probiotics, prebiotics, and synbiotics	Compared with placebo	Probiotics, prebiotics, and synbiotics all resulted in significant amelioration of minimal HE. There were no major adverse events though probiotics and synbiotics were better tolerated than the prebiotic lactulose	Shukla, 2011
Synbiotics	15 patients with liver cirrhosis received synbiotics, another 15 patients received placebo	Daily synbiotic treatment for 7 days consisted of 1010 CFUs of each of the following: Pediacoccus pentoseceus 5–33:3, Leuconostoc mesenteroides 32–77:1, Lactobacillus paracasei subspecies paracasei 19, and Lactobacillus plantarum 2592, with 2.5g betaglucan, 2.5g inulin, 2.5g pectin, and 2.5g resistant starch	Synbiotics resulted in a significantly better Child-Pugh score with a significant improvement in serum bilirubin, albumin concentrations, and international normalized ratio compared with controls	Riordan, 2007
	10 patients with alcoholic cirrhosis and 10 patients with NASH received synbiotics	Synbiotic treatment with 4 capsules daily of a mixture of Lactobacillus acidophilus, L. bifidus, L. rhamnosus, L. plantarum, L. salivarius, L. bulgaricus, L. lactis, L. casei, L. breve, fructooligosaccharides, vitamins B6, B2, B12, D3, and C, folic acid, and minerals over 2 months followed by a one-month washout	Synbiotic treatment was associated with significant improvement in ALT and GGT levels in the alcoholic cirrhosis and NASH groups; however, the benefit from treatment dissipated after a one-month washout period	Loguercio, 2002
	40 patients with NAFLD received synbiotics, and 40 received a placebo	Synbiotic treatment with a daily 500 mg capsule containing Lactobacillus casei, L. acidophilus, L. rhamnosus, L. bulgaricus, Bifidobacterium breve, B. longum, Streptococcus thermophilus, and fructooligosaccharides for 8 weeks	The ultrasound grade decreased significantly with synbiotic supplementation compared to baseline	Asgharian, 2016
	Meta-analysis of 8 trials with 415 patients with NAFLD treated with synbiotics	Compared with placebo or no treatment	Synbiotics improved AST, ALT, total cholesterol, LDL, HDL, triglyceride levels, fasting blood sugar, HOMA-IR, TNF-α, liver stiffness but not BMI or waist circumference	Liu, 2019
Antibiotics	Out of 50 patients with alcoholic liver disease (27 with cirrhosis, 23 without cirrhosis), 24 received paromomycin sulfate, and 25 received a placebo	Paromomycin 1g three times daily or placebo were given over 3–4 weeks	Paromomycin did not improve alcoholic disease in humans	Bode, 1997
	42 patients with biopsy-proven NAFLD (15 steatosis, 27 NASH) received antibiotics, no controls	Rifaximin 400mg three times daily was given for 28 days	Rifaximin improved BMI, endotoxin, IL-10, AST, ALT, GGT, and LDL levels in the NASH group, and ALT levels in the NAFLD group. It did not exert a significant effect on serum levels of TLR-4, IL-1, IL-6, IL-12, or TNF-α in either group	Gangarapu, 2015
	25 patients with NASH received antbiotics, 25 received placebo	Rifaximin 550 mg twice daily was given for 6 months	6 months of rifaximin therapy resulted in a significant reduction in HOMA-IR, ALT, AST, GGT, endotoxin, toll-like receptor-4, IL-6, TNF-α, CK-18, and NAFLD-liver fat score	Abdel-Razik, 2018
	12 patients with cirrhosis received antibiotics, 10 received placebo	Alternating 400mg norfloxacin twice daily for 15 days and 500mg neomycin three times daily for 15 days for a total of 6 months, or placebo for 6 months	Antibiotics improved liver function and Child-Pugh scores	Madrid, 2001
	20 cirrhotic patients with minimal HE received antibiotics, no control group	Rifaximin 550 mg twice daily was given for 8 weeks	Rifaximin improved cognitive function and endotoxemia	Bajaj, 2013

Prebiotics

Prebiotics are complex carbohydrates that cannot be degraded by pancreatic and intestinal enzymes but are ultimately metabolized by gut microflora, which thereby results in promotion of these bacteria and the associated beneficial effects on the host [9,24]. Administration of the prebiotics fructooligosaccharides [24] and pectin [65] results in amelioration of experimental ethanol-induced liver disease in mice. Fructooligosaccharides was additionally found to abrogate small intestinal bacterial overgrowth [24], and pectin restores the abundance of Bacteroides [65]. Prebiotic treatment of 19 patients with NAFLD with 10g inulin daily for 3 months significantly improved AST and ALT levels in a randomized controlled trial [88]. The prebiotic oligofructose showed benefit in multiple smaller clinical NAFLD trials: an 8 week treatment improved AST levels [89]; therapy over 12 weeks reduced leptin, fasting blood sugar, insulin levels, and HOMA-IR [90]; and 36 weeks were associated with a decreased non-alcoholic fatty liver activity score (NAS) on liver biopsy and higher fecal Bifidobacterium and suppressed Clostridium cluster XI and I concentrations [91]. A meta-analysis showed that prebiotics, ie. lactulose, were associated with significant improvement in minimal HE in cirrhotic patients [92].

Synbiotics

Synbiotics are a combination of pro- and prebiotics [9]. A 7-day treatment course with synbiotics (Pediacoccus pentoseceus, Leuconostoc mesenteroides, L. paracasei subspecies paracasei, and L. plantarum with bioactive, fermentable fiber: betaglucan, inulin, pectin, and starch) resulted in a significantly reduced Child-Pugh score with a significant improvement in serum bilirubin, albumin concentrations, and international normalized ratio in patients with liver cirrhosis compared with the placebo-controlled group [93]. The aforementioned meta-analysis by Shukla et al. [92] also showed that administration of synbiotics results in significantly improved minimal HE in cirrhotic patients. Synbiotic treatment with 9 probiotic strains, fructooligosaccharides and a vitamin/mineral mixture over 2 months in 10 patients with alcoholic cirrhosis and 10 patients with NASH was associated with significant improvement in ALT and GGT levels in both groups; however, the benefit from treatment dissipated after a one-month washout period [94]. A randomized controlled trial for NAFLD with 7 species of probiotic bacteria and fructooligosaccharides for 8 weeks resulted in decreased hepatic steatosis on liver ultrasound [95]. A recent meta-analysis of 8 studies involving 415 patients with NAFLD/NASH demonstrated that synbiotics can improve various markers in NAFLD including AST, ALT, total cholesterol, LDL, HDL, triglyceride levels, fasting blood sugar, HOMA-IR, serum TNF-α concentrations, liver stiffness but not BMI or waist circumference [96]. In a recent study, synbiotics (4g fructooligosaccharides twice per day, plus at least 10⁹ CFUs of Bifidobacterium animalis subspecies lactis BB-12 daily) over one year altered the fecal microbiome with higher proportions of Bifidobacterium and Faecalibacterium spp., and reductions in Oscillibacter and Alistipes spp., but did not reduce body weight, liver fat content or markers of liver fibrosis [97].

Antibiotics

Multiple studies have demonstrated the benefits from prophylactic antibiotic use in cirrhotics with and without prior history of spontaneous bacterial peritonitis (SBP) such as prevention of recurrent SBP, improved Child-Pugh status, and improved survival [98–101]. Cirrhosis related to alcohol and non-alcoholic etiologies is associated with small intestinal bacterial overgrowth which worsens with progression of liver disease [102–104]. While selective intestinal decontamination with antibiotics improved liver disease in experimental ethanol-induced liver disease in rodents [28,105], this was not observed in a randomized controlled trial in patients with ALD using the non-absorbable antibiotic paromomycin for 3 to 4 weeks [106]. Rifaximin treatment has been associated with an increase in Bifidobacterium, Faecalibacterium prausnitzii and Lactobacillus [107] that are known to have positive effects on their host. Rifaximin was also linked to improved cognitive function and endotoxemia in minimal HE, which is accompanied by alteration of gut bacterial linkages with metabolites [108]. Antibiotics such as neomycin and polymyxin B [109], or cidomycin [110] improve experimental steatohepatitis in rodents. Trials in humans with rifaximin 1200 mg/day for 28 days [111] or 1100 mg/day for 6 months [112] resulted in marked improvement of NAFLD/NASH with significantly reduced LPS and ALT levels; the 6 months long treatment also lead to amelioration of insulin resistance, inflammatory cytokines, cytokeratin 18 (CK-18), and NAFLD-liver fat score.

In summary, several randomized clinical trials have been carried out evaluating treatments modulating the intestinal microbiome including FMT, pro-, pre-, syn-, and antibiotics. Many trials have been promising. However, they have oftentimes not been confirmed independently employing the same intervention and trial design. Hence, several questions remain such as: What is the best mode of administration for FMT- via oral capsules, nasogastric/-duodenal tube, colonoscopy, or enema? Who should be the donor for FMT? How often should FMT be performed- is a one-time dose sufficient or does it have to be administered repeatedly? For probiotics-containing formulations: Which and how many strains and substrains, and how many CFUs per dose should be given, and how frequently should the doses be administered? Similarly, for prebiotic formulations: What combination of prebiotics, what dose and what frequency of administration should be recommended? The exact treatment duration of the respective modalities still remains to be determined as well.

B. Precision medicine approaches targeting the intestinal microbiome

Farnesoid X receptor (FXR) and fibroblast growth factor 15/19 (FGF15/19)

With the advent of novel technologies, new modes of therapy can be implemented (Table 3). Metagenomic and metabolomic analyses clarified that the dysregulated bile acid metabolism in ethanol-induced liver disease is (at least partially) due to decreased intestinal farnesoid X receptor (FXR) activity; as a precision medicine approach, the intestine-specific FXR-agonist Fexaramine was administered which resulted in improved gut barrier function and reduced ethanol-induced liver disease in mice [113]. Likewise, the intestine-specific FXR-agonist Fexaramine also improves experimental NAFLD with reduced hepatic steatosis and systemic ALT levels, and leads to a decreased intestinal permeability [114]. Similarly targeted approaches were carried out with regard to fibroblast growth factor (FGF)-15/19 (FGF15 is the murine, FGF19 is the human ortholog) whose transcription is regulated by intestinal FXR [113]. Lower intestinal FXR activity in experimental ethanol-induced liver disease results in lower FGF15/19 activity contributing to liver disease via increased rate-limiting enzyme of bile acid synthesis cytochrome P450 Family 7 Subfamily A Member 1 (Cyp7a1) and induction of lipogenesis in the liver [113]. Overexpression of a non-tumorigenic FGF19 variant in mice using adeno-associated viruses resulted in a markedly improved hepatic phenotype in experimental ALD by alleviation of the aforementioned detrimental metabolic changes [113]. Correspondingly, overexpression of this non-tumorigenic FGF19 variant improves experimental steatohepatitis in mice [115]. Decreased circulating FGF19 levels have also been reported in NAFLD patients in multiple studies [116–118]; and NGM282, an engineered FGF19 analogue, over 12 weeks was shown in 2 multicenter, randomized, double-blind, placebo-controlled trials to significantly improve absolute liver fat content as measured by magnetic resonance imaging (MRI)-proton density fat fraction [119], and liver fibrosis and histology in NASH [120]. The FLINT trial showed that the systemic FXR-agonist obeticholic acid over 72 weeks significantly improves liver histology in 50% of the patients with NASH (50 out of 110 patients) versus 23% in the placebo group (33 out of 141 patients) [121]. This was associated with significantly more pruritus in the treatment group with 23% compared with 6% in the placebo group [121].

Table 3.

Precision medicine approaches targeting the intestinal microbiome in alcoholic and non-alcoholic liver disease and related conditions (selected articles)

Type	Intervention Details	Reference
Farnesoid X receptor (FXR) and fibroblast growth factor 15/19 (FGF15/19)	Decreased intestinal FXR activity can be increased by the intestine-specific FXR-agonist Fexaramine which results in improved gut barrier function and alleviated liver damage in experimental ALD in mice	Hartmann, 2018
	Fexaramine also decreases intestinal permeability, hepatic steatosis and systemic ALT levels in experimental NAFLD	Fang, 2015
	Overexpression of a non-tumorigenic human FGF19 variant in mice using adeno-associated viruses leads to a markedly improved ALD by inhibiting the rate-limiting enzyme of bile acid synthesis Cyp7a1 and suppression of hepatic lipogenesis	Hartmann, 2018
	Overexpression of this non-tumorigenic FGF19 variant also improves experimental NASH in mice	Zhou, 2017
	The engineered FGF19 analogue NGM282 over 12 weeks showed significantly improved absolute liver fat content as measured by magnetic resonance imaging (MRI)-proton density fat fraction, and liver fibrosis and histology in human NASH	Harrison, 2018; Harrison, 2020
	The systemic FXR-agonist obeticholic acid over 72 weeks significantly improves liver histology in patients with NASH	Neuschwander-Tetri, 2015
Bio-engineered bacteria	Supplementation with bio-engineered Lactobacillus reuteri to produce and secrete IL-22 induces the IL-22 target gene expressing the antimicrobial protein REG3G in intestinal epithelial cells, decreases bacterial translocation, and improves experimental ALD in mice	Hendrikx, 2019
Fungal microbiome	Targeted treatment of the intestinal fungal dysbiosis in experimental ALD with oral administration of the non-absorbable antifungal agent amphotericin B reduces intestinal fungal overgrowth, decreases β-glucan translocation, and ameliorates ethanol-induced liver disease	Yang, 2017
Bacteriophages	Administration of bacteriophages to kill the pathogenic cytolytic Enterococcus faecalis decreases cytolysin in the liver and abrogates experimental ALD in humanized mice that have been colonized with bacteria from feces of patients with AH	Duan, 2019

Bio-engineered bacteria

Bio-engineering bacteria secreting beneficial metabolites in the gut is another novel precision medicine approach. Intestinal tryptophan metabolites including indole-3 acetic acid (IAA) are reduced in ethanol-fed mice and in patients with alcoholic hepatitis [122]. IAA serves as a ligand for the Aryl hydrocarbon receptor to induce IL-22 in innate lymphoid cells type 3 (ILC3); hence, IL-22 synthesis and secretion by intestinal ILC3 is decreased after ethanol feeding in mice [122]. Supplementation with bio-engineered Lactobacillus reuteri to produce and secrete IL-22 was found to induce the IL-22 target gene expressing the antimicrobial protein REG3G in intestinal epithelial cells, decrease bacterial translocation, and improve ethanol-induced liver disease in mice [122]. Supplementation with IAA resulted in similar effects [122]. Likewise, a bio-engineered E. coli Nissle (SYNB1020) that converts ammonia to l-arginine, reduced systemic hyperammonemia and improved survival in ornithine transcarbamylase-deficient spf^ash mice, and decreased hyperammonemia in a thioacetamide-induced liver injury mouse model [123]. However, it did not lower serum ammonia in patients with liver cirrhosis in a phase 1b/2a study [124].

Mycobiome

The fungal microbiome, or mycobiome, refers to the fungal community in an organism. Although its abundance is less than 0.1% of the microbial community in the human gut [125], it can play a major role in health and disease. Chronic ethanol administration increases intestinal fungal overgrowth and translocation of fungal β-glucan into the systemic circulation in mice, which via the C-type lectin–like receptor CLEC7A (also called Dectin-1) on Kupffer cells induces liver inflammation and liver disease [126]. Fungal dysbiosis can also be seen in patients with alcohol use disorder, with increasing fecal proportions of Candida with more severe liver disease [126]. Targeted treatment of the intestinal fungal dysbiosis with oral administration of the non-absorbable antifungal agent amphotericin B reduced intestinal fungal overgrowth, decreased β-glucan translocation, and ameliorated ethanol-induced liver disease in mice [126]. Candida is the most abundant genus in the fecal mycobiota and significantly more enriched in patients with alcohol use disorder and in patients with alcoholic hepatitis compared with non-alcoholic controls [127]. This is also associated with a lower diversity in the alcohol groups relative to controls. Levels of serum anti-Saccharomyces cerevisiae antibodies (ASCA) as a systemic immune response to fungal products or fungi were significantly higher in patients with alcoholic hepatitis than in patients with alcohol use disorder and non-alcoholic controls; within the alcoholic hepatitis cohort, patients with lower ASCA levels had a significantly improved 90-day survival [127]. This indicates a need for targeted antifungal treatment options in ALD. Not only the abundance of Candida but also presence of its exotoxin candidalysin is of importance in ALD. Candidalysin exacerbates ethanol-induced liver disease in mice [128]. The candidalysin precursor protein extent of cell elongation 1 (ECE1) is detected more frequently in more advanced ALD with 0% of non-alcoholic controls, 4.76% of patients with alcohol use disorder, and 30.77% of patients with alcoholic hepatitis, and it is associated with mortality [128]. Candidalysin might hence be an effective target for therapy in patients with ALD. The mycobiome has not been assessed in patients with NAFLD.

Bacteriophages

Patients with alcoholic hepatitis have increased fecal numbers of Enterococcus faecalis compared with non-alcoholic individuals or patients with alcohol-use disorder [40]. The presence of cytolysin-positive (cytolytic) E. faecalis correlates with severity of liver disease and with mortality in patients with AH. Germ-free mice were colonized with bacteria from feces of patients with AH, and the therapeutic effects of bacteriophages that target cytolytic E. faecalis were investigated. Precision editing of microbiota with bacteriophages decreased cytolysin in the liver and abrogated ethanol-induced liver disease in these humanized mice [40].

In conclusion, with more advanced technologies, more precise interventions with a better side-effect profile can be designed. FMT, pro-, pre-, syn-, and antibiotics have already been evaluated in randomized clinical trials, show efficacy and are overall well tolerated. Even more accurate and “cleaner” techniques such as bio-engineered bacteria or bacteriophages are already in the pipeline, and will change the way ALD and NAFLD/NASH are being treated in clinical practice soon. However, since various studies showed that modulations of the intestinal microbiome might be transient only with an reversion back to baseline after a few weeks to months, longitudinal and longterm studies are necessary to identify which therapy options result in longlasting microbiota changes. Furthermore, most precision medicine approaches such as use of bio-engineered bacteria or bacteriophages have not been trialed in clinical settings yet, and ideally would require large clinical multicenter studies to convincingly demonstrate their beneficial effects in the human as well. It appears possible that in the future, before initiating the aforementioned targeted therapies aimed at the microbiome, a detailed analysis of the patient’s microbiome including metabolic and metagenomic studies will be carried out to determine which precision medicine approach will be the most likely to improve the patient’s health.

Abbreviations:

AH

alcoholic hepatitis

ALD

alcohol-associated liver disease

ALT

alanine aminotransferase

ASCA

anti-Saccharomyces cerevisiae antibodies

AST

aspartate aminotransferase

BMI

body mass index

CFUs

colony forming units

CK-18

cytokeratin 18

CLEC7A

C-type lectin–like receptor

Cyp7a1

cytochrome P450 Family 7 Subfamily A Member 1

DAMPs

danger-associated molecular patterns

ECE1

extent of cell elongation 1

FGF

fibroblast growth factor

FMT

fecal microbiota transplantation

FXR

farnesoid X receptor

GGT

gamma-glutamyl transferase

HE

hepatic encephalopathy

HOMA-IR

homeostasis model assessment-insulin resistance

IAA

indole-3 acetic acid

IFN-γ

interferon gamma

IL

interleukin

ILC3

innate lymphoid cells type 3

LBP

lipopolysaccharide-binding protein

LCFA

long-chain fatty acid

LPS

lipopolysaccharide

MELD

Model for End Stage Liver Disease

MRI

magnetic resonance imaging

NAFLD

non-alcoholic fatty liver disease

NAS

non-alcoholic fatty liver activity score

NASH

non-alcoholic steatohepatitis

PAMPs

pathogen-associated molecular patterns

reg3

regenerating islet-derived 3

SCFA

short-chain fatty acid

SIBO

small intestinal bacterial overgrowth

TLR4

Toll-like receptor 4

TNF-α

tumor necrosis factor-alpha