Current Concepts in the Pathogenesis of Nonalcoholic Fatty Liver Disease

Nahum Méndez-Sánchez; Marco Arrese; Daniel Zamora-Valdés; Misael Uribe


Liver International. 2007;27(4):423-433. 

In This Article


Sanyal et al.[5] found that elevated insulin concentrations fail to suppress adipose fatty acid flux in hepatic steatosis, showing an important level of peripheral resistance to the action of insulin in these patients. It is currently accepted that the association of NAFLD and insulin resistance is almost universal. Although some patients develop hepatic steatosis and even NASH without evidence of obesity and insulin resistance,[7] after excluding patients with drug-induced NAFLD and occult alcohol consumption, the proportion of nonobese and noninsulin-resistant NAFLD patients is minimal.

The precise site of origin and magnitude of insulin resistance is not clearly understood. Impaired insulin action in insulin-sensitive tissues such as muscle, fat and liver results in a number of mutually reinforcing metabolic effects. Thus, while insulin resistance promotes fatty acid accumulation in the liver, the latter causes hepatic insulin resistance characterized by a lack of suppression of endogenous liver glucose production. Therefore, NAFLD should be looked at as a dynamic process that occurs at the crossroad between peripheral and hepatic metabolic alterations, where hepatic steatosis and insulin resistance potentiate each other.

Hepatic steatosis develops as a consequence of dysfunction of several metabolic pathways. An increase in the circulating fatty acid pool seems to be a major determinant in the pathogenesis of fatty liver. However, the role of increased activation of certain transcription factors, the action of adipokines, and derangements of hepatic fat oxidation and very low-density lipoprotein (VLDL) secretion are being increasingly recognized (see Figure 2).

Interrelationship between peripheral insulin resistance and hepatic fat deposition. High calorie consumption in sedentary and genetically susceptible individuals induces lipolysis, tumour necrosis factor-α (TNF-α) expression and hypoadiponectinemia, leading to peripheral insulin resistance and an increased circulating fatty acid pool. Hepatic fat deposition induces insulin resistance per se through abnormal intracellular insulin signalling. Both these processes lead to hepatic insulin resistance and fatty deposition with increased expression of sterol-regulatory element-binding protein-1c (SREBP-1c), CB1 and probably ghrelin, all of them capable of inducing de novo hepatic lipogenesis.

Increased Adipose Tissue-derived Fatty Acid Pool

Adipocytes are metabolically active cells that accumulate fatty acids in the form of triglycerides during excessive calorie intake.[8] Oxidation of fatty acids results in more energy than oxidation of proteins or carbohydrates. During fasting, triglycerides from adipose tissue undergo lipolysis, which releases free fatty acids (FFAs) and glycerol into the circulation for uptake by the liver. In the immediate postprandial state, the pancreas releases insulin, increasing lipogenesis and decreasing lipolysis and fatty acid mitochondrial oxidation. Normally, insulin receptor (IR) and IR substrate 2 (IRS-2, the IRS expressed in the liver) are activated through phosphorylation of its tyrosine residues.[9] However, cytokine signalling blocks tyrosine phosphorylation, and abnormal nonsignal-transducing serine/threonine phosphorylation is observed.[10] Insulin resistance develops after long-term excess energy intake in genetically susceptible sedentary individuals, leading to decreases in insulin's inhibitory effects on peripheral lipolysis and increasing FFAs availability.

Source of Hepatic Fatty Acids in NAFLD

Hepatic steatosis is characterized by the accumulation of triglycerides in both macro and micro vesicles in more than 5% of hepatocytes, predominantly in perivenular hepatocytes, with the periportal areas usually spared.[11] Triglycerides are formed by the esterification of a glycerol moiety with three fatty acid molecules. The net pool of fatty acids available for triglyceride synthesis depends on the balance between its formation and utilization. The deposition of triglycerides in hepatocytes depends on these two processes, along with triglyceride export into plasma.

In healthy individuals, the FFA pool stored in adipose tissue contributes to the majority of the fatty acids that flow to the liver during fasting (-90),[12] especially in the fasting state; however, other sources of fats are important in NAFLD, such as hepatic de novo lipogenesis (DNL) and dietary fatty acids, which can enter the liver by spillover into the plasma FFA pool or through the uptake of intestinally derived chylomicron remnants. Serum FFAs are increased in patients with hepatic steatosis and NASH.[6] FFAs are important mediators of cellular lipotoxicity, probably through lysosomal dysfunction as shown in experimental in vitro studies. Feldstein et al.[13] showed that after stimulation with fatty acids (oleate and palmitate), HepG2 suffered abnormal lipid deposition accompanied by diffusely distributed cathepsin B throughout the cytoplasm in a tumour necrosis factor-α (TNF-α)-dependent manner, demonstrating abnormal lysosomal permeabilization and an increased apoptosis rate.

The sources of hepatic triglycerides are markedly different in NAFLD patients compared with the normal population; only 60% of hepatic triglycerides arise from FFA, while 25% come from hepatic DNL and 15% from dietary fats.[14] These results suggest that hepatic DNL is paradoxically increased in NAFLD. Acetyl-CoA carboxylase (ACC) and fatty acid synthetase (FAS) are the major regulatory enzymes that drive DNL in the liver. These enzymes are upregulated by insulin and ghrelin and downregulated by adiponectin.[15,16]

As mentioned earlier, insulin resistance leads to triglyceride deposition through increased peripheral lipolysis.[17] An association between NAFLD and hepatic insulin resistance seems conceptually clear; however, a causal relationship between hepatic fat accumulation and hepatic insulin resistance has only recently been established. Samuel et al.[18] reported a specific dose-dependent relationship between hepatic fat accumulation and hepatic insulin resistance in an experimental model with no evident adipose tissue insulin resistance. Although no effect on the IR was observed, IRS-1 and IRS-2 tyrosine phosphorylation and protein kinase-B activation was blunted, probably by the activation of protein kinase C-ε, which led to the binding of c-Jun N-terminal kinase (JNK)-1 to IRS-1 and IRS-2. Recent evidence also indicates that, in the face of increased fatty acid delivery to hepatocytes, hepatic DNL is increased in NAFLD as the result of relatively preserved sensitivity of the DNL pathways to insulin and overexpression of sterol-regulatory element-binding protein (SREBP)-1c as a result of hyperinsulinemia and endocannabinoid activation.[19,20]

Role of Transcription Factors, Adipokines and Novel Metabolic Mediators

SREBP-1c. Three SREBPs have been identified: SREBP-1a and SREBP-1c, produced from a single gene through the use of alternate promoters, and SREBP-2 from a separate gene.[21] SREBP-1c is the major isoform expressed in the liver[22] and in tissues involved in energy homoeostasis, and is activated by insulin, liver receptor (LXR)-α, endocannabinoid receptor CB1 and suppressor of cytokine signalling (SOCS)-3, and inhibited by glucagons.[19,23,24,25,26] SREBP-1c expression and SREBP-1c protein levels are diminished in vitro by the addition of fish oil to the diet in experimental models.[27] SREBP-1c leads to upregulated expression of ACC, FAS and stearoyl-CoA desaturase (SCD-1).[28,29,30] Shimomura et al.[31] reported increased expression of SREBP-1c in ob/ob mice. These authors created an experimental model with overexpression of SREBP-1c in the liver.[32] The rats developed lipodystrophy, insulin resistance and hepatic steatosis. Interestingly, the same authors developed a model with overexpression of SREBP-1c only in adipose tissue, observing the development of hepatic steatosis and a three-fold increase in hepatic SREBP-1c expression.[31]

LXR-α. LXRs function as nuclear cholesterol sensors activated in response to elevated intracellular cholesterol levels (in the form of oxysterols, 22(R)-hydroxycholesterol, 24(S)-hydroxycholesterol, 27-hydroxycholesterol and 24(S), 25-epoxycholesterol) in hepatocytes and other cell types.[33] There are two LXRs, which share considerable sequence homology and respond to the same ligands.[34] However, their distribution differs - LXR-α is highly expressed in the liver, adipose tissue and macrophages, whereas LXR-ß is expressed in many tissues.[35] LXR-α induces ACC, FAS, SCD-1 and SREBP-1c transcription through retinoid X receptor (RXR)-α coactivation.[23,36,37,38] Exogenous activation of LXR-α has been shown to induce massive liver steatosis and larger VLDL with a 2.5-fold increase in their serum levels.[39] Interestingly, these phenomena are not associated with increased insulin resistance and are actually accompanied by decreased glucose serum levels through increased peripheral uptake.[40]

Adiponectin. Adiponectin, a novel adipose-specific secretory protein, is an insulin sensitizer and TNF-α antagonist, which exhibits low serum levels in obese subjects.[41,42,43] Experimental studies have demonstrated that the administration of recombinant adiponectin causes reversion of nonalcoholic fatty liver in mice, through inactivation of ACC and FAS, activation of carnitine - palmitoyl transferase (CPT)-I and peroxisome proliferators-activated receptor (PPAR)-α and TNF-α antagonism.[16,44,45] Patients with hepatic steatosis exhibit hypoadiponectinemia independent of insulin resistance, and their adiponectin serum levels correlate with the severity of fatty infiltration.[46,47,48] Adiponectin also prevents apoptotic changes and necrosis in hepatocytes after LPS-mediated injury by reducing TNF-α production and inducing PPAR-α activity.[49]

Endocannabinoid System. The endocannabinoid system plays a central role in the regulation of food intake through hypothalamic receptors.[50] In addition, CB1 knockout mice are resistant to obesity despite high calorie consumption.[51] These observations led to the demonstration of CB1 expression in adipocytes and hepatocytes.[24,52] The endocannabinoid system induces peripheral lipolysis through upregulation of lipoprotein lipase and downregulates adiponectin production.[52] CB1 knockout mice are resistant to diet-induced fatty liver. Wild-type mice with diet-induced fatty livers exhibit a four-fold higher hepatic anandamide concentration than lean controls. Anandamide activates the CB1 receptor in cultured rat hepatocytes, leading to overexpression of SREBP-1c, and increasing hepatic DNL.[24] We have observed immunoreactivity to CB1 in hepatocytes from patients with NAFLD, but not in normal human liver (unpublished data). Endocannabinoid system involvement in NAFLD is shown in Figure 3.

Activation of endocannabinoid receptors CB1 and CB2 in liver cells. CB1 induces hepatic de novo lipogenesis and increased fibrogenesis, while CB2 decreases fibrogenic response and probably reduces chronic inflammatory response. Modified from Zamora-Valdes et al. with permission of the editor.

Ghrelin. Malnourished patients exhibit hyperghrelinemia, and based on the observation that ghrelin induces ACC and FAS expression; ghrelin has been suggested as a perpetuator of fatty liver in these patients.[15,53] NAFLD patients exhibit hypoghrelinemia, and therefore systemic ghrelin is unlikely to participate in hepatic triglyceride deposition.[54] We have studied serum ghrelin and hepatic preproghrelin levels among NALFD patients in our laboratory, confirming the hypoghrelinemia reported in these patients; however, we found a nonsignificant trend towards higher levels of ghrelin mRNA in liver from patients with NAFLD compared with normal livers (0.89 ± 0.94 vs. 1.89 ± 0.59; P=NS), suggesting an autocrine or paracrine effect of ghrelin on triglyceride deposition (unpublished data). Definitively, this finding deserves further study.

Fatty acid oxidation in normal subjects mainly occurs in the mitochondria, through ß-oxidation. CPT-I is the regulatory key enzyme in mitochondrial ß-oxidation, regulating the entry of fatty acids into the outer membrane of the mitochondria after its modification by acyl-CoA synthetase. Mitochondria are responsible for the oxidation of short- to long (C8-C20)-chain fatty acids, contributing to energy generation and producing oxidative phosphorylation. Mitochondrial ß-oxidation is saturated in NAFLD because of the increased availability of fatty acids, leading to negative feedback and excessive production of acetyl-CoA, which enters the Krebs cycle increasing the formation of NADH and FADH2, and thus the delivery of electrons to the respiratory chain, a potential mechanism of reactive oxygen species (ROS) generation.[55,56] Morphological mitochondrial alterations have been reported in patients with NASH[5]; moreover, functional alterations have also been observed in the livers of obese mice and humans and in patients with hepatic steatosis.[57]

Mitochondrial Dysfunction

Patients with NASH have decreased respiratory chain complex activity[58] and an impaired ability to resynthesize adenosine triphosphatase (ATP) after a carbohydrate challenge, transiently depleting hepatic ATP.[59] Their hepatic mitochondria exhibit ultrastructural lesions, with the presence of paracrystalline inclusions in randomly distributed megamitochondria, without variation in its abundance,[5,60] with decreased hepatic mitochondrial DNA levels,[61] decreased protein expression of several mitochondrial DNA-encoded polypeptides and lower activity of complexes I, III, IV and V (ATP synthase), which are partly encoded by mitochondrial DNA, and complex II, which is encoded by nuclear DNA only.[62]

Several factors have been involved in mitochondrial dysfunction pathogenesis, such as FFAs, lipid peroxidation and TNF-α. FFAs could induce an increase in the proton leak in the respiratory chain through the dissociation of R-COOH in R-COO- and a proton, favoured by PPAR-α-induced uncoupling protein (UCP)-2 overexpression and overactivation.[63,64,65,66,67] Lipid peroxidation products alter mitochondrial DNA and react with mitochondrial proteins, decreasing respiratory chain proteins expression, respiratory chain activity and an increased exposition to reactive elements per molecule, forming ROS and leading to more lipid peroxidation and increased production of TNF-α by Kupffer cells.[68,69] TNF-blockade through specific antibodies has been shown to improve mitochondrial dysfunction in experimental NASH models, although the exact mechanism is not clear.[70]

CYP Activity

Microsomal fatty acid oxidation is carried out by the cytochrome P-450 enzymes CYP2E1 and members of the CYP4A subfamily (particularly CYP4A1). Increased availability of fatty acids is a proposed mechanism for the upregulation of CYP enzymes; however, insulin has been shown to downregulate CYP2E1 and CYP4A1,[71] while leptin probably induces CYP2E1, and the CYP4A subfamily is induced by PPAR-α.[72,73] Microsomal ω-oxidation is a minor pathway for fatty acid catabolism in normal individuals; however, significant quantities of dicarboxylic acids can be formed from ω-oxidation of long-chain monocarboxylic fatty acids in NAFLD. Microsomal oxidation of fatty acids can generate ROS and lipid peroxidation as the rate of NADPH consumption increases to reduce oxygen to superoxide and/or H2O2.

CYP2E1 is overexpressed in NAFLD.[74] CYP2E1 knockout mice develop diet-induced steatohepatitis, showing that CYP2E1 deletion neither prevented nor decreased oxidative damage. However, these mice exhibit overexpression of CYP4A1, and monoclonal antibodies against this enzyme prevent and decrease oxidative damage, showing that microsomal oxidation is crucial in experimental NASH.[75] Furthermore, CYP2E1 overexpression in isolated hepatocytes decreases tyrosine phosphorylation and increases serine phosphorylation of IRS-1 and IRS-2 in response to insulin.[76]


PPAR-α is a hepatic fatty acid-sensitive nuclear receptor that regulates the three pathways of fatty acid oxidation at the transcriptional level. Endogenous agonists include almost every major fatty acid, such as palmitic, palmitoleic, linoleic and arachidonic acids; fibrates are the only exogenous agonists employed in clinical practice.[73] PPAR-α transcription is negatively regulated by TNF-α.[77] The transcriptional activity of PPAR-α requires its binding to ligands and coactivation of RXR-α, leading to the induction of mitochondrial, peroxisomal and microsomal fatty acid oxidation enzymes, gluconeogenic enzymes (phosphoenolpyruvate carboxyl kinase), lipogenic enzymes (lipoprotein lipase) and very long-chain FAS.[73] These effects increase the availability of fatty acids for oxidation, leading to a significant reduction in hepatic fatty acids. PPAR-α has been shown to be underexpressed in experimental models of NAFLD, especially in the early phases.[78] Treatment with exogenous PPAR-α agonists prevents NAFLD in fat-fed mice and induces its regression in experimental models of NAFLD.[79]

Triglyceride deposition in the liver might be perpetuated by alterations in its export. This normally occurs by the formation of VLDL through rough endoplasmic reticulum apoprotein B-100, which binds to microsomal triglyceride transfer protein (MTP) and leads to the formation of a "pocket" that will contain triglycerides, giving birth to VLDL.[80]

Insulin inhibits apo-B100 expression, and fatty acids induce its expression.[81,82] ApoB-100 metabolism has been shown to be disturbed in patients with NAFLD, compared with lean and obese subjects without NAFLD; the apoB-100 synthesis rate is markedly reduced (probably induced by hyperinsulinemia), without alterations in albumin synthesis.[83]

MTP knockout mice exhibit microvesicular hepatic steatosis with a three-fold higher triglyceride amount than normal mice but with normal transaminases and without evidence of inflammation.[84] However, MTP knockout mice have a greater necroinflammatory response after toxin-induced liver injury.[85] Amiodarone and tetracycline had been shown to induce hepatic steatosis in humans through inhibition of MTP.[86] A polymorphism in the MTP promoter (G-493T) has been described in Japanese patients with NAFLD.[87]

NASH is characterized by the presence of macro and microvesicular steatosis along with cell ballooning, Mallory bodies, pericellular fibrosis, scattered inflammation and an increased apoptosis rate.[88] Evidence defining which patients with NAFLD will progress to NASH and which will not is scarce. Indeed, host genetic factors, including race and gender among others, influence the outcome of fatty liver but remain poorly defined. The potential role of cytokines such as TNF-α and oxidative stress is reviewed below.

TNF-α Activity

TNF-α expression is one of the earliest events after liver injury and represents a major trigger of the cytokine response. In the normal liver, commonly exposed to endotoxins and exotoxins, TNF-α-target genes are normally expressed at minimal levels. A remarkable fact about this cytokine is that it is one of the major promoters of liver regeneration. In this way, although TNF-α is capable of inducing hepatocellular necrosis, death should not be the outcome expected for normal hepatocytes.[89] Mitochondrial abnormalities, like those observed in NASH patients, provide cellular susceptibility to TNF-α-induced cytotoxic damage.[90]

The role of TNF-α in NAFLD might be a pathogenetic crossroad. A particular difference between patients with hepatic steatosis and NASH are the serum TNF-α levels, which are usually higher in patients with NASH, although the difference does not always reach statistical significance.[47,91,92] A possible explanation for this irregular relationship is that serum TNF-α levels vary during the day. Patients with NASH have intestinal permeability to bacteria and serum endotoxin levels similar to normal subjects; however, their orocecal transit time is longer and small intestinal bacterial overgrowth is found in 50% of patients, compared with ~20% of healthy controls, suggesting that in the presence of persistently or intermittently elevated TNF-α serum levels, susceptible livers might suffer a great deal of cytokine-induced inflammatory damage.[93,94]

Obesity per se induces a striking reinforcement in the hepatocellular response to endotoxin-induced liver injury.[95] In addition, after LPS-stimulation, serum TNF-α levels are significantly greater in NASH patients than in those with hepatic steatosis, despite their BMI, with a positive correlation between production of TNF-α and histological severity.[96] Crespo et al.[97] found significantly higher expression of TNF-α and TNFR1 in patients with NASH compared with obese patients without NAFLD. These results suggest that there is an increased susceptibility to TNF-α-mediated liver injury in patients with NASH, rather than a particular cause-effect relationship for TNF-α in NASH. TNF-α polymorphisms could play a role in this susceptibility trend. A polymorphism of the TNF-α promoter at position 238 is found in ~15% of the general population and in ~30% of NAFLD patients.[98] Carriers of the polymorphism have higher insulin-resistance indices, higher prevalence of impaired glucose tolerance, and triglyceride deposition despite a lower number of associated risk factors for hepatic steatosis.[98]

TNF-α activates inhibitory-κB kinase (IKK), which degrades IκB and allows activation and nuclear migration of nuclear factor (NF)-κB, leading to upregulation of interleukin-6 (IL-6), among other factors.[89] Experimental overexpression of IKK results in profound hepatic insulin resistance and steatosis, probably through abnormal serine phosphorylation of IR and IRS-2.[99,100] SOCS are feedback regulators of cytokine signalling. IL-6 induction of SOCS is associated with insulin resistance at different levels, for instance, SOCS-1 is a specific inhibitor of IRS-2, and SOCS-3 competes for docking sites with IRS-2. Both SOCS-1 and SOCS-3 can also ubiquitinate IRS-2, targeting it for proteasomal degradation.[25] SOCS-3 is also capable of activating SREBP-1c, inducing increased DNL, even in the absence of insulin signalling, by removing STAT3-mediated suppression.[25]

TNF-α also activates JNK-1. Schattenberg et al.[101] demonstrated increased JNK-1 expression in mice fed with a steatogenic diet, and increased adiponectin and resistance to NAFLD in JNK-1 knockout mice. Adiponectin expression correlates inversely with serum TNF-α levels; therefore, TNF-α might induce a loss of adiponectin-mediated suppression of ACC and FAS, with decreased CPT-I activity.[44]

Oxidative Stress

Oxidative stress results from an imbalance between prooxidant and antioxidant compounds, and leads to the oxidative damage of lipids, proteins and nuclear material. Oxidative stress induced by alterations in the mitochondrial respiratory chain, peroxisomal fatty acid oxidation and CYP activity are by far the more likely factors involved in progression to NASH.[102] The mechanisms involved in ROS production have been described above (see Fatty acid utilization) and are summarized in Figure 4.

Oxidative stress in nonalcoholic fatty liver disease. Increased fatty acid availability activates mitochondrial, peroxisomal and microsomal oxidation, leading to increased production of reactive oxygen species, which in turn induces lipid peroxidation (with MDA and HNE production), protein denaturation and DNA damage. MDA and HNE induces: (1) apoB proteolysis, decreasing hepatocellular triglyceride export, favouring steatosis; (2) decreased glutathione available for ROS degradation and (3) upregulated TNF-α and TGF-ß, inducing hepatocellular necroinflammation and stellate cell-induced fibrogenic activity. FA, fatty acids; ROS, reactive oxygen species; MDA, malondialdehyde; HNE, hydroxynonenal; TG, triglyceride; TNF-α, tumour necrosis factor-α; TGF-ß, transforming growth factor-ß.

ROS can attack polyunsaturated fatty acids (PUFAs) and initiate lipid peroxidation, resulting in the formation of long half-life aldehyde by-products such as 4-hydroxy-2-nonenal (HNE) and malondialdehyde (MDA).[55] These products have been shown to propagate oxidative damage and to induce multiple cellular alterations, including microvesicular steatosis.[103,104] HNE and MDA induce Apo-B proteolysis, reducing triglyceride export through VLDL formation.[105] They impair nucleotide and protein synthesis, deplete glutathione and increase TNF-α production, leading to increased necroinflammatory activity and fibrogenesis through the enhancement of hepatic production of transforming growth factor (TGF)-ß1, which activates hepatic stellate cells into collagen-secreting myofibroblasts.[106]


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.