Review article: hepatic steatosis and insulin resistance


Dr A. Lonardo, Unità Operativa di Medicina, Interna e Gastroenterologia, Nuovo Ospedale Civile-Estense di Baggiovara, Modena 41100, Italy.


Hepatic steatosis may be both an adaptive phenomenon and an example of lipotoxicity. Its prevalence ranks in the same order of magnitude of insulin resistance in the general population. Studies support the finding that hepatic steatosis is secondary to insulin resistance and not vice versa. A steatotic liver will further contribute to the development of insulin resistance through impaired clearance of insulin from the portal blood, creating a vicious cycle. Insulin resistance is the leading force in the pathogenesis and natural history of non-alcoholic fatty liver disease. Dysfunction of energetic homeostasis and the interaction of adiponectin, leptin and tumour necrosis factor-α are key events in the pathogenesis of steatosis and insulin resistance. Insulin resistance represents the frame within which hepatic and extrahepatic non-alcoholic fatty liver disease-related clinical manifestations are to be anticipated and interpreted.

Background and aim

Does hepatic steatosis serve any biological purposes?

The accumulation of triglycerides and fatty acids (FA) metabolites in non-adipose tissues is known as lipotoxicity.1 Hepatic steatosis (HS) can be observed in a variety of clinical settings, one of which, non-alcoholic fatty liver disease (NAFLD), has universally been linked to insulin resistance (IR).2 The physiological significance of fatty changes in non-adipose tissues, including HS, remains unknown. However, examples from the animal world have led to the speculation that they might represent an adaptive response to anticipated food shortage.3, 4 Therefore, HS has recently been proposed to represent the liver counterpart of the ‘thrifty genotype’5 but it may also herald ominous biological and clinical consequences in humans. Aim of the present reappraisal is to update relevant issues of HS and IR that have come to the light following the conference held in Den Haag in 2000.6


Is HS a normal finding?

The conventional definition of HS is liver fat in more than 5% of hepatocytes.3 However, the biological rationale for such 5% cut-off has recently been challenged. Indeed a Finnish study reported that 50% of 30 normal men had, on average, only 1.7 ± 0.2% of liver fat content as evaluated through proton spectroscopy.7 Furthermore, 54% of 1039 hepatitis C virus (HCV)-positive patients enrolled in a large anti-viral multicentre trial had 0% HS at liver biopsy.8 These findings – and the observation that any degree of HS correlates with metabolic alterations8– imply that HS is best expressed as a quantitative continuous rather than a dichotomous present or absent variable9 and should probably be regarded as a disease state in itself.

Is IR a normal condition?

The IR is also measured along a continuous scale and no distinct threshold for its development exists.2 The concept of IR is intuitive to the clinician dealing with the diabetic patient necessitating large amounts of ‘exogenous insulin’ to achieve blood glucose control.10 However, IR is also observed in subjects without overt glucose alterations in whom it antedates type 2 diabetes (T2D) by decades. In such subjects, IR manifests with a reduced efficiency of ‘endogenous insulin’ to inhibit hepatic glucose production and to stimulate glucose utilization in skeletal muscle and adipose tissue. This results in compensatory hyperinsulinaemia, which accounts for the elevation of Homeostasis Model Assessment (HOMA)-IR [fasting insulin (μIU/mL) × fasting glucose (mm/22.5)]. Except for transient physiological states (e.g. pregnancy, puberty), IR is an irreversible pathological condition associated with compensatory hyperinsulinaemia.11

Deceptively silent in its early stages, IR may eventually manifest itself clinically as altered (glucose and/or lipid) metabolism with or without cardiovascular disease and/or cancer. Insulin concentrations in the portal blood are twice those in the peripheral blood and therefore it is logical to predict HS to be an early manifestation of IR/hyperinsulinaemia.

The dimensions of the problem

What is the prevalence of HS?

The prevalence of HS in the general population is 20% and that of non-alcoholic steatohepatitis (NASH) approximates 3.5%.12 The prevalence of HS in patients with HCV chronic infection is two- to threefold increased compared with that expected from a chance concurrence of NAFLD in HCV-positive subjects and to the prevalence of HS in chronic hepatitis because of other aetiologies.13 Given that T2D is also increased in HCV-positive subjects, HS has been proposed as the link between HCV infection and IR.14

What is the prevalence of IR?

Irrespective of the definition given, IR syndrome increases with age and is almost always higher in men than in women for a given age.15 Prevalence rates in the 40–55-year age group range from 7 to 36% (WHO syndrome) and from 1 to 22% (EGIR syndrome) in men; and from 5 to 22% (WHO syndrome) and from 1 to 14% (EGIR syndrome) in women.15

It is concluded that the prevalence of HS and that of IR are in the same order of magnitude, which is a further evidence for their non-chance association.3

HS and IR: chicken or egg?

Does IR antedate HS in animal models?

Peripheral IR.  It is best measured through the euglycaemic clamp technique that measures muscle IR – mainly manifests with increased free FA liberation from adipose tissue secondary to the unopposed antilipolytic action of insulin on hormone-sensitive lipase.

Hepatic IR.  Hepatic IR manifested by high fasting insulin levels relative to the concurrent glucose plasma concentrations – mainly manifests with increased hepatic glucose output in the fasting state. Peripheral and hepatic IR (and the results of tests exploring them) are related to each other in clinical conditions such as T2D and obesity, which are common correlates of human HS. However, IR is also found in a variable proportion of non-obese subjects,10, 16 and in rats fed short-term high-fat diets where the development of HS is associated with IR in the absence of obesity.17 Data support the hypothesis that HS leads to hepatic IR by stimulating gluconeogenesis and activating protein kinase (PKC)-epsilon and Jun N-terminal kinase (JNK) 1, which may interfere with tyrosine phosphorylation of insulin receptor substrate-1 (IRS-1) and IRS-2 and impair the ability of insulin to activate glycogen synthase and to translocate the glucose transporter GLUT-4 to the cell surface.2, 17 Another animal model used transgenic mice expressing the HCV core gene.18 These mice exhibited a marked IR, which progressed to overt diabetes after a high-fat diet. A high level of tumour necrosis factor-alpha (TNF-α), altering tyrosine-induced phosphorylation of IRS-1 is considered one of the bases of IR in this model.18

These animal studies indicate that fat diet-induced HS perpetuates IR through impaired postreceptor insulin signalling prior to and independent of the development of obesity. Furthermore, HS appears to be essential for the development of overt diabetes in HCV infection.

IR: a matter of belly or head?

Visceral adipose tissue has adverse physiological and metabolic consequences.19 However, according to Björntorp's hypothesis, the IR syndrome is linked to primary dysregulation of the hypothalamic–pituitary axis. There are striking analogies between growth hormone (GH) deficiency in adults and the metabolic syndrome and between the latter and NAFLD. On this basis, we undertook a study of GH plasma levels in NAFLD. Our data support the hypothesis NAFLD constitutes a GH-deficient state in men.20 Other researchers confirm that progressive NAFLD is prevalent in adults with hypothalamic and pituitary dysfunction and that HS disappears after hormone replacement.21–23

Do interventions that improve IR in humans also affect HS?

The HS consistently improves after pharmacological and non-pharmacological interventions aimed at restoring insulin sensitivity.5, 24 This definitely demonstrates that IR is the cause and not the effect of HS, although the steatotic liver may worsen IR via a reduced clearance of insulin from the portal blood.3

The lesson from islet cell transplantation

Islet transplantation via portal venous delivery has emerged as an effective treatment for metabolically labile type 1 diabetics.25 Most likely secondary to intrahepatic insulin secretion, HS follows intraportal islet allotransplantation, and may therefore be considered a marker of functional islet graft survival.26 HS occurs in 20% of these transplanted patients and completely resolves after graft failure.27 These findings confirm pioneer observations that subcapsular HS develops after intraperitoneal insulin delivery,28 a phenomenon that occurs, once again, in approximately 20% of cases.29

In conclusion, a wealth of evidence strongly supports that IR is the primary (probably hypothalamic) physiopathological abnormality eventually conducive to HS.

Lipotoxicity and lipodistrophy

HS as a manifestation of lipotoxicity

Adipose tissue is the only physiological lipid depot. Intracellular accumulation of triglycerides and FA metabolites [fatty acyl coenzyme A (CoAs), diacylglycerol and ceramide] in non-adipose tissues leads to acquired insulin signalling defects and acquired IR.30 The fact that the lipogenic transcription factor – SREBP-1c which provides the enzymatic machinery for de novo synthesis of fat – is overexpressed in obese rats with defective leptin receptors is further evidence that IR is secondary to the overaccumulation of lipids.1 It has recently been suggested that free FA result in hepatocyte Bax translocation to lysosomes, lysosomal destabilization with cathepsin B release into the cytosol and nuclear factor κ-B (NF-kB)-dependent TNF-α expression.31 Such pathogenic cascade might be the cellular basis accounting for increased hepatocyte apoptosis in NASH, in which it correlates with disease severity.32

The paradox of lipodystrophy

Hepatic steatosis is not confined to subjects with generalized or visceral obesity, but also occurs frequently in lipodystrophies, which are characterized by severe loss of adipose tissue from different regions of the body. The severity of HS seems to be related to the extent of fat loss.9 Those patients with congenital or acquired generalized lipodystrophy often are insulin-resistant diabetic and their HS may progress to cirrhosis, with or without HCV coinfection.9, 33 The mechanism for HS in lipodystrophies may be related to the reduced storage capacity of adipose tissue depots with ensuing diversion of fat for accumulation in other tissues such as the liver. Alternatively, reduced central/peripheral action of leptin might have a role9 as confirmed by efficacy of leptin replacement treatment.34

It is concluded that HS is observed in such diverse clinical phenotypes as the obese and the lipodystrophic patient that share altered fat partitioning and IR, which further reinforces the link between IR and HS.

Does IR cause HS or NASH?

The HS and NASH are associated with IR and the evolution from the former to the latter is deemed to be mediated by additional pathophysiological abnormalities including oxidative stress within the hepatocytes.2 However, two observations support the idea that ‘one hit – namely IR – is enough’ in the pathogenesis of NAFLD.35 The first is that IR is associated with those biochemical pathways leading to generation of reactive oxygen species (ROS) and therefore to the development of oxidative stress. The second is that increasing levels of IR is associated with more advanced NAFLD stages. Oxidative stress can occur when either excess ROS are generated within the hepatocytes or antioxidant defences are defective.36 The main sources of hepatic ROS are the mitochondrial, peroxysomal and cytochrome (CYP) P2E1. The role of mitochondria in the pathogenesis of NAFLD will be reviewed in the next paragraph. In NAFLD, the increase of free FA in the liver may overwhelm mitochondrial β-oxidation capacity, increasing peroxisomal β-oxidation and H2O2 production. CYP2E1 plays a useful physiological role when starvation and/or low carbohydrate diets prevail because of its contribution to the metabolism of FA and its capacity to convert ketones to glucose. However, when the adaptation becomes excessive, adverse consequences prevail: CYP2E1 generates excessive oxygen radicals that may exceed cellular defence systems, and cause tissue damage.37

In conclusion, IR can contribute to all the pathways leading from HS to NASH.38 According to a principle of intellectual parsimony, no extra injuries appear necessary to account for the whole gamut of NAFLD.

HS and IR: a shortage of fuel?

Does HS impair the production of ATP?

Early works demonstrated that ATP homeostasis is altered in human NASH.4,39 Metabolic disturbances in NAFLD are also associated with mitochondrial abnormalities such as increased size and crystalline inclusions. Oxidative damage is the most likely causative process and may result in alterations of mitochondrial DNA, apoptosis and necrosis. Mitochondrial dysfunction may also play a role in those numerous clinical conditions associated with NAFLD, such as hepatocellular carcinoma, lipodystrophy, IR, gut dysmotility, cryptogenic cirrhosis.

Does HS associate with impaired vascular supply?

Microvascular dysfunction and complications resulting from disruption of nitric oxide (NO) signalling and direct metabolic injury to endothelial cells are a well accepted feature of IR.40, 41 It is of interest, therefore, that impaired vascular function has also been consistently observed in HS42, 43 and that it has even been proposed to take part in the pathogenesis of fatty liver syndromes.44 Current views hold that fatty accumulation in the cytoplasm of hepatocytes is associated with an increase in cell volume that results in partial or complete obstruction of the hepatic sinusoid space.44

Taken collectively, these data point to fatty liver as a ‘weaker’ organ and potentially account for the transplantologists’ observation that fatty livers are more susceptible to ischaemia-reperfusion injury.

The good, the ugly and the bad

Strategically modelled by evolution to cope with stressful situations such as famine and infection, adipose tissue – presently deemed to be an active endocrine organ – is a key regulator of both metabolism and inflammation.45 In as much as a steatotic hepatocyte morphologically (and therefore perhaps also functionally) resembles an adipocyte4 understanding the role of the three molecules – adiponectin, leptin and TNF-α– strategically located at the crossroad between inflammation IR and HS is crucial. ‘Adiponectin’46 is an antilipogenic insulin-sensitizing protein produced by adipocytes. Adiponectin has structural resemblance to TNF-α but opposite effects. It enhances FA oxidation and alters hepatic gluconeogenesis; its low circulating levels precede and predict T2D and have also been linked to several other components of the IR syndrome. Circulating adiponectin concentrations are increased by exercise training when body fat is reduced. Adiponectin, which is induced by PPAR-γ agonism, has anti-inflammatory activity through the production of interleukin (IL)-10 and IL-1RA and impaired production of TNF-α and interferon (IFN)-γ.47–49 Serum adiponectin is inversely correlated with liver fat content50 and transaminase activity in males, suggesting that hypoadiponectinaemia may worsen NAFLD.51 Adiponectin administration may be of benefit in controlling the multiple metabolic disturbances present in obese patients, including HS, dyslipidaemia and IR.46‘Leptin’ (=lean)46 is the name given to the protein product of the defective gene discovered in the ob/ob mouse. It is synthesized and secreted by adipocytes. Decreased leptin concentrations act as a signal of negative energy balance and low energy reserves. Mutations in the leptin gene or defects in the leptin receptor result in extreme hyperphagia, obesity, neuroendocrine, reproductive, metabolic and immunological alterations. In NASH patients, leptin levels are elevated52 and are directly correlated with the severity of HS but not with inflammation or fibrosis.53 In animal models leptin is a critical fibrogenic factor whose action may be mediated by increased noradrenaline and transforming growth factor (TGF)-β.54 It is possible that leptin replacement therapy may benefit HS, hyperlipidaemia and IR in patients with low leptin levels.46‘TNF-α54 is produced by endotoxin-activated macrophages and visceral abdominal adipose tissue. TNF-α reduces insulin sensitivity, has proinflammatory and possibly hepatocarcinogenic activity (in rodents). TNF-α-induced cellular IR is mediated by stress-related PKC, including IKK-β and JNK. The same mechanisms that cause IR also promote TNF-α synthesis, perpetuating a self-reinforcing, positive feedback mechanism for sustained TNF-α activity and chronic IR. Increased TNF-α promotes hepatocyte death via apopototic or necrotic route secondary to alterations of cellular ATP levels. TNF-α is a potent inducer of mitochondrial ROS release. Unbuffered exposure to excessive superoxide anion can induce hepatocyte apoptosis, while sudden drops in hepatocyte ATP stores trigger necrotic cell death. TNF-α also initiates fibrosis both by direct activation of hepatic stellate cells and via production of TGF-β.38 TNF-α polymorphisms may also represent a susceptibility genotype for IR, NAFLD and steatohepatitis.55 In addition, TNF-α pathway dysregulation may be involved in increased plasma plasminogen activator inhibitor 1 (PAI-1) observed in obese patients with HS.56

The TNF-α, adiponectin and leptin are closely linked to each other. Increased TNF-α leads (via NF-kB) to inhibited activity of PPAR-γ. As PPAR-γ is required for adipocyte to differentiate sufficiently to make adiponectin, increased TNF-α will finally result in a reduced production of adiponectin, which permits sustained activity of TNF. Furthermore, TNF-α and TNF-α-induced cytokines can also stimulate the production of leptin, which, in turn, inhibits insulin secretion by islet cells. Hence, both TNF-α and leptin perpetuate IR, albeit by different mechanisms.57 A single study evaluating leptin, adiponectin and TNF-α in NAFLD concluded that adiponectaemia is associated with more extensive necroinflammation and may contribute to the development of NASH independent of IR.58

Conclusions and perspectives

The HS may be an adaptive phenomenon and an example of lipotoxicity. Its prevalence in the general population is in the same order of magnitude of IR (HS is secondary to IR). The former will further contribute to the development of the latter through impaired clearance of insulin from portal blood, leading to a positive feedback mechanism. Dysfunction of energetic homeostasis in NAFLD provides a source for potential therapeutic intervention. The interaction of adiponectin, leptin and TNF-α is a key event in the pathogenesis of HS and IR. The latter is the frame within which the spectrum of hepatic and extrahepatic NAFLD-related phenomena and the natural history of this condition are to be anticipated and interpreted.


This study was funded by grants from MIUR (Ministero Istruzione Università e Ricerca) Anno 2002 - prot. 2002062883_001 and Anno 2004 - prot. 2004061213_001. We thank Azienda Ospedaliera Policlinico, AUSL of Modena and family doctors for collaboration.