|Home | About | Journals | Submit | Contact Us | Français|
Non‐alcoholic fatty liver disease (NAFLD) is the most common hepatic disorder in western countries, and its incidence is increasing. This review outlines the significant health burden posed by NAFLD and discusses what is presently known about its pathogenesis, including the roles of the metabolic syndrome, obesity, insulin resistance, hepatic steatosis, reactive oxygen species, inflammatory cytokines and adipocytokines. The way in which NAFLD is clinically diagnosed is described, and areas of uncertainty surrounding its investigation are identified, before discussing the relative merits of the limited treatment options available and looking ahead to potential therapeutic strategies for the future.
Non‐alcoholic fatty liver disease (NAFLD) covers a progressive spectrum of liver pathology, from hepatic steatosis to inflammation and fibrosis characteristic of non‐alcoholic steatohepatitis (NASH) and cirrhosis (figs 11 and 22).1 It is associated with the development of hepatocellular carcinoma (HCC),2,3 a significant proportion of which develops from “cryptogenic cirrhosis”,4 believed to represent the end‐stage of previously undetected NASH.5,6 Epidemiological data suggest that NAFLD is the most common hepatic disorder in western countries,7 and its incidence is increasing.8 A recent study in the United States found that 34% of the population had hepatic steatosis9; reports from the Asia‐Pacific region suggest similarly high levels paralleling the rise in obesity and type 2 diabetes.10
NAFLD is widely regarded as a product of insulin resistance due to its strong association with insulin resistant features also seen in the metabolic syndrome.11,12,13 Similarities in prevalence, clinical features and outcome have led some to argue that NAFLD is an unrecognised component of the latter,14,15 while others have suggested it should be renamed as “metabolic liver disease”.16 In fact NAFLD may precede the development of the metabolic syndrome, implying its presence is an early indicator of a premorbid cardiovascular state.16,17 As for whether the development of advanced liver disease is a valid concern among patients with co‐existing cardiovascular risk factors, El‐Serag et al18 showed an increased incidence of non‐alcoholic chronic liver disease and HCC in patients with diabetes, and de Marco et al19 found that the increase in liver related mortality exceeded the increase in cardiovascular related mortality in a cohort of type 2 diabetics. NAFLD therefore presents a significant, growing health burden in the developed world.
The classic “two‐hit” theory of NAFLD originally proposed by Day and James20 describes peripheral insulin resistance as the initiating event, with an increase in adipocyte lipolysis and hyperinsulinaemia fuelling hepatic triacylglyceride (TAG) synthesis and steatosis.21 Reactive oxygen species (ROS) then drive peroxidation of the accumulated hepatic lipids,22 producing toxic products that damage mitochondria and stimulate further production of ROS.23 Such a positive feedback loop results in cellular damage, activation of Kupffer cells and the generation of proinflammatory cytokines which drive hepatic inflammation,24 with concurrent activation of hepatic stellate cells establishing a pattern of repair that leads to hepatic fibrosis.23,24
However, the link between the development of hepatic steatosis and the generation of ROS is still poorly understood. Hepatic steatosis has been shown to promote hepatic insulin resistance,25,26 and may induce the up‐regulation of carnitine palmityl transferase I as well as a reduced affinity for its physiological inhibitor, malonyl CoA, facilitating the mitochondrial uptake of free fatty acids (FFAs).23,27 With hepatic FFA pools overflowing, such an increase in β‐oxidation could conceivably generate sufficient numbers of ROS to trigger lipid peroxidation.23,27 But can steatosis really drive fibrosis alone?
The natural history of NAFLD would suggest not, as statistically only a small proportion of patients develop advanced liver disease.28,29 Much of what is known about the progression from hepatic steatosis to fibrosis comes from studies of patients with chronic hepatitis C (HCV), 20–30% of whom have been estimated to develop cirrhosis within 10–30 years.30 A number of such studies have demonstrated a significant association between hepatic steatosis and fibrosis.31,32,33 Applying the two‐hit theory, ROS may be produced as a direct effect of the virus on hepatic mitochondria and/or the immune response to viral infection,34 with steatosis providing an abundance of fuel on which such radicals can act.35 In support, steatosis has been associated with markers of oxidative stress in patients with hepatitis C.36 While this provides good evidence for the potential of such a pathogenetic mechanism, it does not explain how ROS may be generated in the absence of a secondary hepatic insult.
Furthermore, a causal relationship between hepatic steatosis and fibrosis has not been established. It is possible that the “progression” from the former to the latter reflects instead the effects of an underlying pathology that produces steatosis and fibrosis concurrently. Insulin resistance is an obvious candidate, and a number of studies have shown it to be a reliable predictor of disease severity.37,38 It underlies the “first hit” of hepatic steatosis in NAFLD, and there is good evidence to suggest that it is responsible for the hepatic steatosis seen in non‐genotype 3 chronic HCV.34,39 Several studies have demonstrated an association between insulin resistance and hepatic fibrosis in cohorts of patients with NAFLD37,40,41 or HCV,42,43,44 and it may promote fibrosis through the following:
Recently it has become clear that insulin resistance can increase serum uric acid levels, and hyperuricaemia may worsen NAFLD.51 However, more studies are required to evaluate the links between insulin resistance and hyperuricaemia, and determine its significance in the context of NAFLD.
Steatosis and insulin resistance have therefore emerged as potentially significant factors in the evolution of the relatively benign fatty liver to the morbid hepatic fibrosis characteristic of advanced NAFLD. However, their complex inter‐relationship and pathogenic significance remain frustratingly unclear,52 and in contrast to the evidence presented above some studies have failed to find an association between steatosis53,54,55 or insulin resistance56 and the progression of liver disease.
Also unclear is the role of the proinflammatory cytokine tumour necrosis factor α (TNF‐α). One study has demonstrated an increase in the hepatic production of TNF‐α in response to diet‐induced obesity and hepatic steatosis,57 and it is known to promote insulin resistance, both directly via serine phosphorylation of IRS‐1 residues downstream of the insulin receptor,39,52,58 and indirectly via the perturbation of the growth hormone axis52 and disruption of β‐cell function.58 Transgenic experiments with the HCV core protein in mice have demonstrated that hepatic insulin resistance following HCV infection is driven by the up‐regulation of TNF‐α and can be reversed by TNF‐α blockade.59 It has been pointed out, however, that other inflammatory liver diseases such as hepatitis B are not associated with insulin resistance despite raised intrahepatic and circulating levels of the cytokine.60 Aside from its potential effects on glucose metabolism, TNF‐α has also been shown to induce mitochondrial abnormalities61 which may promote the production of ROS, and is known to play a role in the recruitment of Kupffer cells, activation of hepatic stellate cells and fibrogenesis in mouse models of NASH.62
In fact the most likely source of TNF‐α in NAFLD is adipocytes.52 Plasma levels have been shown to correlate directly with body fat mass,63 and it has been suggested that the metabolic stress placed on adipose tissue in obesity stimulates macrophage infiltration and localised TNF‐α production, with the cytokine acting in a paracrine fashion to suppress secretion of the adipokine mediator adiponectin.24,58 Adiponectin is known to enhance insulin sensitivity,64 and is predominantly secreted by visceral fat.64 Consequently the expansion of the visceral adipocyte compartment may paradoxically suppress its secretion,65 and NAFLD has been associated with a reduction in plasma adiponectin,65,66,67 explaining the well described observation that abdominal waist to hip ratio predicts insulin resistance and NAFLD better than body mass index, which says nothing about fat distribution.17,65,68 A number of studies have found an association between low levels of adiponectin and hepatic steatosis in patients with NAFLD65,69,70 or HCV.71 Adiponectin is also believed to provide a hepatoprotective role against inflammation and fibrosis,65,66 and the extent of hypoadiponectinaemia has been shown to correlate well with the severity of necroinflammation and hepatic fibrosis in patients with NAFLD.69,72,73
In contrast, levels of the adipocytokine leptin appear to be significantly raised in patients with NASH74,75 or HCV,76,77 an effect which some have ascribed to the obesity‐related phenomenon of “leptin resistance”.78 In both disorders raised levels have been found to correlate with steatosis74,79 and leptin is known to modulate insulin sensitivity,78 as well as being implicated in the development of hepatic fibrosis. Both leptin and leptin receptor deficient rodents either fed choline deficient diets80,81 or treated with profibrotic xenobiotics,82 failed to develop the hepatic fibrosis seen in their normal littermate controls. Leptin has been shown to increase the expression of the profibrotic cytokine TGF‐β in sinusoidal endothelial and Kupffer cells, increase PDGF‐dependent proliferation of hepatic stellate cells,82 and up‐regulate their expression of proangiogenic and proinflammatory cytokines.83 Furthermore, activated hepatic stellate cells have been identified as a non‐adipocyte source of leptin themselves.84 However, some researchers have found no increase in serum leptin concentration between patients with NASH or HCV and controls,85,86 and no correlation between leptin level and hepatic histology in patients with NAFLD85,87 or HCV.86,88,89 One study even describes the reduction of serum leptin with progressive hepatic injury.90
Clearly insulin resistance, hepatic steatosis, reactive oxygen species, inflammatory cytokines and adipocytokines all have a part to play in the development and/or progression of NAFLD. However their precise roles and the sequence of events remain unclear (fig 33).). The disorder shares many similarities with the development of hepatic steatosis and fibrosis in HCV, and the same key players appear to be involved. Nonetheless, the complex inter‐relationship between host risk factors and the cytopathic effect of the virus, especially genotype 3 HCV,91 means we should be wary of drawing comparisons between apparently similar hepatic pathologies. Critically, the trigger that initiates the conversion of benign hepatic steatosis to steatohepatitis in NAFLD remains unknown. Further large scale, prospective, well controlled studies adjusting for a large number of confounding metabolic factors are required to elicit further information about this prognostically important question.
No specific laboratory test exists for the diagnosis of NAFLD; it is primarily a diagnosis of exclusion. Suspicion is typically aroused by the incidental finding of persistently raised serum transaminases, particularly alanine transaminase (ALT), which is known to be a more specific indicator of hepatocyte injury than aspartate transaminase (AST).92 However, the guidelines on what constitutes a clinically significant elevation, and the period of time that must elapse before such an elevation may be considered persistent, are far from clear.
A recent comprehensive review of the significance of elevated serum transaminases in primary care suggests that the first line of investigation should be a detailed history and physical examination to uncover risk factors for, and/or clinical features of, liver disease.93 If risk factors are present, the subsequent course of action is dependent on their nature: some predispose to hepatic pathology directly (eg excessive alcohol consumption), while others predispose to hepatic pathology indirectly (eg intravenous drug use). The authors recommend that patients with a direct risk factor should undergo a six month trial of appropriate interventional therapy, with further investigation reserved for those with an indirect risk factor, those with clinical features, or those in whom such interventional therapy fails to work. These guidelines imply that patients who are obese and/or diabetic without other direct risk factors, indirect risk factors or clinical features of liver disease may be assumed to have NAFLD and treated accordingly for six months without further investigation, the pre‐emptive diagnosis being confirmed by the successful reduction of serum transaminases.
Patients with NAFLD may present with raised serum transaminases that are not associated with any risk factors or clinical features of liver disease. For such patients Smellie et al93 recommend further investigation on the basis of a single measurement of raised ALT >3× upper limit of normal (ULN), while two measurements taken six months apart are required before further investigation is warranted if the raised ALT is <3× ULN, to distinguish between a transient elevation that is clinically insignificant and a persistent rise which may indicate pathological change. However, the range of transient, clinically insignificant elevations in ALT is poorly defined, and there is little evidence to suggest that a threshold of 3× ULN optimises the investigative approach. What is clear is that once innocuous elevations in ALT have been ruled out, the size of the rise gives little or no indication of the extent of liver disease. Indeed, Smellie et al point out the “poor correlation between level of rise and biopsy findings” and hence “do not recommend a lower threshold” for further investigation. At one end of the spectrum, normal serum ALT has been documented in significant numbers of patients with NAFLD,94,95,96 and at the other, raised levels have been found in patients with no evidence of hepatic pathology.97 This suggests that NAFLD is under‐diagnosed in society, and supports the need for an active diagnostic approach that recognises the independent importance of risk factors for the disease. The significance of absolute transaminase levels is unclear, and more research is required to determine the relevance of commonly quoted “threshold” limits.
When the decision to undertake further investigation into raised serum transaminases is made, Smellie et al suggest initially performing a full blood count to screen for excessive alcohol consumption and hypersplenism; an autoantibody count to screen for primary biliary cirrhosis and autoimmune hepatitis; ferritin levels to screen for haemochromatosis; and hepatitis serology to screen for hepatitis B or hepatitis C as the cause. If a diagnosis is not confirmed, a second set of investigations should be initiated, with anti‐endomysial antibody detection to screen for coeliac disease; α1‐antitrypsin measurement to screen for α1‐antitrypsin deficiency; caeruloplasmin and urinary copper levels to screen for Wilson disease; and a liver ultrasound (US) to screen for biliary duct obstruction and a liver tumour. If the above treatable disorders have been ruled out, evidence from a recent large scale study by Skelly et al98 suggests that the most likely cause is NAFLD. They performed liver biopsies on 354 patients with unexplained raised serum liver function enzymes (including serum ALT) >2× ULN for >6 months, and found that 32% of the patients had hepatic steatosis and 34% of the patients had NASH. Given the previously stated poor correlation between the degree of liver enzyme abnormality and extent of liver disease, there is no reason to suggest that the results would have been any different if the cohort had included patients with an unexplained persistent elevation in serum ALT of any size.
From a diagnostic point of view, a liver US performed to exclude biliary duct obstruction or a liver tumour as the cause of an elevation in serum ALT may also be used to detect hepatic steatosis. Several trials suggest that US has a high sensitivity and specificity for the detection of fatty livers. Saverymuttu et al99 selected a cohort of 85 patients that had undergone liver biopsy; they showed that US blindly identified hepatic steatosis in 94% of the 48 patients with the condition, and 100% of the patients for whom steatosis was histologically graded moderate or severe, with a specificity of 84%. Mathiesen et al100 investigated 165 asymptomatic patients with persistently raised serum transaminases and found that US detected biopsy‐proven hepatic steatosis 90% of the time, with a specificity of 82%. However, evidence suggests that US is not a reliable investigation for the diagnosis or grading of fibrosis99,100; a liver biopsy is widely recognised as the only definitive test for NASH and cirrhosis.101
Figure 44 details a suggested algorithm for the investigation of persistently raised serum transaminases.
Given the strong association between obesity, insulin resistance and NAFLD, weight loss would appear to be an appropriate management strategy. The most rational approach involves lifestyle modifications incorporating diet and exercise; a number of studies indicate that such a weight loss regime can successfully reduce raised serum transaminases11,102,103,104 and hepatic steatosis103,105,106 in patients with NAFLD. Kugelmas et al104 studied the effects of weight loss in patients with biopsy‐proven NASH, achieved through a step 1 American Heart Association diet107 with at least 30 minutes light exercise per day. They found a significant reduction in raised serum transaminases at 6 weeks that was maintained at 12 weeks, and also report a significant reduction in plasma levels of IL‐6, an established marker of the severity of the hepatic inflammatory response,108 suggesting that such a regime may also reduce hepatic inflammation. Two case reports109,110 and a recent pilot study by Huang et al111 support this claim, showing that weight loss achieved through a combination of diet and exercise can induce the regression of the histological features of NASH. Furthermore, in the Huang et al study there was a significantly greater reduction in mean waist circumference in the patients showing histological improvement, as well as a greater reduction in visceral fat on CT, showing the importance of reducing visceral fat stores in the treatment of NAFLD.
However, therapies based on lifestyle modification alone are notoriously hard to sustain. A number of studies have shown that weight loss following bariatric surgery can have a beneficial effect on patients with NAFLD. Gastroplasty,112,113 gastric banding,114,115 Roux‐en‐Y gastric bypass116,117 and gastric bypass via biliopancreatic diversion118 have all been shown to induce weight loss with a concurrent improvement in liver biochemistry and histology, including reductions in serum transaminase levels, hepatic steatosis, inflammation and fibrosis. Recently, pilot studies have suggested that pharmacologically induced weight loss may also improve NAFLD. Orlistat, an inhibitor of gastric and pancreatic lipases typically used to manage obesity, and sibutramine, a centrally acting serotonin and noradrenaline reuptake inhibitor which has direct effects on appetite, have both been shown to be effective at inducing weight loss, reducing raised serum transaminases and regressing hepatic steatosis in patients with NAFLD.119,120
Weight loss is an attractive therapy for NAFLD for a number of good reasons. Evidence suggests that it is effective; there are a number of ways in which it may be achieved; and it simultaneously tackles cardiovascular risk factors of the associated metabolic syndrome.121 However, a recent review pointed out the lack of randomised, controlled trials investigating the effects of weight loss on hepatic pathology.122 Some studies have questioned its efficacy; Fan et al123,124 found that weight loss following a low calorie diet reduced hepatic steatosis but had no effect on hepatic inflammatory changes in rodent models of NASH. Others have shown that weight loss can produce a significant deterioration in hepatic function. Severe dieting results in a dramatic increase in peripheral lipolysis which furthers the supply of FFAs to the liver, and has been linked with reduced levels of the antioxidant glutathione, providing the conditions for a paradoxical increase in hepatic inflammation and fibrosis.125 Rapid, diet‐induced weight loss has been shown to precipitate the development and degeneration of steatohepatitis.126,127 Similar concerns exist regarding weight loss achieved through surgery. Jejunoileal bypass for morbid obesity has long been recognised as a primary cause of NASH,128,129 and weight loss following gastroplasty has also been associated with hepatic degeneration.130
Given the difficulty of dietary intervention, the relative risks of bariatric surgery, the scarcity of data surrounding the effects of pharmacologically induced weight loss, and the controversy over the effectiveness of weight loss as a treatment for NAFLD, it is important that alternative therapeutic strategies are developed.
The central role of insulin resistance in the pathogenesis of NAFLD has stimulated considerable interest in the use of insulin sensitisers. Metformin has been shown to eliminate fatty liver disease in insulin resistant ob/ob mice,131 and studies in patients with NAFLD have shown similarly beneficial hepatic effects, including reductions in persistently raised serum transaminases,132,133,134 hepatic fat content,133,134 necroinflammation, fibrosis, and a significant improvement in coexisting positive criteria for the metabolic syndrome.134 Preliminary reports also suggest that thiazolidenediones may be effective at reducing hepatic steatosis, inflammation and fibrosis.135 Alternative approaches to overcoming insulin resistance are to stimulate insulin secretion or target critical events in the development of NAFLD downstream of the insulin receptor. Nateglinide, which stimulates insulin secretion, has been shown to improve liver histology in a cohort of five patients with NASH,136 and L‐carnitine,137 β‐aminoisobutyric acid138 and antisense oligonucleotide inhibitors of acetyl CoA carboxylases,139 all of which increase mitochondrial FFA oxidation, have shown promising results in rodent models.
Antioxidant therapies aim to prevent the lipid peroxidation thought to be central to the progression from hepatic steatosis to NASH. S‐nitroso‐N‐acetylcysteine has been shown to inhibit the development of NAFLD in association with a reduction in lipid peroxidation in rats,140 while a three month course of N‐acetylcysteine reportedly produced a significant reduction in raised transaminases in a cohort of patients with NASH.141 Pentoxifylline, an inducer of hepatic glutathione, is reported to decrease raised serum ALT and hepatic inflammation in diet‐induced steatohepatitis in mice,142 and has also been shown to be effective in reducing steatohepatitis in patients with NASH.143 Some antioxidants may intervene earlier in the pathogenesis of NAFLD, suggesting that either our understanding of the initial events is incomplete or the drugs in question have multiple therapeutic targets. The ROS scavenger manganese tetrakis 4‐benzoic acid porphyrin has been shown to rescue manganese superoxide dismutase knockout mice from hepatic steatosis144 and was recently shown to improve hepatic steatosis in ob/ob mice.145 Ursodeoxycholic acid, another inducer of hepatic glutathione,146 has been shown to prevent the development of hepatic steatosis in rats147 and reduce raised serum ALT in patients with NAFLD.148 Given the apparent hepatoprotective effects of glutathione, the absence of data regarding the effects of S‐adenosylmethionine, which has been shown to reverse decreased hepatic levels in patients with liver disease,149 is notable.
Many of the pharmacological treatments mentioned above may also exert immunoregulatory effects, exhibiting anti‐inflammatory properties through direct or indirect inhibition of pro‐inflammatory cytokines to prevent and reduce hepatic inflammation. Metformin has been shown to inhibit hepatic expression of TNF‐α, and reverse TNF‐α inducible responses that may promote hepatic steatosis and necrosis in insulin‐resistant ob/ob mice.131 L‐carnitine has been shown to reduce the induction of TNF‐α expression by alcohol in rats,150 and pentoxifylline has also been shown to reduce TNF‐α expression in rodent models of NASH.143
Despite the pressing need for a definitive treatment for NAFLD, and the apparent potential for therapeutic intervention, no reliable pharmacological agent has been found to date. This is due, in part, to a lack of well controlled, large scale, randomised trials of the drugs that have shown benefit in preliminary animal or patient studies. There is also a lack of information regarding the effects of such treatments on liver histology. Consequently current treatment places an emphasis on gradual, controlled weight reduction through a balanced diet and regular exercise.
The authors gratefully acknowledge the contribution of Professor Bernard Portmann, Consultant Histopathologist, Institute of Liver Studies, King's College Hospital, London, UK, for supplying histological pictures of hepatic steatosis and non‐alcoholic steatohepatitis.
Competing interests: None.