Pathogenesis of Nonalcoholic Fatty Liver Disease

Introduction

Nonalcoholic fatty liver disease (NAFLD) refers to the ectopic accumulation of fat in the liver (also known as steatosis ) that cannot be attributed to alcohol consumption. Minor amounts of fat deposition in the liver can occur in a healthy subject, but when it occurs in at least 5% of hepatocytes it is considered pathologic. Under the umbrella of NAFLD is a wide spectrum of liver pathology with very different outcomes. When steatosis is the only abnormal finding, the condition is dubbed simple steatosis or, more precisely, isolated steatosis . Classically, isolated steatosis has been considered to be a benign, nonprogressive condition. Conversely, nonalcoholic steatohepatitis (NASH) describes the condition wherein inflammatory activity and cell injury (manifested by hepatocyte ballooning and/or cell death) accompany steatosis. NASH has the potential to progress to liver cirrhosis and its complications.

Classic dogma posits that these dichotomous outcomes of NAFLD reflect interindividual differences in exposures and/or sensitivity to secondary stresses that fatty hepatocytes experience. For many years, this concept was referred to as the two-hit hypothesis . According to this model of NAFLD pathogenesis, the liver must be subjected to two insults for the full spectrum of NAFLD to develop. The first insult leads to steatosis, which sensitizes the liver to a second insult that leads to inflammation, cell death, and fibrosis. This interesting concept was supported by seminal papers which showed that rodents that develop steatosis in the context of obesity, or as a consequence of impaired export of lipids from the liver, are exquisitely sensitive to endotoxin exposure and tumor necrosis factor α (TNF-α)-induced liver injury, and quickly develop steatohepatitis. It was postulated that fatty hepatocytes are unusually vulnerable to a second, proinflammatory “hit” that challenges hepatocyte viability. Subsequent studies revealed that fatty hepatocytes exhibit a number of stress responses (e.g., oxidative stress and endoplasmic reticulum [ER] stress) that might compromise their ability to withstand an added insult.

More recently, the linearity of these concepts has been questioned because it has not been rigorously proven that isolated steatosis always precedes NASH in patients, although some paired biopsy studies have shown progression from isolated steatosis to NASH. This issue will be difficult to resolve because histologic criteria for defining NASH are still debated, and show high interobserver variability. Also, a well performed high-quality liver biopsy corresponds to less than 1/50,000 of the total liver volume, and NAFLD features are not homogeneous across the parenchyma. Thus potential sampling error confounds interpretations of paired liver biopsies.

The role of steatosis per se in NAFLD pathogenesis has also been questioned. Studies in animal models have revealed that not all fat behaves similarly. Triglycerides (the type of fat that accumulates in hepatocytes in livers with isolated steatosis) are now accepted as being the “good guys” that prevent accumulation of toxic fat. Hence, an alternative theory to explain NAFLD pathogenesis is that toxic fat accumulation in the liver occurs in parallel to neutral triglyceride accumulation (i.e., steatosis), and the toxic fat (but not the triglyceride) leads to NASH. According to this theory, isolated steatosis and NASH are considered two independent, eventually parallel, conditions. One problem with this concept, however, is that it fails to explain why the vast majority of patients with isolated steatosis never develop progressive liver disease. The two-hit hypothesis, on the other hand, does not exclude the possibility that “second hits” are lethal to triglyceride-laden hepatocytes, at least in part, because they promote accumulation of toxic fat. Indeed, as mentioned earlier, it has become clear that fatty hepatocytes are coping with various challenges (e.g., oxidative stress, ER stress, perturbed autophagy, and deregulated cell death pathways) that might limit their ability to withstand superimposed threats to their viability (also known as second hits, such as toxic fats). Finally, as initially articulated, the two-hit model for NAFLD progression did not attempt to explain the basis for the first hit, that is to say steatosis. It has since become clear that inflammation can both precede, and promote, steatosis. Moreover, some conditions (e.g., insulin resistance) are able to promote both steatosis and inflammation. Given all of this, it seems that NAFLD pathogenesis and progression most likely reflect the interplay between potentially hepatotoxic exposures and hepatic compensations for these stresses. Adaptations that are necessary to survive an initial stress might inadvertently enhance vulnerability to certain types of subsequent stress. Whether or not hepatocyte death occurs depends on whether or not the liver is able to adapt to the first hepatotoxic stress or becomes exposed to the subsequent stress that is now life-threatening. According to this view, steatosis itself is an early biomarker of hepatotoxic stress. NASH ensues immediately when the adaptive responses to the initial stress are insufficient to maintain hepatocyte viability. Because steatosis is far more prevalent than NASH, the resiliency of hepatocytes apparently permits them to adapt and survive most stresses that result in steatosis. Nevertheless, such adaptations exact a price and “adapted” hepatocytes (marked by their accumulation of neutral lipid) become more vulnerable to other insults that are typically well tolerated by nonsteatotic (i.e., nonstressed) hepatocytes (e.g., ischemia reperfusion, TNF-α). In this context, exposure to the secondary stressors is required to provoke NASH.

The outcomes of NASH are also variable. The disease can remain relatively stable for years, progress to cirrhosis, or regress to isolated steatosis. These heterogeneous outcomes are presumed to reflect differences in the consistency and intensity of exposure to hepatotoxic stresses and variability in the capacity to replace (i.e., regenerate) dead hepatocytes. NASH resolves when injurious stresses subside and all dead hepatocytes have been effectively regenerated. It remains stable when regenerative processes generally match the demand for hepatocyte replacement, and progresses when regeneration is unable to keep pace with cell death. Like potentially hepatotoxic insults and the liver's ability to adapt to such stresses, the effectiveness of regeneration also varies among individuals, and within any given individual over time, because regenerative capacity itself is impacted by dynamic processes (e.g., aging and overall health). Indeed, NAFLD is now seen as one component of a systemic disease that encompasses a spectrum of dynamic metabolic disturbances in which the adipose tissue plays a central role. As such, NAFLD is a biomarker of sick adipose tissue. In some patients with NASH, however, pathologic processes in the liver self-perpetuate, progressing independently of adipose tissue pathology and despite reductions in hepatic fat content. Recent natural history studies of human NAFLD provide some insight into the mechanisms that account for the bewildering complexity this disease. In NAFLD patients, liver fibrosis severity was identified as the only histologic variable that independently predicted clinically meaningful liver outcomes (i.e., death from liver disease, liver transplantation, or liver-specific morbidities). Fibrosis severity, in turn, was predicted by only two variables, namely patient age and severity of liver injury. Thus NAFLD-related liver damage ultimately reflects the balance between forces that threaten hepatocyte survival and those that protect it and an individual's ability to regenerate hepatocytes that die when prosurvival mechanisms become overwhelmed.

This chapter reviews the mechanisms that lead to steatosis, NASH and NASH-associated liver fibrosis/cirrhosis. A clear understanding of physiopathology is paramount for a better clinical management of this disease and to delineate better therapeutic approaches. Enormous progress has occurred since NAFLD was first described 35 years ago, but gaps in knowledge still preclude the development of effective treatment for NAFLD/NASH.

Making Foie Gras

Hepatic steatosis and liver accumulation have been used indiscriminately. However, although triglycerides are the main store of lipids in the hepatocyte, other forms of lipids can accumulate in the liver, such as free fatty acids (FFAs), diacylglycerol, free cholesterol, cholesterol esters, ceramides, and phospholipids. Variations in the concentrations of these other lipids can occur in the context of triglyceride accumulation, suggesting that “simple” hepatic steatosis is really a heterogeneous condition. The availability of FFAs will determine the amount of triglycerides synthesized. In NAFLD, FFAs reach the liver from three main sources: 60% from systemic circulation as nonesterified fatty acids (NEFAs), 15% from portal blood as chylomicron lipoproteins assembled from dietary fat, and 25% from de novo fatty acid synthesis/lipogenesis in the liver ( Fig. 25-1 ). Adipose tissue is the main source of lipids that accumulate in fatty livers. Hence, an adipocentric view of the pathogenesis of NAFLD is replacing a hepatocentric one. Circulating NEFAs enter the liver in a nonregulated way, and consequently liver uptake is a function of FFA plasma concentrations. Adipose tissue is responsible for 80% of FFA plasma pool in the fasting state and 60% in the postprandial state when fat from diet gains importance. Dysfunctional (sick) adipose tissue “metastasizes” fat to the liver. Increased spill out of NEFAs from the adipose tissue occurs in two situations: when the adipose tissue is intrinsically sick and cannot adequately store energy (e.g., as occurs in lipodystrophies); or when adipose tissue sickness is caused by energy surplus that surpasses the natural storage capacity of adipose depots. Insulin resistance (IR) increases the release of NEFAs from the adipose tissue into systemic circulation. Decreased insulin signaling fails to suppress the hormone-sensitive lipase in adipose depots, increasing release of fatty acids (FAs) from triglycerides stored peripherally. Also, by decreasing glucose uptake in the adipose tissue, insulin resistance depletes glycerol-3-phosphate, a factor which is necessary for reutilization of FAs in triglyceride synthesis.

The second source of hepatic fat in NAFLD is de novo lipogenesis. In a healthy lean subject, de novo lipogenesis accounts for 5% of hepatic lipids in the fasting state, and increases in the postprandial period, when availability for lipogenic precursors increases. In a hyperinsulinemic patient with NAFLD, de novo lipogenesis increases threefold in the fasting state, but does not further increase after meals. This suggests that lipogenic enzymes are already maximally up-regulated during fasting in the context of insulin resistance. Insulin resistance–related induction of lipogenic enzymes can be explained by hyperinsulinemia and hyperglycemia. Insulin up-regulates expression of sterol regulatory element–binding protein (SREBP-1c) and peroxisome proliferator–activated receptor gamma (PPAR-γ), which are the major transcription factors that promote expression of most lipogenic enzymes. Glucose activates carbohydrate-responsive element-binding protein (ChREBP), another transcriptional factor that directly up-regulates lipogenic gene expression. Hyperglycemia also activates lipogenesis via indirect mechanisms, including induction of pyruvate kinase expression, and hence glycolysis, thereby providing precursors for lipogenesis. De novo lipogenesis not only promotes steatosis by increasing lipid synthesis, it also inhibits FA oxidation by promoting accumulation of malonyl-CoA, a factor that impedes fatty acid uptake into mitochondria for subsequent degradation. The third source of fat is the diet. The relevance of this source increases when high-fat diets are consumed. Hence, feeding rodents with a high-fat diet is a widely used dietary animal model to induce NAFLD, even when animals ingest isocaloric diets.

Hepatic steatosis results from an imbalance between the production/uptake of fat and its utilization. The fate of FAs in the liver is dictated by three processes: oxidative degradation (mainly in the mitochondria and to a lesser extent in peroxisomes); utilization to synthesize triglycerides that are the stored as inert lipid droplets; and export from the liver as very low-density lipoproteins (VLDLs).

The liver adapts to the overflow of FAs by increasing mitochondrial β-oxidation and mitochondrial biogenesis. That adaptation occurs not only in direct response to FA overflow, but also through the action of adipokines, such as leptin, fibroblast growth factor 21, and interleukin 6 (IL-6), as a consequence of hepatic IR, and because of increased activation of PPAR-α that promotes transcription of lipogenic enzymes. Notably, TNF-α represses PPAR-α expression, causing PPAR-α to decrease in severe liver injury, and this may limit mitochondrial flexibility as NAFLD progresses. Mitochondria are highly specialized organelles that generally extract energy efficiently from fuel. Thus increased FA oxidation in fatty livers might be expected to increase hepatic production of ATP. However, in patients with NAFLD, liver ATP stores are not increased, and hepatic ATP content is reduced in animal models of NAFLD/NASH. Furthermore, mitochondrial responses to transient energy deficits appear to be impaired in NASH, and this enhances hepatic susceptibility to ischemic injury. One mechanism that explains the less efficient ATP production from fuel oxidation is the up-regulation of mitochondrial uncoupling proteins (UCPs). UCP-2 is up-regulated in several animal models of NAFLD/NASH. It is also up-regulated in human NASH and correlates with severity of inflammation and fibrosis. UCPs are proteins localized in the inner membrane of mitochondria. They promote proton leak from the outer to the inner mitochondrial membrane. This dissipates the proton-motive force that normally drives electron transport along the mitochondrial electron transport chain and thus decreases the donation of electrons to molecular oxygen to form superoxide anion and to drive phosphorylation of ADP to generate ATP. UCP-2 is expressed at very low levels in healthy cells, and its major role appears to be constraining mitochondrial production of superoxide anion (a potent reactive oxidant species [ROS]) to protect cells from oxidative damage that might otherwise result from mitochondrial respiration. UCP-2 expression is up-regulated by FAs. FA accumulation also stimulates mitochondrial β-oxidation, thereby generating increased ROS and oxidative stress. Certain products of oxidative stress, such as 4-hydroxynononeal, increase UCP-2 activity. The resultant increases in UCP-2 activity enhance protection from FA-induced oxidative stress. However, high UCP2 activity also has a potential downside because it may induce sufficiently high levels of proton leak to uncouple substrate oxidation (i.e., electron transport) from ADP phosphorylation, thereby limiting ATP synthesis and dissipating energy as heat. This may explain observations that UCP-2 expression is increased in fatty hepatocytes of rodent models of obesity-related insulin-resistance and NASH. It may also explain why these hepatocytes are more vulnerable to a variety of stresses that challenge ATP production ( Fig. 25-2 ). The aggregate data suggest that steatosis does not occur as a consequence of decreased FA degradation, though there may be impairments in the degree of oxidative phosphorylation (i.e., energy production) that result from FA metabolism. In fact, in NAFLD, an increase in mitochondrial FA oxidation without concomitant induction of genes that encode mitochondrial respiratory chain components may further disturb the redox state, leading to mitochondrial exhaustion, as will be discussed later. FA-derived acetyl-CoA can alternatively be used as a substrate for ketogenesis. Ketogenesis provide extrahepatic tissues with an alternative fuel source when carbohydrates are scarce, such as during fasting, but also helps hepatocytes when dealing with energy surplus. In fact, on a high-fat diet, mice with inherited defects of ketogenesis not only develop more steatosis, but also more liver injury. Those mice show increased de novo lipogenesis and gluconeogenesis. Interestingly, hyperinsulinism inhibits ketogenesis, and hence hyperinsulinemic-NAFLD patients are in a state of relative ketogenesis insufficiency.

Mitochondrial adaptation is also limited because the entry of FAs into the mitochondrial matrix through the lipid level–sensitive carnitine shuttle (carnitine palmitoyl transferase I) is inhibited by malonyl-CoA that accumulates during FA excess. To overcome reduced mitochondrial degradation of FAs, FAs are redistributed to other organelles for degradation. In NAFLD, a compensatory increase in peroxisomal FA oxidation occurs. Patients with hepatic steatosis show proliferation and enlargement of hepatic peroxisomes. Peroxisomal lipid degradation has the advantage of allowing unrestricted lipid entry into the organelle. However, compared to mitochondria, peroxisomes are less efficient in extracting energy from the degradation of FAs. Another drawback of peroxisome FA oxidation is the production of hydrogen peroxide, which can perturb redox balance. Healthy peroxisomes are well equipped with antioxidant enzymes, and increasing peroxisomal FA oxidation through a PPAR-α agonist protects mice from NAFLD/NASH. Conversely, animal models with impaired peroxisome function develop spontaneous steatohepatitis and hepatocellular carcinoma.

The liver can also purge excessive fat content by exporting fat as VLDLs. VLDL particles contain a core of triglycerides and cholesterol esters, surrounded by phospholipids and the protein apolipoprotein B (apoB). After translation, apoB is sequentially lipidated in the ER lumen and Golgi apparatus by microsomal triglyceride transfer proteins (MTPs). Complete lipidation is necessary for vesicular flow and secretion of apoB into the plasma. Incomplete lipidation targets apoB for ubiquitination and proteasomal degradation. In insulin-resistant states, including NAFLD, there is evidence of increased VLDL secretion, probably through up-regulation of MTP and decreased degradation of apoB, despite reports of decreased mRNA expression of apoB. However, some genetic diseases associated with NAFLD/NASH result from deficient VLDL secretion. Abetalipoproteinemia is a rare genetic disease with a mutation in the MTP gene that leads to NASH and cirrhosis associated with fat malabsorption, acanthocytosis, and hypocholesterolemia in infancy.

Familial hypobetalipoproteinemia is a genetic disorder that induces the production of a truncated apoB and impairment of VLDL secretion. As a consequence, patients develop severe hepatic steatosis and NASH, independently of obesity and IR. Familial hypobetalipoproteinemia is estimated to occur in 1 in 500 to 1 in 1000 of the population, and clinicians should think of this diagnosis in patients with NASH and low cholesterol levels (i.e., lower than 150 mg/dL total cholesterol and 50 mg/dL HDL-cholesterol). Perturbations in the homeostasis of the phospholipid phosphatidylcholine also hamper VLDL assembly and secretion. Phosphatidylcholine can be synthesized through two pathways: incorporation of choline into phosphatidyl compounds or through three sequential methylations of phosphatidylethanolamine by phosphatidylethanolamine N -methyltransferase (PEMT). Choline deficiency is frequent in the Western world and can promote hepatic steatosis by inefficient outflow of lipids from the liver. Choline deficiency can result from low dietary choline intake, which occurs in more than 90% of the U.S. population. Also, obesity-associated dysbiota can promote increased choline degradation in the gut. Choline deficiency in association with loss of function PEMT polymorphisms synergistically increases the risk for NAFLD.

Notably, monounsaturated FA are preferentially incorporated into triglycerides. As such, saturated FA must be first desaturated through the action of steroyl-coenzyme desaturase 1 (SCD-1). An adaptive increase in SCD-1 and triglyceride synthesis has been described in human NAFLD. SCD-1 appears to exert different effects on different features of NAFLD in rodent models. When fed a NASH-inducing diet, mice deficient in SCD-1 develop less steatosis, but more severe liver injury as a result of accumulation of more toxic saturated FAs, as will be discussed later.

Nonalcoholic Fatty Liver Disease: An Adiposopathy

The idea that the adipose tissue is a simple inert store of fat is obsolete. We know now that the adipose tissue is a complex and plastic organ that has to adapt to very different conditions such as fuel insufficiency and fuel surplus. In that way, adipose tissue controls whole-body energy status. Also, adipose tissue has important endocrine functions, secreting a wide range of adipokines that regulate many physiologic processes such as sensitivity to insulin, appetite, immunity, and reproduction.

Adiposopathy is a new concept claiming that a positive caloric balance and sedentary lifestyle lead to adipose tissue dysfunction in susceptible individuals. The adipocyte responds to chronic energy surplus with maladaptive hypertrophy, in other words, a pathologic increase in cell size. Compared with healthy adipocytes, hypertrophic adipocytes have increased lipolytic capacity. This causes NEFAs to spill out of adipose depots and into the circulation, resulting in their ectopic accumulation in other organs such as the liver, heart, muscle, pancreas, and kidney. Ectopic accumulation of NEFAs increases metabolic stress. Accordingly, adipocyte size strongly correlates with risk for developing type 2 diabetes mellitus, cardiovascular disease, and NAFLD. The increase in the cell size also stresses the adipocyte, rendering it resistant to insulin and less able to store fat by decreasing the concentration of cholesterol in the plasma membrane and inducing IR. Cell enlargement also promotes cytoskeleton disassembly, which further impairs insulin signaling. In addition, large adipocytes up-regulate their expression of hormone-sensitive lipase, lipoprotein lipase, leptin, and β-adrenergic receptors. This change in gene expression has been attributed to increased cross-talk between enlarged adipocytes and the extracellular matrix. The process is thought to be mediated through integrin-β1 signaling that drives extracellular signal–related kinase (ERK _1-2 ) pathway activation to modulate adipocyte metabolism. For example, the up-regulation of β-adrenergic receptors sensitizes enlarged adipocytes to lipolysis induced by catecholamines during fasting.

Other pathologic processes also occur in overtaxed adipose depots. For example, increased FA accumulation in adipocytes promotes oxidative stress via mitochondrial production of reactive oxygen species, NADPH oxidase activation, and down-regulation of antioxidant enzymes. In turn, oxidative stress further decreases insulin sensitivity, deregulates production of adipokines, and induces inflammatory responses. Inflammation exacerbates this oxidative stress, establishing a vicious circle. Obesity also promotes ER stress in the adipose tissue. ER stress worsens IR through c-Jun N-terminal kinase (JNK) pathway-dependent serine phosphorylation of insulin receptor substrate 1 (IRS-1). Another important player in the pathophysiology of adiposopathy is hypoxia. Various mechanisms are thought to contribute to adipocyte hypoxia, including the failure of angiogenesis to keep pace with adipocyte mass expansion, compression of surrounding vessels by enlarged adipocytes, which decreases blood flow, and depletion of the vasodilator nitric oxide due to its consumption during oxidative stress. Adipose tissue expansion also induces extracellular matrix remodeling that results in matrix enrichment with collagen VI. The latter is thought to decrease flexibility of the extracellular matrix, limiting the ability of adipose depots to expand and accommodate enlarged adipocytes. This perpetuates the stress inflicted in the adipose tissue resulting in exacerbated metabolic deregulation. The mechanisms leading to fibrosis in the adipose tissue are not fully understood, but may reflect hypoxia-dependent induction of hypoxia inducible factor 1α and consequent increases in the transcription of several profibrogenic factors, such as connective tissue growth factor (CTGF) and tissue inhibitor of metalloproteinases 1. Also, the profibrogenic factor transforming growth factor β (TGF-β) is up-regulated by mechanical stress, which is triggered in the membranes of adipocytes by expanding lipid droplets. Consistent with this concept, adipose tissue expression of TGF-β increases in obesity.

All of these cellular stressors may culminate in adipocyte death. This is a nonintuitive concept, because we think of obesity as an expansion of the adipose tissue. However, cell death is an important and frequent phenomenon in obesity. Indeed, adipocyte death increases more than 30-fold in obese humans and in animal models of obesity. After feeding mice a high-fat diet for 16 weeks, 80% of adipocytes in visceral adipose tissue die. Hypoxia, ER stress, and mitochondrial dysfunction caused by oxidative stress or direct damage from cathepsin that is released from lysosomes permeabilized by FAs promote cell death. Adipocyte death decreases adipose depot uptake of NEFAs from circulation, and boosts release of NEFAs and other toxic products from adipose depots into the circulation. Also, dying adipocytes initiate a local inflammatory response that recruits macrophages to remove cellular debris, including the fat released from damaged cells. More than 90% of macrophages in obese adipose tissue surround dead adipocytes, producing crownlike structures. The accumulation of macrophages with a proinflammatory M1 phenotype potentiates an inflammatory state that worsens IR and metabolic deregulation. Indeed, macrophage accumulation and crownlike structures correlate with IR, systemic vascular endothelial dysfunction, hepatic steatosis, and NASH.

Lastly, the sick adipose tissue changes its endocrine properties, decreasing the secretion of adiponectin and increasing the secretion of IR-inducers, proinflammatory, and proatherogenic cytokines such as leptin, plasminogen-activator inhibitor 1, TNF-α, IL-6, monocyte chemoattractant protein 1, and angiotensinogen.

Adiponectin is a cytokine that is produced predominantly in the adipose tissue. Paradoxically, adiponectin expression decreases with obesity/expansion of adipose tissue. Lower adiponectin levels also associate with type 2 diabetes mellitus. Importantly, adiponectin inversely correlates with presence of NAFLD, NASH, and severity of fibrosis. Adiponectin protects from steatosis and steatohepatitis in animal models of NAFLD. In fact, adiponectin has insulin-sensitizing, antiinflammatory, and antifibrotic functions. It also protects from steatosis through its effects on whole-body metabolism, improvement of mitochondrial function, and decreased production of reactive oxygen species. Adiponectin is a PPAR-α agonist, ultimately increasing FA β-oxidation. It directly enhances insulin sensitivity by promoting tyrosine phosphorylation of factors that are activated downstream of insulin IRS-1 and Akt, while antagonizing inhibitory serine phosphorylation of these proteins. Adiponectin increases antiinflammatory IL-10 expression and suppresses proinflammatory TNF-α signaling by decreasing TNF-α expression and antagonizing its effects. It also modulates Kupffer cell expression of cytokines by decreasing NF-κB activation and antagonizes IL-1 through increased expression of IL-1 receptor antagonist. In addition, adiponectin inhibits fibrogenesis by decreasing production of TGF-α and through its direct effects on hepatic stellate cells (HSC) which block their transdifferentiation into myofibroblasts.

Leptin is another important adipokine in the pathogenesis of NAFLD. It is the master regulator of appetite, being a potent anorexigen via its action in the hypothalamus. Many genetic models of obesity/NAFLD take advantage of inherited impairment of leptin signaling. Mice genetically deficient in leptin are hyperphagic, develop morbid obesity, severe IR, and severe hepatic steatosis. Besides regulating food intake, leptin has many antisteatogenic actions, decreasing gluconeogenesis and de novo lipogenesis, while increasing hepatic FA oxidation. Thus it is not surprising that leptin-deficient mice develop severe steatosis. Interestingly, despite having mild NASH, leptin-deficient mice are protected from fibrosis, even when submitted to other profibrogenic liver insults. Leptin promotes hepatic fibrogenesis through its direct effects on HSCs. It prevents HSC apoptosis, promotes HSC proliferation, and stimulates HSCs to transdifferentiate into myofibroblasts. Leptin also modulates body fat composition, insulin activity (decreasing insulin production and secretion, while increasing sensitivity to insulin ), thermogenesis, and immune responses. Leptin is produced by adipocytes and hence leptin levels are increased in patients with obesity. However, hyperleptinemia induces partial leptin resistance that inhibits the anorexigenic effects of leptin in the central nervous system while maintaining many of its peripheral effects, including fibrogenesis. Leptin is generally increased in human NAFLD, and some studies (though not all) showed a positive correlation between leptin levels and the severity of hepatic steatosis, inflammation, and fibrosis.

Several other adipokines may also be important in the development and progression of NAFLD, such as resistin, visfatin and retinol-binding protein 4. However, their functions and role in the pathogenesis of NAFLD remain controversial.

In summary, energy surplus induces pressure for adipose tissue to expand. Because adipocytes have a limited capacity to enlarge, they become stressed and undergo programmed cell death, which initiates important inflammatory responses. Increased inflammation promotes insulin resistance in the adipose tissue, deregulation of adipokines production, and spill-out of FAs into the circulation. The adipose-derived FAs then accumulate ectopically in other organs, which, together with the increase in proinflammatory cytokines, induces insulin resistance in muscle and liver, eventually causing NAFLD, cardiovascular, and renal disease ( Fig. 25-3 ).

Nonalcoholic Fatty Liver Disease and Insulin Resistance: The Chicken and Egg Paradox

NAFLD strongly associates with IR and type 2 diabetes mellitus (T2DM). Indeed, NAFLD is two to three times more frequent in patients with T2DM and T2DM is five to nine times more frequent in patients with NAFLD. IR and T2DM increase the risk for developing NASH, progressive liver fibrosis, cirrhosis, and hepatocellular carcinoma. Conversely, patients with NAFLD and T2DM have worse T2DM-related outcomes, with worse glycemic control, and increased risk for renal and ophthalmologic complications. Moreover, patients with NAFLD, and particularly NASH, also have a higher risk of developing T2DM. The aggregate data suggest a causal link between T2DM and NAFLD and vice versa . Although this concept has become dogma in the NAFLD field, some evidence challenges the significance of IR/T2DM in the development and progression of NAFLD. First, neither IR, nor T2DM, is required for NAFLD to develop, because the full NAFLD spectrum (including cirrhosis) can occur in humans who are not IR or diabetic. Also, rodent animal models of NASH with an aggressive/fibrogenic phenotype (e.g., the methionine choline deficient dietary model and genetic models of NASH caused by deficiency of phosphatase and tensin homolog or fatty-acyl CoA oxidase ) do not develop obesity or IR. Conversely, long-term feeding with a high-fat diet induces severe IR, but causes only very mild steatohepatitis with almost no fibrosis. Furthermore, there is discordance between strategies that improve liver histology and strategies that improve glucose metabolism in patients with NAFLD. Indeed, major clinical trials of insulin-sensitizing agents failed to demonstrate significant improvement in liver fibrosis. This is an important negative result because fibrosis is the best predictor of prognosis in patients with NAFLD. On the other hand, an agent that improved liver fibrosis, obeticholic acid, worsened IR.

The most accepted model of NAFLD posits that IR starts in the overwhelmed adipose tissue, where it promotes release of FFAs through increased lipolysis that results from failure to suppress hormone-sensitive lipase. Ectopic accumulation of fat, the perturbed adipokine profile, and the proinflammatory state caused by adipose tissue dysfunction then promotes IR in muscle and liver. In the liver, IR and hyperinsulinism promote steatosis. In liver, insulin-initiated signaling normally inhibits gluconeogenesis and FA oxidation and promotes glycolysis and de novo lipogenesis. In liver, resistance to insulin selectively blocks the actions of insulin in glucose metabolism. Insulin effects on lipogenesis are preserved. Although several hypotheses have been forwarded to explain this discrepancy, the mechanism for the dissociation between sensitivity to insulin control of glucose and lipid metabolism is still not understood. IR promotes hyperinsulinemia to overcome the reduced sensitivity to insulin. Hyperinsulinemia increases de novo lipogenesis that exacerbates IR.

Insulin binds to its receptor inducing its dimerization and autophosphorylation in tyrosine residues 1158, 1163, and 1164 of β-subunits. The insulin receptor recruits and phosphorylates a number of substrates, including IRS-1/2, which then activate both the phosphatidylinositide-3-kinase (PI3K)/Akt and Ras/MAPKs signaling cascades ( Fig. 25-4 ). The former signaling cascade mediates insulin effects on metabolism and prosurvival, whereas the latter mediates effects on mitogenesis and cell growth. Regarding PI3K/Akt signaling, PI3K generates the second messenger phosphatidylinositol (3,4,5)-triphosphate that activates 3-phosphoinositide-dependent protein kinases 1 (PDK-1) and PDK-2, which, in turn, activate the protein kinase Akt (also known as protein kinase B ) through phosphorylation at residues tyrosine 308 and serine 473, respectively. This pathway leads to expression and activation of SREBP-1c, promoting lipid and cholesterol synthesis, and suppression of the gluconeogenesis-promoting forkhead transcription factor Foxo1, resulting in suppression of gluconeogenesis. Akt phosphorylates Foxo1, which promotes translocation of Foxo1 from the nucleus to the cytoplasm where it is ubiquitinated and degraded, thus hampering Foxo1-mediated gene transcription. Akt phosphorylates several other downstream targets, including glycogen synthase kinase 3b, thereby removing its inhibitory effect on glycogen synthase and promoting glycogen synthesis. Akt also inhibits tuberous sclerosis complex 1-2, which then promotes mammalian target of rapamycin complex (mTORC) activation. AKT also phosphorylates the following:

• Akt substrate 160-kDa protein resulting in Glut4 translocation and hence increasing glucose uptake

• Bcl2-associated agonist of cell death for apoptosis inhibition

• Phosphodiesterase-3 for cAMP degradation thereby inhibiting cAMP response element-binding protein (CREB)-regulated transcription coactivator 2, a CREB coactivator that increases hepatic gluconeogenesis

Of note, mTORC1 is a highly conserved protein kinase that controls cell growth and metabolism in response to energy status, promoting protein synthesis and SREBP-1c-dependent lipogenesis, while suppressing autophagy.

Liver steatosis can also induce IR. However, IR does not always accompany hepatic steatosis, suggesting that its occurrence may depend on the type of fats that accumulates in the liver. For example, diacylglycerol, ceramides and saturated FAs seem to be more potent inducers of IR than triglyceride. Lipids can interfere directly with the insulin receptor pathway, inhibiting insulin signaling. Saturated or unsaturated FAs produce diacylglycerol that activate protein kinase C epsilon, which in turn directly binds and inhibits insulin receptor kinase activity preventing IRS-1/2 tyrosine phosphorylation. Alternatively, saturated FA can induce IR through activation of toll-like receptor 4 (TLR-4) and subsequent up-regulation of ceramide synthesis. Ceramide induces activation of protein phosphatase 2A, which directly inhibits insulin pathway at the level of Akt phosphorylation.

Hepatic inflammation/NASH can further worsen IR. The concept that inflammation can induce IR was spawned more than 100 years ago, when salicylates were discovered to have antidiabetic effects. Recently, the mechanisms of salicylate protection were linked to inhibition of NF-κB pathway. Many inflammatory cytokines (e.g., TNF-α, IL-1, several TLR) activate NF-κB, an important transcription factor in NASH. In the canonical pathway of NF-κB, active IKK2 phosphorylates NF-κB inhibitor, IkBα, targeting it for ubiquitination and subsequent degradation. This releases NF-κB from IkB allowing NF-κB to enter the nucleus and promote transcription of proinflammatory genes. The NF-κB pathway elicits IR through several mechanisms. IKK2 induces serine phosphorylation of IRS1-2, thus preventing tyrosine phosphorylation and hence activation of IRS. Cytokines can also activate JNK to induce serine phosphorylation of IRS. IR itself can promote/perpetuate inflammation through abrogation of the inhibitory effect of insulin on Foxo1. Indeed, IR-associated reactivation of Foxo-1 increases expression of proinflammatory cytokines. Lastly, IR and hyperinsulinemia have profibrogenic effects. Direct effects of insulin on HSCs inhibit apoptosis while promoting HSC proliferation and transdifferentiation into activated myofibroblasts.

In summary, peripheral IR triggers the development of NAFLD. The inflammatory milieu caused by obesity then drives hepatic IR and hyperinsulinemia, further worsening hepatic steatosis. Lipotoxicity-induced liver injury worsens inflammation. IR and hyperinsulinemia intensifies liver injury, inflammation and fibrogenesis, establishing a vicious circle between ectopic fat and IR.