Molecular Mechanisms: Connections between Nonalcoholic Fatty Liver Disease, Steatohepatitis and Hepatocellular Carcinoma

Younossi et al. examined 8,515,431 patients from 22 countries in 85 studies and showed the global prevalence of NAFLD to be 25.2%, with the highest prevalence in the Middle East and South America and the lowest prevalence in Africa [18]. The metabolic comorbidities associated with NAFLD are hyperlipidemia (69.2%), obesity (51.3%), metabolic syndrome (42.5%), hypertension (39.3%) and type 2 diabetes (22.5%). The obesity epidemic is closely associated with the rising prevalence and severity of NAFLD. Obesity has been linked not only with simple steatosis but also with advanced diseases such as NASH [19]. Insulin resistance is a driver of the progression of NAFLD, and diabetes mellitus is highly predictive of the progression of simple steatosis to advanced liver fibrosis [20]. The hyperinsulinemia induced by peripheral insulin resistance can stimulate lipogenesis in the liver [21]. Insulin resistance can also lead to an increased release of free fatty acids by adipose tissue [22].

Cytokeratin-18 (CK-18) fragments generated by caspase 3 are an independent predictor of NASH in patients with NAFLD [23]. Caspase-3-generated CK-18 fragments are markedly increased in the blood of patients with definitive NASH [23]. The activation of caspases and apoptosis-related molecules is frequently observed in the liver of NASH patients [12]. NAFLD progresses to NASH in response to elevated endoplasmic reticulum (ER) stress [12,24]. ER stress can drive lipogenesis and steatohepatitis via caspase-2 activation of site 1 protease [24].

Bcl-2 family members also play a role in the progression of NASH. Hepatocellular-localized endothelial nitric oxide synthase (eNOS) deletion exacerbates Western-diet-induced NASH through a decrease in PPARγ coactivator-1α, mitochondrial transcription factor A, Bcl-2-interacting protein-3 (BNIP3) and 1A/1B light chain 3B (LC3B) [25]. Activation of JNK also contributes to the apoptosis that occurs in NAFLD and NASH [12].

Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)-producing natural killer (NK) cells promote a proinflammatory environment in the early stages of NAFLD [12,26]. In patients with either NAFLD or NASH, apoptotic hepatocytes stimulate immune cells and hepatic stellate cells, which contribute to the progression of hepatic fibrosis [12]. Inflammasomes and cytokines can induce apoptosis and contribute to the development of NAFLD and NASH [12]. Oxidative stress, ER stress and autophagy also contribute to the progression of NAFLD and NASH through a modification of hepatocyte apoptosis [12]. Glucose metabolism and lipid metabolism are also involved in the progression of NAFLD and NASH [12]. Thus, apoptosis is an important factor in the pathogenesis of these two conditions.

Intercellular adhesion molecule-1 (ICAM-1)-positive hepatocytes have been observed in NASH patients and reportedly localize in areas with microvesicular fat, although ICAM-1-positive hepatocytes have not been reported in non-NASH patients [27]. Soluble ICAM-1 (sICAM-1) is reported to be significantly higher in NASH patients than in non-NASH patients. Patients with a NAFLD activity score (NAS) greater than four have a higher number of areas of CD68-positive cells and forkhead box P3 (Foxp3)-positive cells than non-NASH patients [27]. In liver tissue with NASH, hepatocytes with microvesicular steatosis express more inflammatory markers than hepatocytes in non-NASH tissues, and in this liver tissue, an increased number of CD68+ cells, such as macrophages and Foxp3+ cells, including regulatory T cells (Tregs), is reportedly observed [27]. There are two types of macrophages, M1 (or “classically activated”) and M2 (or “alternatively activated”). They play an important role in humoral immunity and the response to pathogens, and have pro- and anti-inflammatory properties, respectively [13]. M1-skewed inflammation accompanies diet-induced obesity in the liver and adipose tissue [28]. Liver-resident macrophages and Kupffer cells are lost early after disease onset, followed by a robust infiltration of Ly-6C+ monocyte-derived macrophages in a mouse model fed a methionine-choline-deficient diet [29].

Excess iron and hepatic changes in iron metabolism are associated with the development of NAFLD and NASH [30]. Dietary iron excess leads to hepatic oxidative stress, immune cell activation, hepatocellular ballooning injury and NASH in genetically obese mice [31]. The NOD-like receptor (NLR) family pyrin domain-containing 3 (NLRP3) inflammasome is an intracellular multiprotein complex involved in the production of mature interleukin 1-beta (IL-1β) that induces metabolic inflammation [32]. The NLRP3 inflammasome also plays a crucial role in the progression of NASH in mice. Henao-Mejia et al. demonstrated that the NLRP6 and NLRP3 inflammasomes and the effector protein IL-18 negatively regulate NAFLD/NASH progression [16]. Similar to Asc−/− and caspase-1−/− mice, MCDD-fed Il18−/− animals featured a significant exacerbation of NASH severity [16].

Tim-3 plays a role as an immune checkpoint in the regulation of both adaptive and innate immune cells, including macrophages, and is involved in chronic liver diseases. Tim-3 serves as an important predictor in methionine-deficient-diet (MCD)-induced NASH by regulating reactive oxygen species (ROS) and the associated pro-inflammatory cytokine production in macrophages [33]. TNF receptor superfamily member 4 (TNFRSF4/OX40) deficiency suppresses Th1 and Th17 differentiation, and TNFRSF4 deficiency in T cells inhibits monocyte migration, antigen presentation and M1 polarization [34]. The soluble TNFRSF4 plasma levels are positively associated with NASH. TNFRSF4 is a key regulator of both intrahepatic innate and adaptive immunity and plays a role in the development of NASH [34].

These results indicate that both innate and adaptive immunity are involved in the pathogenesis of NASH.

Wigg et al. reported that patients with NASH have a higher prevalence of small intestinal bacterial overgrowth (SIBO), according to the ¹⁴C-D-xylose-lactulose breath test, and they also have higher TNF-α levels than patients without NASH [35]. Nair et al. reported that higher breath ethanol concentrations are observed in obese female subjects than in nonobese females, suggesting that intestinally derived ethanol may contribute to the pathogenesis of NASH [36]. Wu et al. reported that SIBO may decrease small intestinal motility in NASH rats [37]. They also demonstrated that the effects of cidomycin significantly lowered the serum ALT, AST and TNF-alpha levels in NASH rats [37]. Although the lipopolysaccharide binding protein (LBP) levels and TLR-2 expression were found to be similar, TLR-4/MD-2 expression on CD14-positive cells was higher in the NASH patients than the control subjects [38]. NASH patients have a higher prevalence of SIBO, which is associated with an enhanced expression of TLR-4 and release of IL-8 [38]. In the small intestine and liver of the rat NASH model, it is possible that enhancement of the innate immune response through the TLR4 signal leads to increased production of TNF-α [39]. Wada et al. reported that IP-10 was higher in NAFLD models than in controls and higher in NASH models than in both NAFLD models and controls [40].

In one experiment, the intestinal mucosa barrier function was shown to have changed, resulting in the progression of nonalcoholic steatohepatitis in rats [15]. Intestinal mucosa barrier malfunction may play a role in NASH. Increased intestinal permeability and tight junction alterations are also observed in patients with NAFLD. Miele et al. demonstrated that NAFLD is associated with increased gut permeability and that this abnormality is related to the increased prevalence of SIBO [41]. F11r encodes junctional adhesion molecule A (JAM-A), and its defect leads to intestinal epithelial permeability [42]. A diet high in saturated fat, fructose and cholesterol (a high-fat and cholesterol-deficient (HFCD) diet) led to a significant increase in inflammatory microbial taxa in F11r(-/-) mice compared with control animals. The administration of oral antibiotics or sequestration of bacterial endotoxins with sevelamer hydrochloride reduced mucosal inflammation and restored normal liver histology in F11r(-/-) mice fed an HFCD diet [42]. Both the intestinal and hepatic CYP2E1 induced by binge alcohol consumption seem important in binge alcohol-mediated increased nitroxidative stress, gut leakage and endotoxemia. These factors may alter fat metabolism and inflammation, contributing to hepatic apoptosis and steatohepatitis [43].

Elinav et al. demonstrated that a deficiency of NLRP6 in mouse colonic epithelial cells resulted in the reduction of IL-18 levels and an alteration in the fecal microbiota characterized by expanded representation of the bacterial phyla Bacteroidetes (Prevotellaceae) and TM7 [43]. NLRP6 and NLRP3, both of which include ASC and caspase-1 and involve IL-18 but not IL-1R, result in the development of an altered transmissible, colitogenic intestinal microbial community [44]. Henao-Mejia et al. observed that the increased severity of NASH in Asc- and Il18-deficient mice is transmissible to cohoused wild-type mice [16]. They also demonstrated diet- and cohousing-associated changes in gut microbial ecology by 16S rRNA sequencing [16]. The levels of TLR4 and TLR9 agonists, but not TLR2 agonists, were markedly increased in the portal circulation of MCDD-fed wt (Asc−/−) and Asc−/− mice compared to wt controls, suggesting that TLR4 and TLR9 agonist efflux from the intestines of inflammasome-deficient mice or their cohoused partners through the portal circulation to the liver triggers TLR4 and TLR9 activation, that in turn results in enhanced progression of NASH [16]. Microbiota-induced subclinical colon inflammation may be a determining factor in both the rate of TLR agonist influx from the gut and NAFLD/NASH progression. In comparison with those without SIBO, patients with SIBO had significantly higher endotoxin levels and higher CD14 mRNA, nuclear factor kappa B mRNA and TLR4 protein expression [45]. Patients with NASH had significantly higher endotoxin levels and more intense hepatic TLR4 protein expression than patients without NASH [45]. Although LPS loading resulted in hepatic inflammation in both maintenance-food- and high-caloric-diet-fed mice, high-caloric-diet feeding was more crucially important in the progression of NAFLD during the triggering phase [46].

The bile acids, gut microbiota and metabolome that contribute to the regulation of glucose and lipid metabolism were affected by the Myd88 level in hepatocytes, resulting in altered susceptibility to obesity, such as induction of type 2 diabetes and NASH, in both mice and humans [47]. Lactobacillus paracasei attenuated hepatic steatosis by M2-dominant Kupffer cell polarization in a NASH mouse model [48].

Patients with NASH were shown to have a lower percentage of Bacteroidetes (Bacteroidetes to total bacteria counts) than those with simple steatosis or living liver donors (representing healthy controls) [49]. The inverse and diet-/BMI-independent association between the presence of NASH and the percentage of Bacteroidetes in the stool also supports the notion that the intestinal microbiota may play a role in the development of NAFLD [49].

Yuan et al. reported that a greater abundance of Gram-negative bacteria in the gut microbiome was observed in obese and NASH patients than in normal subjects [50]. Among NASH patients, serum endotoxin is not correlated with disease severity, suggesting that endotoxemia is not required in the pathogenesis [49]. Subjects with NAFLD had slightly higher lipopolysaccharide-binding protein (LBP) (P < 0.001) and EndoCab immunoglobulin G (IgG) (P = 0.013) levels than those without NAFLD [51]. These endotoxin markers are associated with NAFLD in the general population but do not have a major effect on NASH and/or fibrosis. People with modest alcohol consumption may have lower serum endotoxin levels than those with heavy alcohol consumption.

A meta-analysis of four randomized trials involving 134 NAFLD and NASH patients revealed that probiotic therapies can reduce AST, ALT, total cholesterol and TNF-α, and improve insulin resistance in these patients [52]. Modulation of the gut microbiota may afford a novel treatment for NAFLD [52]. Mosapride citrate, a gastroprokinetic agent, showed a protective effect against MCD diet-induced NASH development in a mouse model, with a possible involvement of increased fecal lactic acid bacteria, protection against colon inflammation and an elevated plasma-glucagon-like peptide-1 (GLP-1) concentration [53].

Nakano et al. observed that a history of appendectomy was significantly more frequent in NAFLD patients with advanced fibrosis than in those without fibrosis [54]. Appendectomy may thus be one of the risk factors for advanced fibrosis in NAFLD [54]. In adult patients with NAFLD, dysbiosis is associated with altered bile acid homeostasis, which renders patients at increased risk of hepatic injury [55]. Gut microbiota alterations are closely correlated with dysregulated bile acid levels in the liver and feces of streptozotocin- and high-fat-diet-induced NASH–HCC mice [56]. Calcium reduces liver injury in mice on a high-fat diet, as well as inducing alterations in the microbial and bile acid profiles [57]. Intestinal bacteria may play an important role in the development of NAFLD and NASH [58,59,60,61,62,63,64,65], and they may constitute a therapeutic target. Further studies will be needed to determine the therapeutic utility [66,67,68,69].

To identify DNA sequence variations that contribute to interindividual differences in NAFLD, Romero et al. performed a genome-wide association study of nonsynonymous sequence variations (n = 9229) in a multiethnic population [85]. One PNPLA3 variant (rs738409; I148M) was significantly associated with increased hepatic fat levels (P = 5.9×10⁻¹⁰) and with hepatic inflammation (P = 3.7×10⁻⁴) [84]. This variant was most common in Hispanics, who are most susceptible to NAFLD. The hepatic fat content was > twofold higher in patients with homozygous PNPLA3-148M than in noncarriers [85]. PNPLA3 encodes a 481-amino-acid protein of unknown function that belongs to the patatin-like phospholipase family [86].

The identification of a second allele of PNPLA3 (i.e., S453I) that was independently associated with hepatic fat content further supports a role for PNPLA3 in determining hepatic triglyceride levels [85]. Yuan et al. also reported that the PNPLA3 variant (rs2281135, T) is associated with the plasma levels of certain liver enzymes [87]. Obese Southern Europeans carrying the PNPLA3-148M allele have increased indices of liver damage independent of proxy measures of insulin resistance [88]. Kawaguchi et al. also found that Matteoni type 4 NAFLD is both a genetically and clinically different subset from the other spectra of the disease and that the PNPLA3 variant is strongly associated with the progression of NASH in Japanese patients with NAFLD [89,90].

Singal et al. performed a meta-analysis of nine studies with 2937 patients and found that PNPLA3 was associated with an increased risk of HCC in patients with cirrhosis (OR 1.40, 95% CI 1.12–1.75) [91]. Upon subgroup analysis, PNPLA3 was found to be an independent risk factor for HCC in patients with NASH or ALD-related cirrhosis (OR 1.67, 95% CI 1.27-2.21). Liu et al. also reported that bearing the PNPLA3 rs738409 c.444C>G minor allele (encoding the I148M variant) was associated not only with a greater risk of progressive steatohepatitis and fibrosis but also with HCC in patients with NAFLD [92]. Ueyama et al. revealed that the PNPLA3 G allele and JAZF zinc-finger 1 (JAZF1) rs864745 G allele are associated with non-HBV and non-HCV-related HCC in Japanese patients with type 2 diabetes mellitus [93].

NAFLD is characterized by a massive accumulation of lipid droplets (LDs) [94]. Su et al. identified 17β-HSD13 as a pathogenic protein involved in the development of NAFLD [94]. One splice variant (rs72613567: TA) in HSD17B13 that encodes the hepatic lipid droplet protein hydroxysteroid 17-beta dehydrogenase 13 was found to be associated with reduced levels of ALT and AST [95]. This variant was associated with a reduced risk of NAFLD (by 17% (95% CI, 8 to 25) in heterozygotes and by 30% (95% CI, 13 to 43) in homozygotes) and nonalcoholic cirrhosis (by 26% (95% CI, 7 to 40) in heterozygotes and by 49% (95% CI, 15 to 69) in homozygotes) [95].

Chen et al. reported that lower HSD17B13 in peritumoral tissues was associated with worse recurrence-free survival and overall survival in HCC patients, although they mainly studied viral-hepatitis-associated HCC [96]. They also showed that HSD17B13 delayed G1/S progression in HCC cells [96]. Patients carrying rs72613567, a splice variant with an adenine insertion (A-INS), exhibited HSD17B13 levels that decreased proportionally with allele dosage [97]. Whole-transcriptome genotype profiling also showed an overrepresentation of immune-response-related pathways. The HSD17B13 rs72613567 A-INS allele reduces the risk of NASH and progressive liver damage [97].

HSD17B13 has retinol dehydrogenase (RDH) activity, with this enzymatic activity dependent on lipid droplet targeting and cofactor binding sites [98]. May et al. reported an association of variants of HSD17B13 with certain specific features of NAFLD histology and identified the enzyme as a lipid droplet associated RDH [98]. The HSD17B13 minor allele rs6834314 was significantly associated with increased steatosis but decreased inflammation, ballooning, Mallory–Denk bodies and liver enzyme levels in 768 adult Caucasians with biopsy-proven NAFLD, as well as cirrhosis, in the general population [98]. Yang et al. reported that the HSD17B13 rs72613567 loss of function variant protected against HCC development in patients with ALD [99].

Namjou et al. reported that three SNPs in the PNPLA3-SAMM50 region, rs738409, rs738408 and rs3747207, displayed the strongest association with NAFLD [100]. They also identified certain novel loci for NAFLD disease severity, including one locus near IL17RA (rs5748926, p = 3.80 × 10^{− 8}) that was associated with the NAS score, and another near ZFP90-CDH1 that was associated with fibrosis (rs698718, p = 2.74 × 10^{− 11}) [100].

Klotho beta (KLB) mediates the binding of fibroblast growth factor 21 (FGF21) to the FGF receptor (FGFR). FGF21-KLB-FGFR signaling plays an important role in multiple hepatic metabolic systems [101]. Ji et al. reported that KLB SNPs were related to obesity and hepatic inflammation, which may be involved in the pathogenesis of NAFLD [101]. Xu et al. reported that the TM6SF2 E167K variant was associated with NAFLD in Northeast China [102]. A meta-analysis showed a correlation between the rs1501299 polymorphism in the adiponectin gene and the risk of NAFLD [103]. Kawaguchi et al. found the rs17007417 in dysferlin (DYSF) (p = 5.2 × 10⁻⁷, OR (95%CI) = 2.74 (1.84–4.06)) to be a SNP associated with NASH–HCC [104]. Rs17007417 in DYSF was significantly different in NASH–HCC cases compared with controls, as well any of the four Matteoni subgroups [90]. Further studies will be needed to determine the effects of these genetic variants on hepatocarcinogenesis.