Fatty Liver Disease


alcoholic, nonalcoholic, fatty liver, steatosis, steatohepatitis, obesity, diabetes, metabolic syndrome


  • Chapter Contents

  • Spectrum of fatty liver disease 309

    • Steatosis 309

    • Steatohepatitis 310

    • Fibrosis and cirrhosis 313

  • Alcoholic liver disease 314

    • Epidemiology 314

    • Pathogenesis 315

    • Factors affecting individual susceptibility to alcoholic liver disease 318

    • Animal models 319

    • Clinical features 319

  • Pathology of alcoholic liver disease 320

    • Alcoholic steatosis 320

    • Alcoholic foamy degeneration 321

    • Alcoholic hepatitis 321

    • Vascular and microvascular lesions 323

    • Perivenular fibrosis 323

    • Alcoholic cirrhosis 324

    • Other morphological features of alcoholic liver injury 326

  • Diagnosis and prognosis of ALD: role of liver biopsy 327

  • Treatment of ALD and histopathological features of resolution 328

    • Abstinence from alcohol 328

    • Medical management 329

    • Liver transplantation 329

  • Nonalcoholic fatty liver disease 329

    • Historical perspectives 330

    • Epidemiology 330

    • Natural history 331

    • Factors modifying disease progression 332

    • Genetic modifiers 332

    • Pathogenesis 333

    • Clinical features 334

    • Diagnosis 334

    • Role of liver biopsy 335

  • Pathology of nonalcoholic fatty liver disease 336

    • Adult NAFL and NASH 336

    • Paediatric NAFL and NASH 342

    • NAFLD in special populations 344

    • Hepatocellular carcinoma 344

    • Grading and staging in NAFLD and NASH 346

  • Differential diagnosis of NAFLD 349

    • Alcoholic liver disease 349

  • NAFL and NASH concurrent with other liver disease 350

    • Autoimmune hepatitis 350

    • Chronic hepatitis C 350

      • Chronic hepatitis B 350

  • Histopathological features of prognostic significance 351

  • NAFL and NASH in other clinical settings 352

    • Drug- and toxin-induced fatty liver disease 352

    • Allograft liver 352

    • Total parenteral nutrition 353

    • Nutritional disorders 353

    • Inherited metabolic diseases 353

    • Coeliac disease 353

    • Miscellaneous aetiologies 353

  • Treatment of NAFLD 353

    • Weight reduction and lifestyle modification 353

    • Drug therapy 353

    • Liver transplantation for NASH-related cirrhosis 354

    • Histopathological features of NASH resolution 354

Spectrum of fatty liver disease

Thomas Addison of northern England, known for his 1836 description of the eponymous disease, was the first to use the term ‘fatty liver’ in the English medical literature, most likely referring to alcoholic liver disease (ALD). Three years later, Karl Rokitansky from Vienna observed that fat accumulation in the liver may be aetiologically related to cirrhosis. It was subsequently recognized that ALD comprises a pathological spectrum of which the earliest lesion is the accumulation of lipid in hepatocytes— fatty liver or steatosis . In a proportion of individuals, this is associated with necroinflammation and ballooning of hepatocytes, development of cytoplasmic inclusions (Mallory–Denk bodies) and a variable and mixed inflammatory infiltrate. This form of the disease is also accompanied by fibrosis with a predominantly perivenular and pericellular distribution. This constellation of histological changes is referred to as steatohepatitis . Further progression of the disease is associated with the development of a micronodular cirrhosis.

It is now clear that these features are not specific for ALD. Steatosis is very common and seen in a myriad of conditions. Furthermore, in the 1970s and 1980s, several groups reported that the entire histological spectrum seen with alcohol could occur in patients with morbid obesity, after jejunoileal bypass surgery and as an adverse drug reaction with several agents. Since then, it has become apparent that steatosis and steatohepatitis are frequently found in patients with type 2 diabetes mellitus and the metabolic syndrome. This whole spectrum is referred to as nonalcoholic fatty liver disease (NAFLD); patients with ‘simple’ steatosis have nonalcoholic fatty liver (NAFL) and those with hepatitis have nonalcoholic steatohepatitis (NASH). Some suggest that these terms should be subclassified according to aetiology, such as metabolic syndrome steatohepatitis (MESH), drug-induced steatohepatitis (DISH), chemotherapy-associated steatohepatitis (CASH) and toxin-associated steatohepatitis (TASH). These subclasses have received limited support, however, and are not widely adopted.

There are both similarities and differences between alcoholic and nonalcoholic fatty liver disease. In this chapter, we consider some of the features that are common to both. We then review the pathogenetic and pathological features of each and discuss how the two can be distinguished clinically and histopathologi­cally. It should be noted, however, that there is increasing evidence of synergistic relationships between ALD and NAFLD. Furthermore, steatosis (or the underlying metabolic abnormalities that cause steatosis) is now recognized to be an important cofactor in the development and progression of a number of chronic liver diseases.


Steatosis is characterized by the accumulation of lipid, primarily triglyceride, within the cytoplasm of hepatocytes. It is a common finding in liver biopsy specimens. The significance of small amounts is uncertain; it is often a nonspecific finding and may be a feature of the aging liver. More extensive fat accumulation occurs in a large number of primary hepatic disorders and in a variety of systemic conditions. Other forms of lipid accumulation, such as the phospholipidoses, cerebrosidoses and gangliosidoses, are not reviewed in this chapter.

In normal liver, lipid accounts for approximately 5% of total wet weight. This can increase to as much as 50% in steatosis, resulting in marked hepatomegaly (up to 5 kg). On cut section at autopsy or in explant specimens, the liver has a pale-yellow appearance and a greasy consistency ( Fig. 5.1 A ); the colour is largely caused by retained carotenes. Histologically, in routinely fixed material, the fat in hepatocytes is seen as cytoplasmic vacuoles, since the lipid is lost during processing. Lipid can be demonstrated in cryostat sections using Oil Red O or Sudan Black or in tissue that has been post-fixed in osmium tetroxide.

Figure 5.1

Fatty liver. A, Gross appearance showing the yellow colour contrasting with a normal liver. B, Macrovesicular perivenular steatosis in an alcoholic patient; there is no fibrosis (Masson trichrome stain). C, Alcoholic foamy degeneration in which both macrovesicular and microvesicular steatosis are present in the perivenular and midzones. There is also canalicular and intracellular bilirubinostasis. D, Note that the nucleus remains in a more central position in the microvesicular pattern, whereas it is displaced when a macrovesicle is present. ( C and D, Haematoxylin and eosin [H&E] stain.)

Two major patterns of steatosis are recognized on light microscopy: macrovesicular and microvesicular . In macrovesicular steatosis the hepatocytes are distended by a single droplet, which displaces the nucleus ( Fig. 5.1 B ). Uncomplicated macrovesicular steatosis has generally been regarded as a benign and fully reversible condition, although this concept has been challenged, and again, evidence suggests that it acts synergistically with other liver toxins in the induction of injury. By contrast, diffuse microvesicular steatosis is generally a serious condition ( Fig. 5.1 C and D ), with hepatic dysfunction and coma, and is often associated with impaired β-οxidation of lipids.

The distinction between macrovesicular and microvesicular steatosis is not absolute, and in some biopsies one may see a mixed pattern. This may be of prognostic importance in ALD. In some biopsies, multiple medium-sized vacuoles are seen, occasionally referred to as mediovesicular steatosis. In animal models there is a gradual transition from a microvesicular form to large-droplet steatosis by coalescence of the cytoplasmic lipid vacuoles. An autopsy study of adults with no history of liver disease has suggested that the prevalence of microvesicular steatosis may be underestimated. On the other hand, not all tiny droplets in hepatocytes on light microscopy represent microvesicular fat deposition.

Macrovesicular steatosis is the most common form; Table 5.1 outlines the major causes. The distribution of fat is variable. In alcoholic steatosis the fat is generally first seen in perivenular zones, whereas that seen in wasting diseases (e.g. HIV/AIDS), in severe protein–calorie malnutrition and after steroid therapy starts in the periportal zones. In our experience, however, there may be no apparent zonality in these conditions.

Table 5.1

Causes of predominantly macrovesicular steatosis

  • Alcoholic liver disease

  • Nonalcoholic fatty liver disease and nonalcoholic steatohepatitis

  • Obesity

    • Metabolic disorders of insulin resistance

    • Metabolic syndrome, syndrome X, insulin resistance syndrome

    • Type 2 diabetes

    • Alström syndrome, Bardet–Biedl syndrome

    • Polycystic ovary syndrome

    • Lipodystrophy syndromes; PPARγ mutations

    • Leptin deficiencies or resistance

    • Hypothalamic obesity; Prader–Willi syndrome

  • Drugs and toxins (see also Chapter 12 )

    • Nifedipine

    • Diltiazem

    • Tamoxifen

    • Oestrogens

    • Corticosteroids

    • Industrial solvent dimethylformamide

    • Paint thinners and solvents

    • Petrochemical exposure

    • Rapeseed cooking

    • Cocaine

  • Viral hepatitis

    • Hepatitis C virus

    • Hepatitis B virus (less often)

    • Lábrea fever (hepatitis D)

  • Disorders of lipoprotein metabolism

    • Abetalipoproteinaemia

    • Familial hypobetalipoproteinaemia

    • Dorfman–Chanarin syndrome

    • Pseudoneonatal adrenoleucodystrophy

  • Nutritional disorders other than obesity

    • Total parenteral nutrition

    • Protein–energy malnutrition (kwashiorkor)

    • Pancreatic insufficiency with bone marrow suppression, Shwachman syndrome

    • Obesity surgery: jejunoileal bypass, gastric bypass, gastroplasty

    • Rapid weight loss

    • Postsurgical: biliopancreatic bypass, small bowel resections

  • Systemic disorders

    • Cachexia, febrile illnesses, heat stroke

    • Inflammatory bowel disease

    • Weber–Christian disease

    • Cystic fibrosis

    • Pituitary dysfunction; craniopharyngioma

  • Non-insulin-related metabolic or endocrine disorders

    • Wilson disease

    • Galactosaemia

    • Tyrosinaemia

    • Hereditary fructose intolerance

    • Cystinuria

    • Sandhoff disease

    • Pituitary dysfunction

    • Hypothyroidism

  • Other

    • Hepatic ischaemia

    • Small bowel diverticulosis with bacterial overgrowth

    • Age

Table 5.2 lists the causes of microvesicular steatosis; the first to be recognized was acute fatty liver of pregnancy. Microvesicular steatosis was subsequently identified as a key feature in Reye syndrome, and more recently it has been described with a variety of therapeutic agents. A feature common to all forms of microvesicular steatosis is impairment of mitochondrial β-oxidation. This is the process by which acyl-coenzyme A (CoA) derived from long-, medium- and short-chain fatty acids is progressively cleaved into acetyl-CoA units. The process leads to the generation of reducing equivalents for oxidative phosphorylation and involves five complexes found within the inner mitochondrial membranes; some of the subunits of these complexes are encoded by mitochondrial DNA (mtDNA). Some inherited metabolic conditions associated with microvesicular steatosis, notably some of the mitochondrial cytopathies, are associated with mtDNA mutations or deletions. Drug-induced microvesicular steatosis may involve direct inhibition of β-oxidation by drug metabolites. Chronic alcohol abuse may lead to acquired mtDNA deletions and may be associated with so-called alcoholic foamy degeneration.

Table 5.2

Causes of microvesicular steatosis

  • Acute fatty liver of pregnancy

  • Reye syndrome

  • Drugs

    • Salicylate toxicity

    • Sodium valproate toxicity

    • Tetracycline toxicity

    • 2-Arylpropionic acids (e.g. ketoprofen)

    • Amineptine

    • MDMA (ecstasy)

    • Didanosine

    • Fialuridine

    • Stavudine

    • Linezolid

    • l -asparaginase

    • Propofol

    • Acute iron toxicity

    • Aflatoxins

  • Jamaican vomiting disease

  • Multiple hornet stings

  • Alcoholic foamy degeneration

  • Inherited urea cycle disorders

  • Inherited disorders of fatty acid metabolism

  • Mitochondrial cytopathies

  • Wolman disease

  • Cholesterol ester storage disease

  • Hepatitis D infection in Amazonian Indians (Lábrea fever)

  • Bacillus cereus toxins

  • Navajo neuropathy

  • Pearson syndrome

Focal fatty change

Steatosis is normally a diffuse process, but a more localized form was first described in 1980. Since then, steatosis has become increasingly recognized, as revealed by imaging techniques such as ultrasound or computed tomography, and may simulate metastatic disease. The lesions may be single or multiple, are typically “geographical” in shape but may sometimes be nodular, and they have occasionally been associated with other hepatic abnormalities such as cirrhosis. However, the basic architecture of the liver is usually preserved, and although some patients have factors predisposing to generalized macrovesicular steatosis, the mechanisms leading to a nonhomogeneous distribution of fat are not clear. Local hypoxia has been postulated to play a role, however, as has a direct, local effect of insulin from an aberrant pancreatic vein. The lesion must be distinguished from other intrahepatic, lipid-rich lesions, such as lipoma, angiolipoma, myelolipoma and coelomic fat ectopia. Rarely, the converse can be observed, with areas of sparing in an otherwise diffuse steatosis ; this generally occurs adjacent to the gallbladder or near the porta hepatis in areas receiving blood from an aberrant gastric vein.


Steatohepatitis refers to a histological constellation of findings with evidence of additional modes of cell injury, cytoskeletal disruption, cell death and accompanying inflammation. These changes are frequently accompanied by fibrosis, and this stage of liver disease is recognized as a precursor for development of cirrhosis. Although the degree to which such changes differ depends slightly on the aetiology of steatohepatitis, those discussed next are common to all.

Hepatocellular ballooning

Ballooned hepatocytes are a common feature in steatohepatitis and are currently considered an essential finding for the diagnosis of steatohepatitis. These cells are usually increased in size, have a rounded outline and a rarified, clear, ‘oedematous’ cytoplasm and may have a hyperchromatic nucleus and thickened cell membrane ( Fig. 5.2 ). However, smaller hepatocytes may share similar cytoplasmic alterations and can be considered ‘ballooned’. Ballooned hepatocytes are not specific for fatty liver disease and can also be seen in viral hepatitis and chronic cholestasis. Ballooning may contribute to the development of hepatomegaly and may have a direct functional effect; some studies reported a correlation between hepatocyte enlargement and intrahepatic pressure in alcoholic fatty liver disease, although this was not confirmed by others.

Figure 5.2

Hepatocyte ballooning. Many of the cells in this field are enlarged with rarefaction of the cytoplasm. Some of the cell swelling is the result of the accumulation of fat, but there is also excess cytoplasmic fluid. (H&E stain.)

Ballooned hepatocytes are recognized most easily when they contain Mallory–Denk bodies, formerly referred to as ‘Mallory’s hyalin’ or ‘alcoholic hyaline’ (see next section). However, not all ballooned hepatocytes contain these intracytoplasmic cytoskeletal aggregates, and some may contain fat droplets. Ultimately, ballooning may be a structural manifestation of microtubular disruption. Whether this feature represents cells undergoing lytic necrosis (as in viral hepatitis), oxidative damage or possibly an adaptive change similar to ‘brown fat’ remains to be resolved. Altered intracytoplasmic cytoskeletal contents of hepatocytes, i.e. keratins 8 (K8) and 18 (K18), are well recognized. On light microscopy, ballooned hepatocytes typically are enlarged and have altered, reticulated cytoplasm. Lackner et al. have proposed the use of the K8/18 immunohistochemical (IHC) stain as a method for verification of hepatocyte ‘clearing’ resulting from loss of diffuse intracytoplasmic keratins in ballooning. This stain also highlights the aggregated K8/18 in Mallory–Denk bodies. This immunostain is routinely available, and its application in practice may be helpful ( Fig. 5.3 ).

Figure 5.3

Ballooning can be detected by the lack of cytoplasmic keratins 8 and 18 (K8/18) as well as the presence of clumped keratin aggregates (K8/18 immunohistochemical stain).

Mallory–Denk bodies

Mallory first described aggregates of amorphous, eosinophilic material in hepatocytes in alcohol-related liver disease in 1911. These inclusions are now often referred to as Mallory–Denk bodies (MDBs) in recognition of Helmut Denk’s work in this field. MDBs are a feature of both alcoholic and nonalcoholic steatohepatitis but are also seen in association with prolonged cholestasis, which occurs in diseases such as primary biliary cirrhosis, Wilson disease, Indian childhood cirrhosis, focal nodular hyperplasia and hepatocellular carcinoma. MDBs have been produced experimentally using a variety of agents, including griseofulvin, diethylnitrosamine and 3,5-diethoxy carbonyl-1,4 dihydrocollidine (DDC).

Three ultrastructurally distinct forms of MDBs have been described : type 1 has bundles of filaments in parallel arrays, type 2 is seen as clusters of randomly oriented fibrils, and type 3 is identified as granular or amorphous substance containing only scattered fibrils. The filament thickness varies from 5 to 20 nm, depending on the method of measurement. Type 2 MDBs are the type seen most frequently in alcohol-related liver disease ( Fig. 5.4 ). Purified from human liver, MDBs are composed predominantly of protein; five polypeptide bands have been detected by electrophoretic analysis, with molecular weights ranging from 32 to 56 kD. Small amounts of carbohydrate have also been detected. MDB formation is thought to result from derangement of the intermediate filament component of the cytoskeleton of the hepatocyte, but they are not composed solely of intermediate filament proteins. Antibodies to MDBs react with the cytoplasmic filament system of normal hepatocytes (K8, K18), and they also contain a number of unique antigenic determinants, together with K19 and K20. The keratin molecules may be hyperphosphorylated. MDBs from alcoholic and nonalcoholic liver disease have been shown to share a common antigenic determinant.

Figure 5.4

Mallory–Denk body (MDB) type 2. A, The MDB consists of a mass of randomly orientated filaments. Several mitochondria, of varying sizes and shapes, are also seen. (Magnification ×25,100) B, Higher magnification shows the randomly arranged filaments (×57,000).

Hypotheses concerning the pathogenesis of MDB formation are outlined in a number of excellent reviews. Extensive literature addresses the biology of MDBs in animal models, as well as gender and strain differences in their formation. Rather than reflecting a passive process of collapse of intracytoplasmic structures, MDBs result from concerted, active metabolic pathways that include transamidation, hyperphosphorylation and ubiquitination of abnormally folded intermediate filaments associated with the chaperone heat shock proteins. A protein product of an early-response gene, p62 (sequestosome-SQSTM1) associates with MDBs; this process may represent a hepatoprotective sequestration of the abnormal proteins into biologically inert inclusions.

The possible detrimental effects of alcohol and its metabolic products on proteasome dysfunction have been highlighted. Hirano et al. have designed an in vitro system in which MDBs can be generated; this is likely to lead to further opportunities to establish their pathogenesis. In addition to IHC stain to K8/18, anti-ubiquitin and anti-p62 antibodies can be used to demonstrate the presence of MDBs ( Fig. 5.5 ).

Figure 5.5

Immunoreactivity of Mallory–Denk bodies to p62.


Ballooned hepatocytes in steatohepatitis may be a manifestation of progressive injury leading to lytic necrosis. Such cells, however, including those containing MDBs, are not destined to immediate cell death, and animal studies in particular have shown that these sublethal changes may persist for some time, even months. This has implications in the diagnostic setting, where histological changes of alcoholic hepatitis can remain for such duration despite abstinence. Cell death in steatohepatitis also occurs by apoptosis ( Fig. 5.6 ), and a relationship exists between apoptotic cell numbers and the severity of disease. Serum levels of K18 fragments from apoptotic hepatocytes are now being used as surrogate markers to predict severity. Confluent and bridging necrosis are rarely seen in fatty liver disease but can be observed in severe ALD—so-called central sclerosing hyaline necrosis (see later)—and was seen in some cases of steatohepatitis after jejunoileal bypass surgery. The inflammation that accompanies the cell injury is variable in both intensity and cellular nature. In most fatty liver diseases, lobular inflammation is mixed, with lymphocytes, macrophages and neutrophil polymorphs all being present ( Fig. 5.7 ). The number of neutrophils is generally greater in alcoholic steatohepatitis than in NASH, but in both, they may be seen to encircle ballooned hepatocytes containing MDBs, so-called satellitosis ( Fig. 5.8 ). It has been postulated that this may occur through a chemotactic effect of the inclusions, although current concepts suggest that it more likely is associated with local chemokine release, including interleukin-8 (IL-8). Portal tract inflammation may occur in all forms of steatohepatitis. It is predominantly lymphocytic in NASH ( Fig. 5.9 ) but may be mixed in ALD. Biopsies from a subset of patients with ALD show a predominantly portal-based inflammation, possibly reflecting another mechanism of injury.

Figure 5.6

Alcoholic steatohepatitis. The hepatocytes show fatty change, and focal necrosis is accompanied by a mixed inflammatory cell infiltrate. An apoptotic cell (centre) is also seen. (H&E stain.)

Figure 5.7

Steatohepatitis. In this field there is a mixed lobular inflammatory infiltrate. Neutrophil polymorphs are present, but there are also lymphocytes and occasional activated Kupffer cells. (H&E stain.)

Figure 5.8

Satellitosis. This micrograph shows a single hepatocyte which contains a poorly formed Mallory–Denk body. The cell is surrounded by neutrophil polymorphs. (H&E stain.)

Figure 5.9

Portal inflammation in nonalcoholic steatohepatitis. This is mild and not accompanied by significant interface hepatitis. (H&E stain.)

Fibrosis and cirrhosis

In common with other forms of chronic liver disease, the necroinflammation of steatohepatitis is normally accompanied by hepatic fibrosis, which reflects an imbalance between the production and the degradation of extracellular matrix (ECM). In the normal liver the sinusoidal ECM consists predominantly of a basement membrane-like matrix containing non-fibril-forming collagens, types IV and VI, and small amounts of collagen types I and III, whereas in fibrosis and cirrhosis, as in cirrhosis of other aetiologies, fibronectin, laminin and type III collagen, with lesser amounts of type I collagen, as well as proteoglycans, tenascin, decorin and biglycan, are deposited in the space of Disse. The cell biological basis for this exaggerated repair process is discussed in Chapter 1 and in a number of excellent review articles on hepatic fibrogenesis.

Mechanisms that may be specific for alcohol-induced injury have been reviewed by Siegmund and Brenner and Wang et al. Fatty liver diseases are associated with several highly characteristic but not entirely specific patterns of fibrosis. The early lesion may be that of perivenular fibrosis ( Fig. 5.10 A ), whereas a pericellular fibrosis (‘chickenwire’ pattern of fibrosis) may be seen in association with the perivenular fibrosis ( Fig. 5.10 B ). The perivenular/pericellular fibrosis constellation is generally seen in association with active cell damage, surrounding ballooned cells, but it may be seen in the absence of significant necroinflammation; in this setting it may reflect previous ‘acute’ events with subsequent partial resolution.

Figure 5.10

Hepatic fibrosis. A, Mild perivenular fibrosis and mild pericellular fibrosis in zone 3. B, More extensive pericellular fibrosis is seen as thin strands that are located in the perisinusoidal space of Disse and surround cords of hepatocytes. (Picrosirius red stain.)

The perivenular changes may be accompanied by portal and periportal fibrosis in all forms of fatty liver disease. Isolated portal fibrosis can also occur in both alcoholic and nonalcoholic fatty liver disease. In progressive disease, bridging fibrosis ensues with the formation of central-central, central-portal and portal-portal septa and the formation of a predominantly micronodular cirrhosis. The nodules are characteristically small in alcoholic cirrhosis; this may partly result from the inhibitory effect of alcohol on liver cell regeneration. In end-stage disease, particularly that associated with alcohol, there may be large amounts of parenchymal extinction. This is likely to reflect secondary vascular events (see later discussion and Chapter 1 ).

Alcoholic liver disease


Excess alcohol consumption is the principal cause of liver-related mortality in Western countries. Archaeological records of the earliest civilizations show that the history of alcohol dates back to about 10,000 bce . Many reviews of the epidemiology of alcoholic liver disease (ALD) document a direct relationship between alcohol consumption per capita and mortality from alcoholic cirrhosis. In addition, alcohol, even when not causing overt liver disease, can hasten the progression of other forms of chronic liver disease in terms of progression, risk of malignancy and morbidity; this is particularly true of chronic hepatitis C (CHC). Thus the true impact of alcohol on health and the economic burden worldwide are certainly underestimated. There was evidence of a decline in the prevalence of ALD in some countries, but this appears to have stabilized, and in others (e.g. United Kingdom) there is a steady rise, with alcohol responsible for more than 80% of cirrhosis cases in Scotland.

Although alcohol is undoubtedly a direct hepatotoxin, it is now accepted that both genetic and environmental factors influence the susceptibility of the individual to alcohol-associated liver injury. Racial differences in alcohol metabolism and possibly susceptibility are associated with genetic polymorphisms of alcohol dehydrogenase and acetaldehyde dehydrogenase.


Early biochemical studies showed a direct metabolic effect of alcohol to explain the development of steatosis; this could be reproducibly induced in nonalcoholic individuals and laboratory animals. The steatosis was shown to be reversible on cessation of alcohol. Because only some alcoholic individuals progress to steatohepatitis, the concept of a two-hit phenomenon was postulated to explain the development of more severe injury. The situation, however, appears more complex. Steatosis itself may not be entirely ‘benign’ and may contribute to the necroinflammation; furthermore, some of the factors previously thought to be potential mediators of the second hit have now been shown to contribute to the steatosis. Recent studies have attempted to define the proteomic signature of alcoholic hepatitis. Despite years of ongoing research, the mechanisms that contribute to disease progression remain poorly understood. Several excellent recent reviews summarize current paradigms for alcohol-induced injury.

Alcohol metabolites and associated biochemical disturbances

Alcohol metabolism and its associated disturbances can at least partly explain some features of alcohol-induced injury. Alcohol is readily absorbed from the gastrointestinal tract and is distributed throughout the body in proportion to the amount of fluid in the tissues. Less than 10% is eliminated through the lungs, kidneys and sweat; the remainder is oxidized predominantly in the liver, which probably explains the marked metabolic disturbances that occur there.

Three pathways for alcohol metabolism exist in the liver. The principal pathway involves cytosolic alcohol dehydrogenase (ADH), which catalyses the oxidative metabolism of alcohol to acetaldehyde. Hydrogen is transferred from alcohol to the cofactor nicotinamide adenine dinucleotide (NAD), converting it to the reduced form, and acetaldehyde is produced. The generation of excess reducing equivalents (NADH) in the cytosol results in a marked shift in the oxidation-reduction (redox) potential, as indicated by the increased lactate/pyruvate ratio. Some of the hydrogen equivalents are transferred from the cytosol to the mitochondria via several shuttle systems.

Acetaldehyde generated by the metabolism of alcohol is oxidized to acetate; again, this occurs predominantly in the liver. It occurs through the action of mitochondrial aldehyde dehydrogenase (ALDH2). Some of the acetaldehyde and most of the acetate are excreted by the liver into the bloodstream and metabolized peripherally. Patients with chronic alcoholism have accelerated alcohol metabolism and higher levels of blood acetaldehyde than nondrinkers. The metabolism of acetaldehyde produces excess NADH in liver mitochondria, which in turn leads to a decrease in β-oxidation of long-chain fatty acids by inhibiting long-chain 3-hydroxyacyl-CoA dehydrogenase (LCHAD) activity.

Chronic alcohol consumption results in high levels of acetaldehyde, which is highly toxic due to a variety of mechanisms, including glutathione depletion, oxidative stress, lipid peroxidation and formation of adducts with proteins.

Chronic alcohol consumption accelerates ADH-related alcohol metabolism to a limited extent. ADH activity does not increase, however, and a hypermetabolic state has been proposed as a possible mechanism. There are multiple isoenzymes of ADH, which are divided into five major classes and encoded by at least seven different gene loci.

The microsomal ethanol-oxidizing system (MEOS) is a cytochrome P-450-dependent pathway and in particular involves the ethanol-inducible cytochrome P-450 2E1 (CYP2E1). CYP2E1 was initially thought to be the only isoenzyme involved in ethanol metabolism, but ethanol can also induce CYP1A1, CYP3A and CYP4A. Increased CYP2E1 activity after chronic alcohol consumption is probably the major mechanism for the increased rates of alcohol clearance from the blood and metabolic tolerance that develop in regular drinkers. The increased xenobiotic toxicity and carcinogenicity seen in association with chronic alcohol consumption, in both humans and various animal models, can be explained largely by the induction of CYP2E1, which leads to enhanced metabolism of a wide variety of agents, e.g. paracetamol (acetaminophen), that are metabolized by this isoenzyme.

Catalase, located in peroxisomes, plays only a minor role in alcohol metabolism, possibly accounting for <2% of overall alcohol oxidation, although there is some evidence for increased metabolism of alcohol by peroxisomes in chronic drinkers.

Alcohol metabolism may cause lactic acidosis, with a reduced renal capacity for uric acid excretion producing a secondary hyper­uricaemia. There is also impaired carbohydrate metabolism with reduced gluconeogenesis from amino acids; this may produce hypoglycaemia. Of particular importance to the development of steatosis is the related impaired fat metabolism: H + ions replace two-carbon fragments derived from fatty acids as the main energy source in hepatocyte mitochondria and also depress the citric acid cycle. There is decreased fatty acid oxidation; an increase in α-glycerophosphate also occurs, with a consequent increase in trapping of fatty acids, and an accompanying increased synthesis of triglycerides. In longstanding alcohol abuse, protein synthesis is depressed, which together with impaired liver cell secretory function, causes retention of lipoproteins and contributes to the accumulation of fat in hepatocytes.

Molecular mechanisms of alcohol-induced steatosis

The biochemical events associated with alcohol metabolism just outlined are now thought to be insufficient to fully explain the initiation and perpetuation of alcohol-induced steatosis. In animal models, for example, the redox shifts are attenuated over time, but the steatosis remains. A number of alternative pathways have been identified that regulate lipid metabolism and with which alcohol has been found to interact ( Fig. 5.11 ).

Figure 5.11

Alcoholic metabolism, transcriptional control of lipid metabolism and endoplasmic reticulum (ER) stress; outline of mechanisms. AMPK, Adenosine monophosphate-activated protein kinase; FA, fatty acid; MTP, microsomal triglyceride transfer protein; PPARα, peroxisome proliferator-activated receptor α; SREBP-1, sterol regulatory element-binding protein 1.

Peroxisome proliferator-activated receptor α (PPARα) is a nuclear hormone receptor which regulates the oxidation and transport of fatty acids. PPARα agonists have been shown to reverse steatosis in alcohol-fed mice. Furthermore, PPARα knockout mice display impaired fatty acid oxidation with increased serum free fatty acids and develop steatosis. Liver injury is more severe in alcohol-treated knockout animals compared to wild-type controls. PPARα plays an important role in fatty acid transport through its effects on microsomal triglyceride transfer protein (MTP), a key molecule in the assembly of very-low-density lipoproteins (VLDLs) before export, and alcohol-induced alterations to PPARα activity could contribute to lipid accumulation through this route.

The adenosine monophosphate (AMP)-activated protein kinase (AMPK) enzyme is a master regulator of metabolism that acts as a sensor for various forms of cellular stress, causing increased activity of energy-generating metabolic pathways, including fatty acid oxidation. AMPK has been found to be decreased in alcohol-fed animals, and over time this was associated with accumulation of SREBP. The effect of alcohol on AMPK may therefore contribute to both increased fatty acid synthesis and decreased fatty acid oxidation, contributing to steatosis.

Sterol regulatory element-binding proteins (SREBPs) represent a family of transcription factors that regulate the genes for enzymes responsible for the synthesis of cholesterol, fatty acids and triglyceride. SREBP-1 isoforms are involved in regulating fatty acid synthesis while SREBP-2 controls cholesterol synthesis. Alcohol feeding of mice led to increased intrahepatic levels of SREBP-1 which correlated with triglyceride accumulation. AMPK inhibits SREBP-1c, and therefore reduced AMPK levels caused by alcohol may contribute to elevating SREBP-1 levels and activity in steatosis.

Alcohol also interferes with liver methionine metabolism. Normally, methionine is converted to S -adenosylmethionine and then to S -adenosylhomocysteine and finally to homocysteine; conversion back to methionine involves methionine synthase. Homocysteine and S -adenosylhomocysteine are elevated in alcoholic patients and are thought to induce the endoplasmic reticulum (ER) stress response. This involves the accumulation of unfolded and misfolded proteins triggering the ‘unfolded protein response’. The ER stress response may play an important role in the pathogenesis of alcohol-induced steatosis but may also contribute to hepatocyte cell death through activation of proapoptotic genes.

Sirtuin-1 is another master metabolic regulator. In experimental animals, alcohol exposure led to decreased levels of SIRT-1, blocking the deacetylation of SREBP with the resultant increase in SREBP-1c, and in turn increased fatty acid synthesis.

There is a growing list of additional pathways that have been implicated in the development of alcohol-induced steatosis. This includes factors such as tumor necrosis factor alpha (TNFα), long considered to be involved in necroinflammation but now also implicated in steatosis. Others that may either exacerbate or protect from lipid accumulation include adiponectin, IL-6, plasminogen activator inhibitor 1, endocannabinoids and complement factors.

Mitochondrial abnormalities, oxidant stress and lipid peroxidation

Nassir and Ibdah recently reviewed mitochondrial alterations in alcohol-related liver disease. Impairment of liver mitochondrial function after chronic alcohol exposure is well documented. Isolated mitochondria from experimental animals following chronic alcohol administration are more likely to increase the permeability transition in response to a range of cellular stress. Oxidative phosphorylation is impaired because of a defect in protein synthesis encoded by mtDNA. As noted later, mtDNA mutations may be acquired, and morphological changes may occur in the structure of the organelle itself, manifest histologically by the formation of giant mitochondria ( Fig. 5.12 ). The cytokine IL-6 may play a critical role in mtDNA repair after alcohol-induced injury.

Figure 5.12

Giant mitochondria containing crystalline inclusions. A number of normal-sized mitochondria are also seen, some with inclusions. (×17,500.)

As noted earlier, chronic ethanol ingestion induces CYP2E1; this results in enhanced lipid peroxidation and increased rates of reactive oxygen species (ROS) production. Mitochondria are both the source and the target of ROS. Other sources of ROS include endotoxin-activated Kupffer cells and neutrophil polymorphs. Chronic ethanol ingestion simultaneously results in enhanced acetaldehyde production, caused by CYP2E1 induction, and depletion of glutathione, which impairs a major defence mechanism against oxidative stress. Comprehensive reviews further discuss the role of oxidative stress.

Mitochondrial dysfunction can also promote both apoptosis and hepatocyte necrosis. Apoptotic cells can be seen in liver biopsies from patients with ALD (see Fig. 5.6 ), and DNA end-nick labelling has been used to confirm the presence of apoptotic bodies. Apoptosis in ALD is probably triggered by oxidative stress as well as by a variety of cytokines. Autophagy may be a defence mechanism within hepatocytes that protects against alcohol-induced apoptosis (and steatosis).

Protein hyperacetylation

Experimental studies have demonstrated that exposure to alcohol can cause global hyperacetylation. This is associated with a reversible, post-translational modification of the ε-amino groups of lysine residues, shown to modulate a diverse range of cellular processes, from transcriptional activation to microtubule stability. The lysine acetylation is controlled by two counteracting enzymes, histone acetyltransferases and histone deacetylases. The precise role of such molecular alterations in ALD is uncertain, but recent work has demonstrated that acetyl-CoA generated from acetate (derived from alcohol metabolism) increases acetylation of proinflammatory gene histones in macrophages, enhancing their inflammatory response.

Innate immunity

Rao et al. reviewed the now considerable evidence supporting the concept that endotoxaemia, resulting from alcohol-induced leakiness of the small intestine and bacterial translocation, plays an important role in the pathogenesis of ALD. The importance of perturbations in the gut–liver axis in the induction of steatosis, necroinflammation and fibrosis is now well accepted, including increasing evidence that altered gut microbiota may be important. Endotoxins, which consist of lipopolysaccharide (LPS) derived from the walls of gram-negative bacteria, normally permeate only to a very limited extent across the mucosa of the gut. Any that is transported in the portal venous system is cleared by hepatocytes and Kupffer cells through a number of important LPS recognition systems that include the binding of LPS to surface receptors, enzymatic degradation of the lipid A moiety and inactivation of LPS itself.

When LPS levels exceed those that can be dealt with by homeostatic mechanisms, they can activate Kupffer cells through Toll-like receptors (TLRs), in particular through binding to TLR4, leading to cytokine production and generation of ROS. Animal studies have demonstrated an increased susceptibility of female rats to alcohol/endotoxin-induced liver injury. Acetaldehyde has been shown to increase intestinal permeability by altering the distribution of tight junction proteins and adherens junction proteins on enterocytes. Alcohol may also exert this effect indirectly through increased levels of microRNA-212 that decreases ZO-1 expression. Numerous human and animal studies of ALD have demonstrated an increased production of proinflammatory cytokines such as TNFα (produced primarily by activated Kupffer cells), IL-1, IL-6 and osteopontin. The effects of TNFα are responsible for many of the clinical features seen in alcoholic hepatitis (e.g. fever, neutrophilia, even hypotension in severe cases).

Kupffer cells that have been activated by endotoxin show enhanced production of macrophage inflammatory protein 2 (MIP2), which is a potent chemotactic agent for polymorphs. Endotoxin also upregulates the expression of intercellular adhesion molecule 1 (ICAM1) on hepatocytes. The alcohol-related influx of endotoxin into the circulation is thought to lead to an enhanced interaction between adhesion molecules on neutrophils and ICAM1 on hepatocytes, thus contributing to liver injury and to the migration of polymorphs into the liver. Furthermore, monocyte secretion of MIP1 and MIP1αI is increased in alcoholic hepatitis.

Disturbances of IgA metabolism, with elevated total serum immunoglobulin A and a characteristic linear deposition of IgA in the hepatic sinusoids, are well described in ALD. IgA has also been shown to trigger the secretion of TNFα by monocytes and synergistically enhance the endotoxin-induced secretion of this cytokine.

The complement system forms an important part of innate immunity, and the possible role of its components in the pathogenesis of alcoholic hepatitis has been studied. The complexities of the interactive role of cytokines and inflammatory mediators is discussed in several comprehensive reviews.

Adaptive immunity and autoimmune phenomena

Many immunological disturbances occur as a result of alcoholic liver injury and may be considered secondary events. Nevertheless, some morphological features suggest that acquired immunological mechanisms sometimes play a primary role in alcohol-related liver injury.

The neutrophil polymorph infiltrate seen in and around hepatocytes, which contain Mallory–Denk bodies, could indicate a local Arthus reaction with humoral sensitization to MDBs. An antibody to MDB antigen has been reported in alcoholic hepatitis but was not demonstrated in patients with fatty liver or inactive cirrhosis. Hypergammaglobulinaemia in ALD is characterized by a polyclonal increase in all classes of immunoglobulins, but with serum IgA levels being particularly increased and correlating closely with the severity of liver damage.

Low titres of autoantibodies have been reported in ALD, with antinuclear factor in 12–17% and anti-smooth muscle autoantibodies in 12–27% of patients. IgG and IgA antibodies to liver cell membrane antigens (LMAs) have been demonstrated in all stages of alcoholic liver injury. IgG class antibodies reacting with ethanol-altered rabbit hepatocytes were detected in more than 70% of patients with various stages of ALD but in only 25% of those with nonalcoholic liver disease.

A number of studies have suggested that an immune response is directed against acetaldehyde–protein adducts formed during the metabolism of alcohol. Different proteins have been detected in these adducts in hepatocytes and include a 37-kD protein in the cytosol and CYP2E1 in microsomes. IHC studies have demonstrated the presence of protein–acetaldehyde adducts in perivenular hepatocytes at an early stage of alcoholic injury. In addition, malondialdehyde and acetaldehyde can react together with proteins to form hybrid adducts termed ‘malondialdehyde–acetaldehyde adducts’ (MAAs). Albano and Vidali have reviewed the complex interactions between lipid peroxidation and the induction of autoantibody formation against adducts.


Alcohol-induced liver injury selectively affects the perivenular region in the early stages. It has been postulated that a relatively lower oxygen tension in this zone may exaggerate the shift in redox potential that accompanies alcohol metabolism. Hypoxia has been shown to induce perivenular necrosis in rats chronically fed ethanol. Israel et al. likened the acceleration of alcohol metabolism with an associated enhancement of oxygen requirement to a ‘hypermetabolic state’, and this has been the basis for animal and human studies of propylthiouracil.

Lieber et al. suggested that impaired oxygen utilization, rather than lack of oxygen supply, and enhanced alcohol metabolism in this zone are factors in alcohol-associated liver injury. Fluctuations in blood flow have also been suggested as a mechanism for hypoxia-related injury in animal models for ALD; French comprehensively reviewed the role of hypoxia in liver injury

Nutritional factors

A number of reviews have addressed the interaction among nutritional factors, alcohol and liver disease in humans. Primary malnutrition caused by poor diet is well recognized in the chronic drinker. Moreover, secondary malnutrition is now well recognized in drinkers in whom the diet is adequate.

Mezey examined the role of dietary fat in ALD; the severity of alcoholic steatosis is clearly related to the amount of fat in the diet. Patients with biopsy evidence of ALD have lower levels of polyunsaturated fatty acids in their livers than chronic drinkers who have morphologically normal livers.

The recognition that metabolic and nutritional factors contribute to the pathogenesis of NASH (see later) has stimulated renewed interest in the role of nutritional factors in the pathogenesis of ALD. There is now clear evidence that obesity potentiates the severity of alcohol-related hepatic injury, although it is not yet clear whether the effect is additive or synergistic. Naveau et al. have shown that adipose tissue is inflamed in patients with ALD and that the amount correlates with the severity of liver injury, postulating that this may be another source of proinflammatory cytokines. In various animal models, manipulation of the type and amount of fat and carbohydrate in the diet can influence the development and severity of liver injury.

Factors affecting individual susceptibility to alcoholic liver disease

Lelbach clearly demonstrated a relationship between the risk of developing cirrhosis and the dose and duration of alcohol consumption. Nevertheless, even with an intake as high as 226 g/day for a mean duration of 11.4 years, only 25% of drinkers develop cirrhosis. On the other hand, a daily intake as low as 20 g for women and 40 g for men has been associated with development of cirrhosis.

A study of the total lifetime alcohol intake in patients with biopsy-proven ‘NAFLD’ showed that 13% had a significantly higher alcohol intake than obtained in the original history, suggesting that the alcohol intake may have been sufficient to have at least been contributing to the pathogenesis of the liver injury. This type of study confirms the need for a reliable alcohol history to enable both the clinician and the pathologist to differentiate between alcoholic and ‘nonalcoholic’ liver disease.

The pattern of drinking may be important: an open population study in Italy of 6534 participants, who did not have evidence of viral hepatitis, suggested that drinking alcohol outside mealtimes and drinking different types of alcohol increased the risk of ALD.

Analysis of the quantitative relationship between alcohol consumption and cirrhosis led Sorensen to consider whether alcohol abuse has a permissive rather than a dose-related effect. Sorensen et al. suggested that the risk of cirrhosis at a given level of alcohol consumption is higher or lower, depending on the action of other factors. Such cofactors could change the threshold for alcohol-associated liver injury, as well as the level of risk with higher alcohol consumption. The group studied developed cirrhosis at a constant rate of 2% per year. This observation led to the suggestion that other contributing factors affect approximately one in every 50 drinkers every year.

A meta-analysis of 23 published liver biopsy studies of 5448 drinkers from many countries confirmed the greater susceptibility of women to the hepatotoxic effects of alcohol. The increased susceptibility to ALD has the following manifestations:

  • 1.

    Women tend to present with more severe disease, often associated with a lower daily intake of alcohol for a shorter duration.

  • 2.

    Women, particularly those younger than 45, have a higher incidence of alcoholic hepatitis and a worse long-term prognosis, even if they abstain.

  • 3.

    Women with alcoholic cirrhosis have a higher mortality.

It is not unusual to see liver biopsies from women in their 20s and 30s that show severe alcohol-related liver injury. Progression to cirrhosis, liver failure and death can occur in 1–2 years if drinking continues ( Fig. 5.13 ).

Figure 5.13

Alcoholic hepatitis and cirrhosis in a 32-year-old woman who continued to drink and died several months later from bleeding varices. A, The liver shows steatosis and alcoholic hepatitis (H&E stain). B, Established micronodular cirrhosis is apparent (picrosirius red stain). A liver biopsy 2 years earlier showed only fatty change and mild pericellular fibrosis.

Obesity is a risk factor for ALD. Similarly, the increased prevalence of NASH in women, which in turn is related to the increased prevalence of obesity and diabetes mellitus type 2, may have some role in the increased susceptibility of women to ALD. Excess weight (body mass index [BMI] ≥23 in women and ≥27 in men) has been reported as a risk factor for ALD.

Thurman observed that (1) alcoholic liver injury involves the activation of Kupffer cells by endotoxin; (2) female rats are more sensitive than male rats to alcohol-induced injury, possibly from increased levels of endotoxin in the serum and ICAM-1 expression in hepatic sinusoidal-lining cells; and (3) oestrogen increases the sensitivity of Kupffer cells to endotoxin. These observations provide further insights into factors that may contribute to the greater susceptibility of women to alcohol-induced liver injury. Frezza et al. suggested that the increased susceptibility of women could be linked to gender-related differences in first-pass metabolism. Eagon has reviewed other possible mechanisms to explain the gender differences.

Numerous studies have demonstrated racial differences in alcohol-metabolizing enzymes, suggesting that polymorphisms at the alcohol and aldehyde dehydrogenase loci not only play a role in genetic predisposition to alcoholism and alcoholic liver injury but also could act as protective factors against alcohol abuse. An atypical ALDH2 isoenzyme ( ALDH2*1/*2 ) is widely prevalent among Japanese and Chinese populations; the ‘deficiency’ of ALDH2 results in a lower rate of elimination of alcohol, which is mainly responsible for their alcohol ‘sensitivity’, manifesting as facial flushing, sweating, headache and increased pulse rate. This sensitivity to alcohol generally acts as a deterrent against drinking, which no doubt accounts for the lower incidence of alcoholism and alcohol-related diseases in Oriental races. For a more detailed discussion about the correlation between genetic variants in alcohol-metabolizing enzymes, see reviews by Crabb et al., Stickel and Osterreicher and Lieber.

Polymorphisms of genes encoding for ADH, ALDH and CYP2E1 appear to play only a minor role in the susceptibility of Caucasians to ALD. Polymorphisms in cytokine genes may be more important. For example, polymorphisms in the promoter region of IL-10 have been described. A clinical study of 287 heavy drinkers (>80 g alcohol daily for at least 10 years), compared to 212 controls, showed that the presence of the A allele at position −627 in the IL-10 promoter was associated with an increased risk of advanced ALD. A Japanese study found that genetic polymorphisms of IL-1β was associated with ALD development.

Although past studies of the association between chronic hepatitis B virus (HBV) infection and alcohol-induced liver injury reported conflicting results, a number of studies have shown an increased incidence of serological markers for hepatitis B in patients with ALD, suggesting increased susceptibility to infection. In addition, some patients appear to have an accelerated rate of progression to cirrhosis caused by the combined effects of chronic viral infection and alcohol, although this association is much less pronounced than with chronic hepatic C virus (HCV) infection.

There is an increased prevalence of HCV infection in heavy drinkers ( Fig. 5.14 ). In many, but not all, of the clinical studies, the liver injury was more severe in drinkers who were HCV positive. In addition, chronic HCV infection accelerates the rate of progression to cirrhosis and increases the risk of hepatocellular carcinoma developing in patients with ALD.

Figure 5.14

Chronic hepatitis C and alcoholic liver disease. A heavy infiltrate of mononuclear cells is seen in a portal tract. There is also evidence of interface hepatitis and mild acinar hepatitis. The liver also shows moderate macrovesicular steatosis, and elsewhere there was mild alcoholic hepatitis. (H&E stain.)

The potentiation of acute drug-associated liver injury by alcohol is well recognized. The potential dangers of paracetamol (acetaminophen) ingestion, even in therapeutic doses, is now a well-recognized problem in chronic drinkers, but the idea of an alcohol–drug interaction causing chronic liver disease is a more recent concept. Leo et al. observed hepatic fibrosis in the rat after long-term administration of ethanol and a moderate dose of vitamin A. Hall et al. produced hepatic fibrosis and cirrhosis in rats by feeding ethanol in the Lieber–DeCarli diet together with exposure to low-dose carbon tetrachloride (CCl 4 ) vapour for 10 weeks. Relatively low doses of alcohol may well potentiate insidious liver injury by therapeutic doses of prescription drugs, nonprescription agents (e.g. vitamin A) and environmental toxins. This type of interaction may explain some cases of chronic liver injury seen in association with relatively low doses of alcohol and may account for some cases of ‘cryptogenic cirrhosis’.

Chronic alcohol consumption is a risk factor in methotrexate-associated fibrosis and cirrhosis in patients with psoriatic arthropathy. It is also thought to increase the risk of liver injury in rheumatoid arthritis patients receiving low doses of oral methotrexate.

Both primary (hereditary) and secondary iron overload can potentiate alcohol-related liver injury ( Fig. 5.15 ). Activation of hepatic stellate cells by iron and increased oxidative stress are factors involved in the accelerated progression to cirrhosis in patients with homozygous genetic haemochromatosis who are chronic drinkers.

Figure 5.15

Hereditary haemochromatosis and alcoholic liver disease. A, The liver shows the features of established micronodular cirrhosis, with moderate fatty change. Grade 4 siderosis with brown granules of iron are seen in the hepatocytes. Iron is also seen in bile duct epithelium and portal tact macrophages. B, Another area of the same liver. The iron is seen as blue granules, and the fibrous tissue is stained red. (Perls/Sirius red stain.)

Animal models

Animal models have been widely used to study the effects of alcohol on the liver, yielding a wealth of information on pathogenetic mechanisms and factors that potentiate or protect against alcohol-related liver injury. Despite the variety of animal species, the innumerable dietary manipulations and even the development of the intragastric tube-feeding model, none of the animals has ever developed an alcoholic hepatitis-like pattern of injury. Although some of the early animal studies used primates, most investigators now use rodent models. Mathews et al. summarize the most common approaches, including a chronic + binge ethanol model in which neutrophil infiltration occurs, but ballooning degeneration with MDB formation is not a feature. Wilkin et al. recently reviewed the place of transgenic models of alcohol-induced injury. Currently, one of the main values of animal models is in the identification of novel therapeutic targets and the evaluation of new treatments before initiation of clinical trials. More recently, several groups have explored the use of zebrafish models for high-throughput analysis of the molecular pathogenesis of ALD.

Clinical features

The clinical features of the various histological parts of the ALD spectrum are highly variable and range from entirely asymptomatic forms to an acute presentation in liver failure. Those with minimal symptoms are more likely to have earlier, potentially reversible injury. Presentation is often with symptoms and signs unrelated to the liver, such as nonspecific digestive symptoms or psychiatric illness. Physical examination may be normal but may show signs associated with alcohol misuse (plethora, tremulousness, aggressive behaviour). There may be signs of chronic liver disease, including spider naevi, palmar erythema and oedema. Other features may include Dupuytren contractures, parotid swelling and effects of alcohol-related injury on other organs, such as hypertension and atrial fibrillation. Clinical signs and history cannot be relied on to predict the histological features and stage of the disease. Guidelines for the investigation and treatment of suspected ALD have been published by the American Association for the Study of Liver Diseases. Katoonizadeh et al. have drawn attention to the clinical features of acute-on-chronic alcoholic liver failure, suggesting that these patients have a poorer prognosis than those with decompensated cirrhosis; biopsy features are helpful in making this distinction, in particular the presence of ductular bilirubinostasis.

Pathology of alcoholic liver disease

Alcoholic steatosis

Steatosis, the earliest and the most common manifestation of alcoholic liver injury, is seen in up to 90% of patients presenting for treatment of chronic alcoholism. The spectrum of clinical manifestations ranges from asymptomatic hepatomegaly, through nonspecific digestive symptoms, to hepatic failure. Sudden death may occur in alcoholic patients and has been attributed to alcohol withdrawal. At autopsy, however, the liver may show severe fatty change, suggesting that these patients may have died from acute liver failure caused by severe microvesicular steatosis; the possible role of other metabolic complications (e.g. hypoglycaemia) cannot be excluded.

Fatty change occurs predominantly in the perivenular zone and is seen initially in hepatocytes adjacent to the terminal hepatic venule. As the liver injury progresses, fatty change can be seen in hepatocytes in all zones. Fat disappears from the hepatocytes in 2–4 weeks following abstinence from alcohol but may persist in portal tract macrophages.

Initially, fat droplets appear to be membrane bound, presumably by endoplasmic reticulum. As the droplets become larger, they fuse, forming non-membrane-bound droplets. Fat droplets are seen as clear, intracytoplasmic vacuoles in haematoxylin and eosin (H&E)-stained sections of processed tissue ( Fig. 5.16 ). To retain fat in tissue specimens, formalin-fixed liver material can be post-fixed in osmium tetroxide; the osmicated fat is seen as black droplets in unstained sections. H&E and ‘routine’ special stains can still be performed ( Fig. 5.16 B ). The amount of fat in sections of osmicated tissue can be readily quantified (e.g. by digital image analysis). As the fat droplets enlarge, they are seen predominantly as single, large vacuoles that displace the nuclei of the hepatocytes.

Figure 5.16

Macrovesicular fatty change. A, Single, large fat droplets are seen in most of the hepatocytes; the liver cell nuclei are situated peripherally. B, Osmicated hepatic tissue. The osmicated fat is seen as black droplets of varying size. (H&E stain.)

Hepatocyte necrosis and inflammation are not usually seen at the steatosis stage, other than in association with lipogranuloma formation. Rupture of distended hepatocytes leads to the release of fat and an inflammatory response comprising lymphocytes, macrophages and occasionally eosinophils ( Fig. 5.17 ). Rarely, true epithelioid granulomas are seen, and serial sectioning may be required to demonstrate the presence of fat droplets. Lipogranulomas predominantly occur in the region of the terminal hepatic venules and are seen most frequently in severe fatty change. Small amounts of fibrous tissue may be present. However, lipogranulomas usually disappear without sequelae.

Figure 5.17

Lipogranuloma. Mildly fatty liver with a lipogranuloma in the region of a terminal hepatic venule. (H&E stain.)

Alcoholic foamy degeneration

The term ‘alcoholic foamy degeneration’ is used to describe alcohol-induced microvesicular fatty change occurring in the absence of alcoholic hepatitis. Uchida et al. described 20 patients, all of whom recovered rapidly once alcohol was withdrawn. They considered this to be a purely degenerative process since it occurred in the absence of inflammation. The clinical and biochemical features are highly suggestive of extrahepatic biliary obstruction. The liver shows striking microvesicular fatty change that is maximal perivenularly but may extend into this midzone ( Fig. 5.18 ). Some macrovesicular fat droplets may also be present. Bile is frequently seen in perivenular hepatocytes and canaliculi, whereas Mallory–Denk bodies (MDBs) and an infiltrate of neutrophil polymorphs are usually minimal or absent. Perivenular fibrosis is usually present, as is a small amount of perisinusoidal fibrosis. Electron microscopy showed widespread damage or loss of organelles, particularly mitochondria and endoplasmic reticulum.

Figure 5.18

Microvesicular fatty change. Most of the hepatocytes are distended by large numbers of small fat droplets that surround centrally situated nuclei. Several cells contain slightly enlarged mitochondria, which are stained pink. (Chromotropic aniline blue [CAB] stain.)

Alcoholic hepatitis

Beckett et al. used the term ‘acute alcoholic hepatitis’ to describe a clinicopathological syndrome. Many subsequent studies have shown that a wide variety of clinical features and biochemical abnormalities may accompany the same morphological pattern of liver injury. The term ‘alcoholic steatonecrosis’, used by some authors, is synonymous with alcoholic hepatitis.

Alcoholic hepatitis can only be reliably diagnosed morphologically. A liver biopsy study of patients presenting for treatment of alcoholism revealed alcoholic hepatitis in 17%. Alcoholic hepatitis may be asymptomatic, but it is usually associated with nonspecific digestive symptoms, hepatomegaly and increased liver enzyme levels. About 25% of patients with severe liver injury show evidence of liver failure or hepatic encephalopathy. Since alcoholic hepatitis may be asymptomatic, and up to 39% of patients have established cirrhosis at first presentation, the true incidence of alcoholic hepatitis remains to be determined. Histological alcoholic hepatitis is said to occur less often in Japan than in other parts of the world.

The liver injury is characterized by steatosis, liver cell necrosis and a neutrophil polymorph-rich infiltrate. MDBs are usually seen, but their presence is not needed for the diagnosis. Unlike fat droplets, MDBs persist in liver cells for many months after alcohol consumption ceases. Giant mitochondria may be seen on light microscopy as eosinophilic, globular inclusions ( Fig. 5.19 ) but occasionally may be needle-shaped cytoplasmic inclusions. Where the liver injury is very mild, the presence of giant mitochondria may be a diagnostic clue of chronic alcohol consumption.

Figure 5.19

Giant mitochondria (megamitochondria) within perivenular hepatocytes (arrows) in a patient with a history of excess alcohol intake. Note that the shape of giant mitochondria is variable; some can be needle shaped.

When mild, only occasional foci of liver-cell necrosis are seen in the perivenular regions accompanied by a slight neutrophil polymorph infiltrate; a few enlarged hepatocytes may contain MDBs, and minimal pericellular fibrosis may also be present. Occasionally, however, liver biopsies from patients with a history of moderate to heavy alcohol consumption show a nonspecific pattern of injury characterized by mild to moderate hepatocyte injury—necrosis and/or apoptosis—with minimal or absent ballooning degeneration, and accompanied by a mononuclear cell infiltrate with few or no polymorphs.

In fully developed alcoholic hepatitis, hepatocyte necrosis is more widespread and sometimes confluent. Hepatocyte enlargement is a prominent feature, and these ballooned hepatocytes frequently contain MDBs ( Fig. 5.20 ). The neutrophil polymorph infiltrate is often concentrated around hepatocytes containing MDBs (see Fig. 5.25 B ), so-called satellitosis. Monoclonal antibodies to K8 and K18, ubiquitin and p62 can be used to confirm the presence of the inclusions (see Figs 5.5 and 5.20 D). These will also detect small MDBs that are not readily apparent in routinely stained sections. A review of 700 articles on MDBs found a mean prevalence of 65% in alcoholic hepatitis and 51% in alcoholic cirrhosis.

Figure 5.20

Alcoholic hepatitis. A, The architecture is disturbed, and there is a considerable liver cell necrosis with an associated neutrophil polymorph infiltrate. Macrovesicular fat droplets are present in some hepatocytes. This 30-year-old woman had a 12-year history of heavy drinking. (H&E stain.) B, Numerous, slightly enlarged hepatocytes contain Mallory–Denk bodies (MDBs), and a heavy neutrophil polymorph infiltrate is seen in proximity to these cells. A marked degree of liver cell necrosis is apparent, and a few large fat droplets are seen. (H&E stain.) C, Ballooned hepatocytes show immunoreactivity for the ligand sonic hedgehog (Shh) that can initiate Hedgehog pathway signalling and promote fibrogenesis. D, MDBs. A monoclonal antibody to ubiquitin reacts strongly with MDBs.

Steatosis is usually present but is variable in severity; it may be absent in patients hospitalized for several weeks before the liver biopsy. Unlike in NASH, presence of steatosis is not a diagnostic requirement. Pericellular fibrosis is often present but may be limited to the perivenular zones in the early stages of ALD.

Other features may include lipogranulomas, apoptosis (acidophil bodies), induced hepatocytes with a ‘ground-glass’ appearance due to the proliferation of smooth endoplasmic reticulum, oncocytic hepatocytes, bile stasis, Kupffer cell proliferation and enlargement due to ingested lipid and ceroid, ‘microscopic cholangitis’ and a mild mononuclear cell infiltrate in the portal tracts. Features indicating liver regeneration include mitotic figures in hepatocytes, microregenerative nodules and a ductular reaction.

Sclerosing hyaline necrosis

Sclerosing hyaline necrosis, described by Edmondson et al., is not a separate pattern of alcohol-induced liver injury but rather forms part of the morphological spectrum of alcoholic hepatitis, characterized by extensive perivenular liver cell necrosis associated with the deposition of fibrous tissue ( Fig. 5.21 ). The terminal hepatic venules may become occluded, and portal hypertension can occur in the absence of cirrhosis. Initially, fibrosis is seen only in the perivenular region and in association with foci of necrosis (see Fig. 5.10 ).

Figure 5.21

Sclerosing hyaline necrosis. Mild fatty change, mild alcoholic hepatitis and several Mallory-Denk bodies are apparent in the region of a hepatic venule. Increased amounts of fibrous tissue are seen in the parenchyma adjacent to the venule. (H&E stain.)

With progression, more extensive pericellular fibrosis is seen. More severe bridging necrosis between adjacent terminal hepatic venules or between terminal hepatic venules and portal tracts results in condensation of pericellular fibrosis tissue and the formation of septa. Elastic fibres, which stain with orcein, can be seen in active fibrous septa, but not in areas of passive collapse of the reticulin framework.

Vascular and microvascular lesions

Goodman and Ishak reviewed 200 autopsy cases of ALD and described three types of venous lesions: lymphocytic phlebitis, phlebosclerosis (from perivenular fibrosis gradually obliterating vein lumens) and veno-occlusive lesions, characterized by intimal proliferation, fibrosis of the vein wall and varying degrees of luminal obliteration. Lymphocytic phlebitis was noted in 16% of patients with alcoholic hepatitis and in 4% of those with cirrhosis. Phlebosclerosis was found in all cases of alcoholic hepatitis and cirrhosis. Veno-occlusive lesions ( Fig. 5.22 ) were found in 52% of patients with alcoholic hepatitis and portal hypertension, totally occluded veins were found in 47%, and partial occlusion of varying severity was found in 74% of cirrhotic patients. Portal hypertension correlated significantly with degree of phlebosclerosis and veno-occlusive change. In biopsy material, Burt and MacSween confirmed that phlebosclerosis was a universal finding in alcoholic hepatitis and cirrhosis, but they found veno-occlusive lesions in only 10% of 256 biopsies and lymphocytic phlebitis in 4%. These occlusive venous lesions may contribute to the atrophy of hepatic parenchyma and functional impairment. Robles–Medranda et al. described a Budd–Chiari-like syndrome in a patient with advanced ALD.

Figure 5.22

Veno-occlusive lesion in alcoholic liver disease. Note the marked intimal proliferation producing considerable narrowing of the lumen of a hepatic vein branch. (Verhoeff–van Gieson stain.)

Schaffner and Popper first introduced the term ‘capillarization of hepatic sinusoids’. This phenomenon is now recognized as resulting from disruption of the normal ECM in the space of Disse and replacement by mainly type I collagen, but also basal lamina-like material containing laminin and type IV collagen, all of which are produced by activated hepatic stellate cells. This process occurs in early alcoholic liver injury and is seen first in the perivenular zone. Capillarization of the sinusoids results in a significant barrier between the blood and the hepatocyte, which may provoke further activation of stellate cells and may be a factor in hepatocyte dysfunction and injury, as well as reducing the transport of solutes from the sinusoidal blood to the hepatocyte via the space of Disse. This phenomenon is often accompanied by structural changes in the liver sinusoidal endothelial cells (see Chapter 1 ). The term ‘defenestration’ is used to describe changes in the fenestrae that lessen the porosity of the sinusoidal lining. Scanning electron microscopy (SEM) studies of needle biopsies from noncirrhotic alcoholic patients show evidence of defenestration in the perivenular zone, occurring in both the presence and the absence of collagenization of the space of Disse. Horn et al. demonstrated a positive correlation between defenestration and the occurrence and localization of subendothelial basal laminae, and between the presence of a basal lamina and the occurrence of collagen in the space of Disse, again suggesting that the phenomenon of defenestration is involved in the pathogenesis of hepatic fibrosis. The process of defenestration has been associated with increased vascular resistance in the sinusoidal bed, which suggests that alterations in the hepatic sinusoidal lining may be involved in the pathogenesis of portal hypertension. Fraser’s group made the interesting suggestion that defenestration of the hepatic sinusoids could be a cause of hyperlipoproteinaemia and could also contribute to hepatic steatosis.

Perivenular fibrosis

Baboons fed alcohol chronically have been observed to progress from the fatty liver stage to cirrhosis, without an intermediate stage of alcoholic hepatitis. Perivenular fibrosis was present in association with alcoholic steatosis. A similar lesion occurs in humans, and it is now accepted as an intermediate stage in the development of alcoholic cirrhosis and distinct from alcoholic hepatitis.

Perivenular fibrosis has been defined as fibrosis extending around at least two-thirds of the perimeter of the terminal hepatic venule, the fibrous rim measuring over 4 μm in thickness ( Fig. 5.23 ). Serial liver biopsy studies have shown that patients with perivenular fibrosis at the steatosis stage are likely to have progressive liver injury if drinking continues. In the study of Worner and Lieber, 13 of 15 patients with perivenular fibrosis in the first biopsy developed more severe disease within 4 years; nine had more extensive fibrosis, one incomplete cirrhosis and three established cirrhosis. Morphometric studies of ‘early alcoholic liver disease’ failed to demonstrate fibrous thickening of the walls of the terminal hepatic venules. Another study confirmed the presence of perivenular fibrosis as an early lesion in moderate to heavy drinkers ; 38.9% of males with a daily intake of 40–80 g alcohol for an average of 25 years had perivenular fibrosis, and 44.4% had perivenular fibronectin deposition. Since only 20% of the heavy drinkers had cirrhosis, the authors suggested that factors in addition to the dose and duration of alcohol consumption must contribute to the progression from early perivenular fibrosis to cirrhosis.

Figure 5.23

Perivenular fibrosis. Alcoholic fatty liver without evidence of alcoholic hepatitis. An abnormally thick rim of fibrous tissue surrounds the terminal hepatic venule; fibrous tissue also extends focally into the surrounding liver in a pericellular pattern. Several hepatocytes contain large mitochondria (discrete, pink globular masses). (CAB stain.)

Ultrastructural studies have shown myofibroblast proliferation around the terminal hepatic venule, occurring in association with perivenular fibrosis ( Fig. 5.24 ). Perivenular fibrosis is thought to be the first lesion in a sequence of events which leads ultimately to the development of cirrhosis. The recognition of perivenular fibrosis at the steatosis stage may permit the identification of patients who are likely to have progressive liver injury if they continue to drink.

Figure 5.24

An activated myofibroblast in the space of Disse. Prominent rough endoplasmic reticulum is seen in the cell, but fat droplets are absent. Bundles of collagen are seen in the space of Disse closely apposed to the outer surface of the cell. (×12,700.)

Alcoholic cirrhosis

A World Health Organization (WHO) group defined cirrhosis as a diffuse process characterized by fibrosis and the conversion of the normal liver architecture into structurally abnormal nodules. Micronodular cirrhosis, which is the most common type of cirrhosis seen in association with alcohol, is characterized by remarkably uniform-sized regenerative nodules, most of which are <3 mm in diameter. Bands of fibrous tissue completely surround the regenerative nodules; the terminal hepatic venules are not recognizable, but new vessel formation is apparent within the fibrous tissue ( Fig. 5.25 A ).

Figure 5.25

Alcoholic cirrhosis. A, ‘Inactive cirrhosis’. There is complete loss of the normal architecture and replacement by small regenerative nodules that are completely surrounded by broad bands of fibrous tissue. (Van Gieson stain.) B, Steatosis superimposed on established micronodular cirrhosis (H&E stain).

Biopsies of cirrhotic livers may also show features of alcoholic hepatitis; the hepatocyte injury occurs predominantly at the periphery of the regenerative nodules. A variable mixture of neutrophil polymorphs, lymphocytes, plasma cells and macrophages is seen in the fibrous tissue. The presence of alcoholic hepatitis usually means continued drinking, even though there may be little or no fatty change. Autopsies sometimes reveal an unexpected finding of ‘inactive cirrhosis’ in patients who have been drinkers in the past ( Fig. 5.25 A ), but if drinking continues up until the time of death, steatosis, with or without the features of alcoholic hepatitis, may be seen ( Fig. 5.25 B ).

In the past a more portal-based chronic hepatitis has been described in alcoholic patients because of the presence of interface hepatitis, but it now appears that superimposed HCV (and to a lesser extent HBV) in fection explains many of the reported cases. A ductular reaction is frequently seen in ALD; special stains and IHC studies have demonstrated glycogen, α1-antitrypsin (α1-AT) and glucose-6-phosphatase in the ductular cells. IHC studies using a variety of monoclonal and polyclonal antibodies to keratins have demonstrated the presence of the bile duct keratins K7 and K19 in the ductular cells ( Fig. 5.26 ). The ‘proliferating’ ductules seen in the portal tracts and periportal regions, in alcoholic cirrhosis as well as cirrhosis due to other aetiologies, were previously regarded as regenerating hepatocytes showing ductal metaplasia. An alternative explanation is that these proliferating bile ductular cells are the progeny of ‘proliferating’ hepatic stem cells (see Chapter 1 ). Small epithelial cells and ‘proliferating’ bile ductules often become particularly prominent in cirrhotic livers after abstinence from alcohol. IHC studies of a wide range of human liver disease, including ALD, have provided convincing evidence in support of the concept that these ‘reactive’ ductules are the progeny of hepatic, and possibly bone marrow, stem cells. Jung et al. have demonstrated a key role for Hedgehog (Hh) signalling pathways in the generation of the ductular reaction in ALD.

Figure 5.26

Proliferation of ductules and small epithelial cells in alcoholic liver disease cirrhosis. Explanted liver shows numerous small epithelial cells and ductules in the portal tracts and at the periphery of regenerative nodules. (Immunostain for biliary epithelial cell keratins 7 and 19.)

Cirrhosis may become macronodular, particularly if drinking ceases. The regenerative nodules then vary greatly in size, measuring up to several centimetres in diameter; many such nodules contain portal tracts and terminal hepatic venules that are abnormally related to each other ( Fig. 5.27 ). In some cases the architectural changes are best described as ‘incomplete septal cirrhosis’. A mixed micronodular and macronodular cirrhosis, with a variable proportion of micro- and macronodules, may be found at autopsy or at transplantation.

Figure 5.27

Macronodular cirrhosis. A large macroregenerative nodule contains several rudimentary portal areas. (Sirius red stain.)

Large regenerative nodules may occasionally have clinical and radiological features suggestive of hepatocellular carcinoma. Nagasue et al. reported several such cases and proposed the term ‘hepatocellular pseudotumour’ in the cirrhotic liver to describe these nodules. Focal nodular hyperplasia-like lesions are also described.

Steatosis and alcoholic hepatitis are reliably diagnosed by needle biopsy of the liver, but cirrhosis may be underdiagnosed because of sampling difficulty, particularly when an aspiration-type needle is used. Macronodular cirrhosis may be suspected but cannot be confidently diagnosed by needle biopsy because of the large size, by definition >3 mm in diameter, of the nodules.

Globules of α1-AT are frequently seen in hepatocytes at the periphery of regenerative nodules; this accumulation is considered to be a consequence of impaired protein secretion. However, studies in liver transplant units have revealed a higher-than-predicted prevalence of heterozygote α1-AT deficiency in patients undergoing transplantation for a variety of diseases, including alcoholic cirrhosis. This observation suggests that α1-AT deficiency may potentiate other forms of liver disease, including ALD. On rare occasions, ALD and homozygous α1-AT deficiency may coexist ( Fig. 5.28 ). Again, a possible interaction between the two aetiological factors may lead to cirrhosis at a relatively early age. Copper may also accumulate in hepatocytes in ALD; this is most likely a secondary cholestatic phenomenon. Copper bound to protein is seen as periodic acid-Schiff (PAS)-positive, diastase-resistant globules that stain positively with orcein and other, more specific stains for copper such as rhodanine.

Figure 5.28

Alcoholic steatohepatitis and coexisting α1-antitrypsin deficiency. Many of the hepatocytes contain diastase-resistant periodic-acid Schiff (PAS)-positive globules. In addition, there is fatty change, hepatocyte necrosis and inflammation.

Oxyphilic granular hepatocytes, which have been termed ‘hepatic oncocytes’, are frequently seen at the periphery of regenerative nodules in alcoholic cirrhosis. Oncocytes and induced hepatocytes can be differentiated from classic ground-glass hepatocytes which contain hepatitis B surface antigen (HBsAg) by orcein staining or immunohistochemistry.

Other morphological features of alcoholic liver injury

Cholestatic syndromes

Alcoholic liver disease may present with clinical and biochemical features that are strongly suggestive of extrahepatic biliary obstruction. Features include jaundice, right upper quadrant pain and tenderness, hepatomegaly and marked elevation of serum bilirubin, and alkaline phosphatase and cholesterol levels. Biliary obstruction caused by gallstones or alcoholic pancreatitis can be excluded by ultrasound, transhepatic cholangiography, endoscopic retrograde cholangiopancreatography (ERCP) or magnetic resonance cholangio­pancreatography (MRCP).

A high index of clinical suspicion of intrahepatic cholestasis is required to avoid unnecessary surgery in patients with ALD, because the risk of postoperative hepatic and renal failure is considerable. Severe cholestasis has been described in association with: fatty liver ( Fig. 5.29 ), alcoholic foamy degeneration, alcoholic hepatitis, decompensated alcoholic cirrhosis and Zieve syndrome, which comprises alcoholic steatosis, jaundice, hyperlipidaemia and haemolytic anaemia.

Figure 5.29

Fatty liver and cholestasis. The hepatocytes show fatty change, and bile is present in some canaliculi. (H&E stain.)

There is a significant correlation between tissue cholestasis and mortality in ALD. The possibility that jaundice occurring in patients with a history of excess alcohol consumption may be caused by nonalcoholic liver disease, such as viral hepatitis or drug-induced cholestasis, must also be considered.

In patients with a severe fatty liver who develop a cholestatic syndrome, the liver biopsy may show portal tract changes with oedema, increased prominence of marginal bile ducts and a mild to moderate cholangitis with a neutrophil polymorph infiltrate, so-called microscopic cholangitis.

Portal tract changes

Portal fibrosis has not been considered to result from classic alcohol injury. However, it may sometimes be a feature; Morgan et al. found that the presence of portal fibrosis correlated with a previous history of viral hepatitis or episodes of acute pancreatitis. Michalak et al. suggested that portal fibrosis is part of progressive ALD. In their experience, portal and septal fibrosis was more frequent than perivenular fibrosis. Furthermore, portal fibrosis appeared to contribute to a greater proportion of overall fibrosis than did perivenular scarring. These intriguing findings merit further investigation.

Increased numbers of portal tract macrophages are seen frequently in ALD at virtually all stages. Also, macrophages greatly enhanced lysosomal enzyme release compared with normal liver, suggesting that cytotoxic mediators released by these activated cells may be contributing to liver injury. Rakha et al. described ‘more than mild’ portal inflammation in 47% of biopsies from patients with alcoholic fatty liver disease.

Hepatic siderosis

Excess stainable iron is found in both hepatocytes and Kupffer cells in many patients with ALD. In one study, 57% of patients had mild siderosis and 7% had grade 3–4 siderosis. The explanation for the excessive iron accumulation is still unclear but suggests increased iron absorption, attributed to a variety of factors: the high iron content of some alcoholic beverages, in particular home-brewed beer, which in sub-Saharan Africa is made in iron pots ; a direct effect of alcohol on the small intestine ; haemolysis associated with spur cell leading to increased iron absorption ; and upregulation of the transferrin receptor. It is now generally accepted that only mild siderosis, grade 1–2, is alcohol related, so-called alcoholic siderosis. Patients with ALD in whom iron accumulation is massive (grade 3–4) may also have hereditary haemochromatosis (see Fig. 5.15 ). Nevertheless, iron overload can be seen in extrahepatic tissues of patients with end-stage ALD in the absence of HFE mutations.

In alcoholic siderosis, the iron-containing hepatocytes are distributed randomly, and the iron granules are often few in number. This contrasts with the zonal distribution of iron, with a periportal predominance, and often a pericanalicular distribution in the hepatocytes in hereditary haemochromatosis.

Before the identification of genetic abnormalities involved in iron metabolism, alcoholic siderosis and hereditary haemochromatosis were differentiated by biochemical measurement of the hepatic iron concentration and calculation of the hepatic iron index (HII; hepatic iron concentration divided by patient’s age). An HII >2 is virtually diagnostic of hereditary haemochromatosis, but calculation of the HII has been largely replaced by genotyping. If genotyping is not available, however, and when marked iron overload is an unexpected finding in a liver biopsy specimen, the hepatic iron, as demonstrated by the Perls method, can be measured by image analysis, or the remainder of the liver tissue can be removed from the paraffin block and the iron concentration measured biochemically in the dewaxed specimen.

The liver of chronic drinkers who are homozygous or heterozygous for the HFE (C282Y) mutation, or one of the less common mutations involved in iron overload, may show evidence of both hepatic iron overload and alcohol-related liver injury. A biopsy study of 206 patients with HFE -associated haemochromatosis showed that cirrhosis was almost nine times as common in those who drank >60 g of alcohol daily than in those who drank <60 g. In such patients, episodes of alcoholic hepatitis, in which iron is released from dead hepatocytes, can lead to an altered distribution of liver iron, with haemosiderin appearing in Kupffer cells and portal tract macrophages.

Both primary and secondary hepatic iron overload appear to potentiate liver injury in humans. Similarly, dietary iron overload potentiated alcohol/hepatotoxin-mediated fibrosis and cirrhosis in an animal model. Possible mechanisms for this potentiation of hepatic fibrosis included increased amounts of lipid peroxidation, iron-induced activation of hepatic stellate cells and iron-dependent activation of nuclear factor κB (NFκB) in Kupffer cells, leading to the production of proinflammatory cytokines. The effects of alcohol on iron storage diseases of the liver have been reviewed.

Porphyria cutanea tarda

Alcohol is thought to hasten the onset of the hepatic and cutaneous manifestations of porphyria cutanea tarda; alcohol withdrawal is followed by a dramatic clinical and biochemical improvement. The hepatocytes may contain needle-shaped cytoplasmic inclusions of uroporphyrin which show brilliant-red autofluorescence under ultraviolet light; this is specific for porphyria cutanea tarda. Variable degrees of siderosis are usually present in the liver, and there may also be evidence of ALD.

Hepatocellular carcinoma

Hepatocellular carcinoma (HCC) develops in 5–15% of patients with alcoholic cirrhosis, usually in association with macronodular cirrhosis ( Fig. 5.30 ). In Western countries, alcohol may be the most common cause of HCC, accounting for 32–45% of cases. Tumours are often seen in association with a combination of alcoholic and viral liver disease (HBV and HCV related) but are also frequently described in patients with negative markers for viral hepatitis. Some evidence now indicates that alcohol may cause genetic alterations. In addition, alcohol can act as a co-carcinogen because of its ability to induce the hepatic microsomal P-450-dependent biotransformation system, in particular the ethanol-inducible isoenzyme cytochrome P-450, leading to enhanced activation of a variety of procarcinogens present in food, tobacco smoke and alcoholic beverages. A study of American patients with cirrhosis, with and without HCC, showed that alcohol, tobacco and obesity were independent, synergistic risk factors for HCC development.

Figure 5.30

Hepatocellular carcinoma arising in a liver showing macronodular cirrhosis. This 58-year-old man had a past history of heavy drinking but had abstained for several years. (H&E stain.)

Alcohol is recognized as an important potentiating factor for HCV-associated HCC. Yamauchi et al. reported a 10-year cumulative HCC occurrence rate of 80.7% in patients with alcoholic cirrhosis who were anti-HCV positive and drank >120 g alcohol daily. In contrast, 18.5% of those with alcoholic cirrhosis alone developed HCC over 10 years, and the 10-year rate for tumours in HCV-related cirrhosis in nondrinkers was 56.5%. One review of alcohol and HCC reported that chronic alcohol consumption >80 g/day for >10 years increases the HCC risk almost fivefold, and that alcohol doubles the risk of this tumour in CHC compared with the risk with the virus alone.

Autopsy studies in the past, as well as recent prospective studies of liver biopsies, have suggested that dysplastic nodules, which develop within cirrhotic nodules, especially macroregenerative nodules (>5 mm diameter), may be the precursor lesion for HCC. A prospective liver biopsy study of radiologically identified hepatic nodules concluded that an increased ratio of nuclear density >1.5, clear-cell change and small-cell change are features indicating a high risk for evolution to HCC.

Lee et al. evaluated large-cell change, seen as foci of hepatocellular enlargement, nuclear pleomorphism, hyperchromasia and multinucleation, and concluded that this feature is an independent risk factor for HCC, with an estimated odds ratio of 3.3. However, large-cell change is not thought to be a direct precursor of HCC, but rather an indicator of increased risk of tumour development. A comment about the presence or absence of both small-cell and large-cell change should always be included in reports of cirrhotic livers.

Recent epidemiological studies have demonstrated an increased prevalence and incidence of intrahepatic cholangiocarcinoma and a relationship to the presence of cirrhosis. Interestingly, Wu et al. described biliary intraepithelial neoplasia lesions in the livers of patients with alcoholic cirrhosis, suggesting that these may be precursor lesions for the development of biliary malignancy.

Diagnosis and prognosis of ALD: role of liver biopsy

Historically, liver biopsy was seen as an important diagnostic tool in the management of patients with alcohol-related liver disease. With emerging methodologies for noninvasive measurement of fibrosis by surrogate markers and elastography, as well as the inherent invasive nature of the biopsy procedure, its use in practice has declined. Indeed, the most recent guidelines from the European Association for the Study of the Liver (EASL) and American Association for the Study of Liver Diseases (AASLD) differ in their recommendation for the role of biopsy in the management of patients with alcoholic hepatitis. Proponents of its use point to the dual or multiple pathologies or unsuspected diagnoses; in a study by Levin et al., only 80% of patients with a heavy alcohol intake were found to have alcohol-associated liver injury. The other 20% had various types of non-alcohol-related liver disease, including cholangitis, viral hepatitis, granulomatous hepatitis, passive venous congestion and nonspecific changes. However, other studies have shown that a clinical diagnosis of alcohol-related liver injury was significantly associated with a histological diagnosis of ALD, with a 98% specificity and a 79% sensitivity. Nevertheless, a liver biopsy may be justified in patients with suspected ALD but in whom the clinical evidence for alcohol misuse may be in doubt. Biopsy may also be used to assess the patient for possible therapeutic intervention and to monitor the effects of various treatment modalities. Furthermore, biopsy may be part of the workup of some patients being considered for orthotopic liver transplantation (OLT), particularly if cofactors are suspected (e.g. iron overload, which may require treatment before OLT). Biopsy can be used to confirm or refute alcohol abstinence, although as noted earlier, some histological features (e.g., MDBs) can be long-lasting.

A biopsy can also be of value in establishing the stage and grade of the disease; this may contribute to an assessment of prognosis. It has been demonstrated, for example, that in patients with hepatic decompensation, transjugular biopsies can be used to identify a subset of patients with superimposed alcoholic hepatitis who may benefit from a trial of therapeutic intervention.

A number of demographic, clinical and laboratory findings can be used to predict outcome in patients with ALD. These include persistent alcohol misuse, female gender, tobacco use, concomitant obesity and type 2 diabetes or viral hepatitis. Clinical scoring systems have been developed to predict mortality in acute alcoholic hepatitis.

The histopathologist is frequently asked to assess prognosis and the likelihood of the liver injury being reversible on the basis of the liver biopsy findings. Semiquantitative scoring systems have been described for assessing histological severity in NASH (see later); these are yet to be fully validated for ALD. Alcoholic steatosis without evidence of hepatocyte necrosis or fibrosis was once thought to be a ‘benign’ condition with little risk of progression to cirrhosis. This was questioned by Teli et al., who followed 88 patients with alcoholic steatosis for a mean of 10.5 years. Seven developed fibrosis, and another nine became cirrhotic; however, eight of the nine were known to have continued to drink heavily. A mixed macro- and microvesicular pattern and giant mitochondria in the original biopsies were found to be independent histological predictors of progression. Continued alcohol consumption by patients with these histological features on biopsy is associated with a 47–61% risk of progressive liver disease. Dam–Larsen et al. reported severity of steatosis on liver biopsy in alcoholic fatty liver disease to be a predictor of progression. In a more recent study from Denmark of patients with histologically proven alcoholic steatosis or steatohepatitis, those with fatty liver disease had an increased risk of developing cirrhosis and liver-related mortality compared to a matched reference population; the risk was double when steatohepatitis was present.

Histological features that indicate a risk of progression to cirrhosis include the following:

  • 1.

    Severity of the fatty change, particularly if it is microvesicular, and the extent of the hepatic fibrosis in livers seen at the fatty liver stage

  • 2.

    Severity of hepatocyte necrosis and extent of pericellular and perivenular fibrosis ; the formation of fibrous septa with elastic fibre deposition and architectural disturbance are poor prognostic indicators.

  • 3.

    Diffuse parenchymal disease, which can also have an acute fulminant course

  • 4.

    Widespread obliteration of hepatic venules

  • 5.

    Widespread Mallory–Denk body formation

  • 6.

    Presence of marked intraparenchymal cholestasis

A validated alcoholic hepatitis scoring system has been described in which the degree of polymorph infiltrate, fibrosis stage, cholestasis and megamitochondria predicted 90-day mortality.

Harinasuta and Zimmerman and Christoffersen et al. found that in noncirrhotic livers, parenchymal fibrosis appeared to be more marked when MDBs were present. The 30-day mortality for patients with severe alcoholic hepatitis exceeds 40%. Clinical scoring systems have been developed to predict mortality in acute alcoholic hepatitis. Patients with alcoholic hepatitis are likely to have progressive injury leading to cirrhosis ; in one study, 50% of patients who continued drinking developed cirrhosis in 10–13 years. Pares et al. documented three variables that independently increase the risk of progression to cirrhosis: severity of the initial histological injury, continuation of drinking and female gender. Morgan concluded that survival is significantly reduced in women, in elderly persons and with continued drinking.

Chedid et al. followed 281 alcoholic patients prospectively for 4 years to assess prognosis. The worst prognosis, 35% survival at 48 months, was seen in patients with alcoholic hepatitis superimposed on cirrhosis. Mathurin et al. showed that steatosis, fibrosis stage and presence of steatohepatitis were independent predictive factors of fibrosis progression. Jepsen et al. demonstrated that the 5-year mortality rates for patients with alcoholic cirrhosis may be as high as 85%; many have overt clinical complications at diagnosis (although again, some may be asymptomatic). In a retrospective study of 192 alcoholic patients, fibrosis stage was the main predictor of long-term survival in patients with early/compensated ALD. In decompensated ALD, gender, biochemical markers of liver failure and extent of pericellular fibrosis predicted long-term survival.

Numerous studies have shown that abstinence can prolong survival in alcoholic cirrhosis. Traditionally, cirrhosis has been thought to be an irreversible process, but as discussed in Chapter 1 , this concept has now been challenged. Cirrhosis in animal models is reversible, provided that the inducing aetiological agent is removed and sufficient time allowed for the liver to return to its normal structure. There are reports of reversal of established cirrhosis in humans. This is a rare occurrence, and reported cases are open to some doubt, particularly since micronodular cirrhosis can become macronodular after the removal of the aetiological agent, in this case long-term abstinence from alcohol, and a follow-up needle biopsy (taken from within a macroregenerative nodule) may fail to detect the presence of cirrhosis.

Treatment of ALD and histopathological features of resolution

The ideal goal should be prevention, rather than finding better ways to treat ALD. The recognition of inherited and acquired factors that increase the risk of development of alcohol-related liver injury has the potential to reduce considerably the prevalence of ALD, provided that the patients with risk factors (e.g. methotrexate, chronic HCV infection) can be persuaded to abstain from alcohol.

Abstinence from alcohol

Abstinence from alcohol remains the best way of preventing progression of liver injury as well as prolonging survival in patients with established cirrhosis. More vigorous counselling about the need for abstinence, including the risk of progressive liver disease should drinking continue, could significantly reduce the prevalence of ALD.

Patients who are required to abstain from alcohol, usually for at least 6 months, as a condition of eligibility for liver transplantation may have sufficient return of function, due to cessation of alcohol-induced liver injury and liver regeneration, that transplantation can be delayed or avoided altogether. Examination of explant tissue from patients undergoing transplantation for alcoholic cirrhosis affords an opportunity to observe the changes that occur after abstinence. The architectural changes include possible remodelling of the cirrhosis. In addition, we have noted frequent cytological changes in hepatocytes, with enlargement and a prominent ‘glassy’ eosinophilic cytoplasm ( Fig. 5.31 ).

Figure 5.31

End-stage alcoholic liver disease; explant tissue. Hepatocytes in a cirrhotic nodule show a glassy cytoplasm with accentuated cell boundaries, but there is no current steatohepatitis. (H&E stain.)

Medical management

The treatment of alcohol-related liver disease, in particular alcoholic hepatitis, is controversial. Corticosteroids are used for carefully selected patients with severe alcoholic hepatitis. When tests confirm the absence of coexisting infection, corticosteroids modulate the immunological disturbances and downregulate cytokines, such as TNFα and IL-8 production. Pentoxifylline is also thought to inhibit the synthesis and activity of TNFα and has been shown to be beneficial. However, in a recent large, multicentre randomized trial, pentoxifylline did not improve survival in patients with alcoholic hepatitis, in contrast to prednisolone, which reduced 28-day mortality but did not affect medium- or long-term mortality. S- adenosyl- l -methionine , a glutathione precursor, can restore mitochondrial glutathione and significantly lessen hepatocyte injury. Propylthiouracil can suppress the ‘hypermetabolic state’, as measured by increased liver oxygen consumption. Polyenylphosphatidylcholine , a mixture of polyunsaturated fatty acids, is another agent that decreases alcohol-induced oxidative stress. In baboons, its use has been shown to lessen hepatic fibrosis, presumably by decreasing the amount of breakdown products of lipid peroxidation, although this has yet to be shown as beneficial in humans. In recent years, alcoholic hepatitis has also been treated with biologics, in particular the anti-TNFα monoclonal antibody infliximab, but the results of two large, randomized controlled trials were disappointing. Other agents proposed in ALD treatment include colchicine, anabolic steroids, TLR4 antagonists, IL-1 receptor antagonists (anakinra), amlodipine and several complementary and alternative medicines, as reviewed by O’Shea and McCullough. Ongoing clinical trials of targeted treatment include metadoxin, emricasan, rifaximin, probiotics and zinc.

The contribution of dietary deficiencies to the pathogenesis of ALD has been highlighted (see earlier discussion). Most studies of dietary modification to enhance or lessen alcohol-induced liver injury have been in animal models; it is interesting to speculate about the possibility of dietary modification as a means of lessening the risk of ALD in humans. The dietary extremes used in models, however, are unlikely to be palatable to humans and are therefore unlikely to provide protection from alcohol-related liver injury. For a detailed discussion of nutrition and ALD, see the review by Halsted.

Liver transplantation

Although in the past the role of liver transplantation (LT) for ALD has been controversial, most current transplant units, at least in developed countries, perform LT on carefully selected patients. The controversy related to LT for ALD has abated largely because of the growing acceptance of alcohol as a cofactor or ‘permissive agent’ in a wide variety of inherited and acquired forms of liver disease. As discussed earlier, excess alcohol consumption can hasten the progression of other liver diseases (e.g. chronic HBV/HCV infections, hereditary haemochromatosis), with or without overt features of alcohol-related injury in the liver. Thus the belief that ALD is entirely self-inflicted and LT cannot be justified has been modified in recent years. In most centres, LT for patients with end-stage ALD is restricted to those who are cirrhotic and showing signs of decompensation but with evidence of recent abstinence. Several centres have advocated LT for the treatment of acute alcoholic hepatitis, but this has not been widely adopted.

The outcome of LT in ALD is similar to that in other forms of end-stage liver disease. Much emphasis has been placed on the selection of patients for transplant on the basis of likelihood of post-transplant sobriety, and a variety of prognostic models have been developed. However, in one study of 78 critically ill patients who had OLT for ALD, neither the preoperative length of sobriety nor alcohol rehabilitation predicted survival. Although some liver transplant units regard the consumption of any alcohol as a relapse, others differentiate harmful drinking from other forms of drinking.

Anecdotal reports of recurrent alcoholic liver disease abound, but there are few published reports. Hepatic steatosis is the feature seen most frequently in the transplanted liver after resumption of drinking ( Fig. 5.32 ). Two of the follow-up studies describe steatosis, without evidence of alcoholic hepatitis or cirrhosis, in post-transplant liver biopsies from patients consuming alcohol. Burra et al. also describe fatty change, but with pericellular fibrosis, and suggest that these two features are ‘the most relevant histological signs of heavy alcohol intake’ after LT. In contrast, Conjeevaram et al. reported severe alcoholic hepatitis in 50% of recidivists; a 45-year-old woman developed cirrhosis within 21 months of LT and was subsequently retransplanted.

Figure 5.32

Post-transplant liver biopsy from a 45-year-old man with presumed ‘cryptogenic’ cirrhosis who started drinking again after the transplant. The liver shows moderate macrovesicular fatty change without evidence of hepatitis or fibrosis. (H&E stain.)

Nonalcoholic fatty liver disease

Nonalcoholic fatty liver disease (NAFLD) is the most frequent cause of persistently elevated liver enzymes in North America and Europe. It is generally accepted to be an increasingly common cause of end-stage liver disease and also an independent risk factor for cardiovascular disease and stroke. As its name suggests, NAFLD develops in the absence of excessive alcohol consumption. To discriminate NAFLD from alcohol-related liver injury, an alcohol consumption threshold of 20 g/day for women and 30 g/day for men is adopted. Although a range of causative agents have been implicated (see Tables 5.1 and 5.2 ), ‘classic’ NAFLD is generally recognized to represent the hepatic manifestation of the metabolic syndrome, characterized by obesity, insulin resistance or type 2 diabetes, dyslipidaemia and hypertension.

Reflecting its growing importance and increasing consumption of health care resources, there has been an exponential growth in the study of NAFLD epidemiology, clinical associations, pathophysiology and therapy. Microscopic examination remains central for both diagnostic and research purposes. Biopsies with ‘simple’ steatosis (or steatosis and inflammation only) are diagnosed as nonalcoholic fatty liver (NAFL) and those with steatohepatitis as nonalcoholic steatohepatitis (NASH). The increasing incidence of obesity, calculated to approach 40% of the U.S. population by the year 2030, and the attendant rise in insulin resistance and type 2 diabetes in up to 10% of obese patients, are concerns for significant health and economic impacts of these disorders.

Historical perspectives

An association of fatty liver with obesity was first made by Bartholow in 1885. Pepper was first to observe fatty liver in a diabetic patient in 1884, and in 1938, Charles Connor described the histopathological features of fatty liver disease in diabetic patients and its association with cirrhosis development, highlighting an aetiological link. NAFLD was recognized by pathologists decades ago as a distinct, potentially progressive liver disease with histological similarities to alcohol-related liver disease in obese patients. In the 1950s, studies documented possible links of fatty liver with morbid obesity, nutritional disorders, diabetes and cirrhosis. Thaler documented the presence of steatosis in 26.5% of his series of 10,900 liver biopsies in a 1975 review. This insightful review discussed putative or known processes and associated culprits that may result in hepatic steatosis and included many currently discussed: reduction of lipid export (malnutrition; certain medications), increased lipid delivery to the liver (weight reduction therapy, small bowel bypass surgery, overnutrition, hyperlipoproteinaemia), increased hepatic lipid synthesis (overnutrition, alcoholism, hyperlipoproteinaemia) and reduced lipid oxidation (alcoholism). Thaler recognized the link of obesity and ‘diabetic fatty liver’; although earlier studies had shown an increased prevalence of cirrhosis in diabetic patients, he suggested that evidence did not support a cirrhotogenic role for fatty liver and that other, unidentified ‘factors’ were important for this progression.

Histopathological studies in patients with clinically apparent liver disease related to obesity but not alcohol, with diabetes, and who later developed glucose intolerance and/or diabetes, as well as evaluation of liver biopsies from patients undergoing surgical bypass procedures for morbid obesity, documented lesions already known in alcoholic fatty liver disease and now recognized as NAFL and NASH. Previously, the entity was referred to by investigators that emphasized either the histological findings or presumed aetiology: nonalcoholic fatty liver with alcoholic hyaline, nonalcoholic fatty hepatitis, fatty liver hepatitis, diabetic hepatitis, steatofibrosis, ‘alcohol-like’ liver disease, nonalcoholic steatonecrosis, pseudoalcoholic liver disease and idiopathic steatohepatitis.


Adult disease

The absolute worldwide prevalence of NAFLD is not known. Estimates vary among populations and are influenced by ethnicity, dietary patterns and method of ascertainment. Arguably the most accurate noninvasive technique to detect NAFLD is 1 H-magnetic resonance spectroscopy ( H-MRS) quantification of hepatic triglyceride content (HTGC), with NAFLD defined as an HTGC >5.56%. Applying these criteria to a population sample of 2349 North American adults, the Dallas Heart Study reported an overall prevalence of approximately 31% across ethnicities (45% of Hispanics, 33% of Caucasians, 24% of African Americans). More recently, a meta-analysis incorporating >8.5 million individuals across 45 studies determined that NAFLD affects approximately 25% of the adult population worldwide. Although prevalence varied somewhat between continents, it was strikingly similar in the United States and Europe (24.1% and 23.7%, respectively), with higher levels reported in the Middle East, South America and Asia (31.8%, 31.5% and 27.4%, respectively) and lower levels in Africa (13.5%). The prevalence of NAFLD was also influenced by gender (42% white males versus 24% white females).

Because of the reliance on histological assessment for its detection, potentially influencing selection of cases for liver biopsy, estimates of NASH prevalence are subject to greater selection and ascertainment bias. The most robust of the limited data available come from histological assessment of apparently healthy prospective liver donors, in whom the prevalence of NASH was estimated at 3–16% in Europe and 6–15% in North America. In the meta-analysis, 7–30% of histologically characterized NAFLD patients without a clear indication for liver biopsy exhibited features of NASH. This equates to a NASH prevalence of 1.5–6.5% of the global general population.

As expected, prevalence of NAFLD increases substantially in groups with recognized risk factors for metabolic syndrome. In an unselected southern European population sample, 91% of obese patients (BMI ≥30 kg/m 2 ), 67% of overweight patients (BMI 25–30 kg/m 2 ) and 25% of normal-weight individuals had radiological evidence of NAFLD. Similarly, 73–97% of bariatric patients had NAFL and 25–33% had NASH. Among patients with type 2 diabetes, 40–70% had NAFLD.

Although the prevalence of most liver diseases is stable, the prevalence of NAFLD is rising and placing an increasing burden on already overstretched health care systems worldwide.

Paediatric disease

Paediatric NAFLD is now recognized as the most common cause of chronic liver disease in children worldwide. The increased incidence of NAFL and NASH in children currently is not restricted to Western populations and has paralleled the dramatic increase in obesity and type 2 diabetes mellitus in this age-group. Lifestyle changes, including a shift to more sedentary, less physically demanding activities, as well as dietary exposure to high fats and refined sugars, are implicated. Despite these facts, obesity and obesity-associated liver disease remain under-recognized by primary caregivers.

The prevalence of NAFLD in obese children and adolescents in developed countries based on elevated liver enzymes or ultrasound has been estimated at 10–77%, although ultrasound screening and elevated alanine transaminase (ALT) levels both underestimate the true prevalence of fatty liver. A histological autopsy study showed an overall prevalence of fatty liver of 9.6% in children in southern California, with older age in the paediatric range, obesity and Hispanic ethnicity being major risk factors. Paediatric NAFLD is reported to increase with body weight, up from 13–15% to 30–80% in obesity. Boys are more often affected by NAFLD than girls: 12.4% versus 3.5% in a screening study and 14.4% versus 7.4% in obese children ; both studies used serum ALT. Ethnic susceptibility is also varied in children; as in adults, obese African American children had decreased prevalence of clinical NAFLD in a series from an urban clinic based on ALT elevation. The National Health and Nutrition Examination Survey (NHANES) showed elevated serum ALT in 7.4% of Caucasian children, 11.5% of Mexican American children and 6.0% of African American children. Literature reviews indicate that similar factors and markers of inflammation are present in paediatric NAFLD as in adult disease.

Weiss et al. reported increased incidence of clinical features for cardiovascular risk factors associated with insulin resistance, now identified by a unified term, ‘metabolic syndrome’, and with severity of obesity in 490 children and adolescents, none of whom was diabetic or taking medications to lower blood pressure, lipids or glucose. All the findings were independent of age or pubertal status, confirming that well-recognized features of metabolic syndrome in adults are present in children. In their 2-year follow-up of 77 patients, metabolic syndrome persisted in all affected patients, and eight had progressed to type 2 diabetes. Schwimmer et al. further demonstrated that each feature of metabolic syndrome was more likely to occur in 150 obese children with NAFLD than in 150 obese children without NAFLD.

Beyond the classic metabolic syndrome-related forms of NAFLD, inherited conditions that share the common features of obesity and impaired glucose tolerance, insulin resistance or type 2 diabetes include Bardet–Biedl syndrome, Alström syndrome and Prader–Willi syndrome. A case report of a lean child with an inherited multisystem triglyceride breakdown and storage disorder, Dorfman–Chanarin syndrome, documented steatohepatitis and cirrhosis. Hypothalamic dysfunction with subsequent obesity is a recognized complication of many forms of treated childhood cancer. The rare disorder pseudoneonatal adrenoleukodystrophy, caused by homozygous deficiency of acyl-CoA-oxidase, an enzyme essential for fatty acid oxidation, is associated with hepatic steatosis. NASH has also been described in Werner syndrome, a rare autosomal recessive disease characterized by the premature onset of multiple age-related disorders.

Natural history

Adult disease

With obesity and insulin resistance rapidly becoming the norm in the general population of many countries, a large proportion of the population have NAFLD. Paradoxically, however, only a minority progress to advanced liver disease characterized by hepatic fibrosis, cirrhosis or hepatocellular carcinoma (HCC) and thus experience liver-related morbidity. NAFLD is therefore characterized by substantial interindividual variation in disease severity and outcome, with liver-related mortality occurring in <5% of patients and being the third most common cause of death after cardiovascular disease and extrahepatic malignancy.

Historically, dogma has dictated that, within the NAFLD spectrum, ‘simple’ steatosis (NAFL) is largely a benign condition with minimal risk of progression or clinical sequelae, and that NASH is the progressive and prognostically relevant form of the disease. Although much remains to be learned about the natural history and prognostic features of NAFLD, several recent ‘serial liver biopsy’ and ‘longitudinal follow-up’ studies have provided new insights into disease natural history that have important implications for the routine management of patients. Biopsy studies provide detailed information on how histological features of NAFL and NASH evolve and the rate at which fibrosis progresses. Follow-up studies describe the clinical disease course in patient cohorts and progression to specific endpoints (e.g. death, transplantation).

A systematic review and meta-analysis of cases with serial liver biopsy data, comprising 411 histologically characterized NAFLD patients (150 NAFL, 261 NASH) drawn from 11 small studies, demonstrated the dynamic nature of hepatic fibrosis in NAFLD. During more than 2145 person-years follow-up, 34% of patients exhibited fibrosis progression, 43% showed stable disease, and disease regression was observed in the remaining 23%. NAFL patients without fibrosis on baseline biopsy exhibited fibrosis progression at a rate of 0.07 stage/year on average, whereas NASH patients progressed more rapidly, at 0.14 stage/year. These rates equate to a one-stage increase in fibrosis over 14 years in NAFL and 7 years in NASH. Of great clinical relevance, disease progression occurred irrespective of whether the index biopsy showed NAFL or NASH; about 20% of patients showed rapid fibrosis, with progression from stage F0 to F3/4, whereas the remaining 80% had minimal progression. These findings were independently validated by a subsequent large, single-centre UK study. In 108 NAFLD patients with serial liver biopsy data covering a median interval of 6.6 years, 42% had fibrosis progression, 40% had stable disease, and disease regression was seen in 18%. No significant difference in the proportion exhibiting fibrosis progression was observed between patients with NAFL (37%) and those with NASH (43%) on index biopsy, confirming that the distinction between NAFL and NASH is of limited prognostic value. However, all patients in whom fibrosis progression was observed also exhibited features of NASH on the follow-up biopsy, suggesting that although NASH may not be present in the early phases of disease, it is a necessary pathogenic driver of fibrosis progression. The development of type 2 diabetes between biopsies was associated with greater fibrosis. Indeed, 80% of the patients with NAFL on initial biopsy who subsequently showed fibrosis progression had also developed type 2 diabetes in the interval between biopsies, compared with 25% of nonprogressors.

Longitudinal cohort studies that have followed patients representing the full spectrum of histological NAFLD at enrollment have provided important insights into long-term disease outcomes, including liver transplantation (LT) and death. Relative to a population reference sample, a cohort of 229 European NAFLD patients showed a 29% increase in all-cause mortality during an average 26 years of follow-up. The presence of NAFLD was associated with a modest increased risk of cardiovascular disease (1.29-fold) but a 3.2-fold increased risk of cirrhosis and a 6.5-fold increased risk of HCC. In multivariate analysis, only fibrosis stage at enrollment was significantly associated with overall long-term mortality, cardiovascular disease or cirrhosis. In a cohort of 619 North American patients with a median follow-up of 12.6 years, the long-term risk of LT or liver-related death was detectable even at mild stages of hepatic fibrosis. This risk increased with fibrosis stage: compared to patients with stage F0 disease (no fibrosis), those exhibiting F1/2 (mild) fibrosis had a 11.2-fold increased risk of LT or death, and those with F4 (advanced fibrosis/cirrhosis) had an 85.8-fold increased risk. Once again, the presence or stage of fibrosis, rather than discrimination of NAFL or NASH at index biopsy, was of greatest prognostic value.

It is believed that the presence or stage of fibrosis is of greater prognostic value than discriminating NAFL and NASH (despite NASH being pathogenetically necessary for fibrogenesis) because there is a greater chance of fibrosis being present in patients with NASH than in those with NAFL, which affects the outcome of multivariate analyses. This is corroborated by two additional studies, which examined 118 patients with biopsy-confirmed NAFLD followed for a median of 21 years and 209 NAFLD patients with a median 12 years of follow-up. In the former, no difference in overall or liver-related mortality was seen between NAFL and NASH patients, but those who died were more likely to have evidence of fibrosis, often stage F2 or greater. In the latter study the presence of NASH only correlated with liver-related mortality if fibrosis was included in the definition. Indeed, when the individual histological features were analyzed, only stage F3/4 was independently associated with liver-related mortality, with a 5.68-fold hazard ratio. Studies showing that noninvasive scoring systems such as the NAFLD Fibrosis Score (discussed later) are capable of predicting liver-related events, LT and death support the prognostic importance of fibrosis in NAFLD.

Based on current evidence, these findings are likely the result of a dynamic bidirectional flux between the states of NAFL and NASH, while changes in fibrosis stage occur much more slowly. Thus at liver biopsy, some patients exhibit only NAFL but will be developing NASH and, if susceptible, will ultimately progress to advanced fibrosis and experience adverse liver outcomes. In contrast, others exhibit NASH but may be regressing to NAFL. Therefore the presence or absence of NASH at biopsy provides limited long-term prognostic information. It does appear, however, that patients with only very mild/moderate steatosis in the absence of inflammation may be at the lowest risk of progression. Also, fibrosis progression in most cases is slow (F0 to F1 in ~8 years), although a subgroup of patients are ‘rapid progressors’ and will quickly progress three or even four stages over 2–6 years. In the most recent study, those with NASH were on average 9 years older than patients with NAFL, and 44% of NAFL patients had developed NASH during a median 8 years of follow-up. NASH may therefore simply reflect longer duration of disease, and NASH usually develops after steatosis.

Hepatocellular carcinoma may develop in a background of NASH-related cirrhosis (see later HCC section). Small series and numerous case reports indicate that HCC may also develop on a background of noncirrhotic NASH, but the extent to which this occurs is not well defined.

Paediatric disease

A long-term follow-up of 66 children over 20 years showed the potential for rapid progression to cirrhosis in NASH (4.75–6.6 years in two of five cases with multiple repeat biopsies), recurrence of ‘aggressive NASH’ in a transplanted liver and lower overall survival of affected children compared with age- and gender-matched children.

Factors modifying disease progression

As already discussed, NAFLD is characterized by substantial interpatient variation in disease severity and outcome. To appreciate why this occurs, NAFLD should be considered a complex disease trait in which environmental factors (e.g. dietary constituents, intestinal microbiota) act on a polygenic background of multiple subtle, interpatient genetic and epigenetic variations that determine susceptibility. Together, these determine disease severity. NAFLD likely is initiated by dietary excess, but genetic and epigenetic factors clearly contribute and modify individual response to the challenges of calorific excess and metabolic stress.

Environmental factors, principally dietary (diets high in fat and/or fructose or low in antioxidants) but also related to the intestinal microbiome, have been identified as significant modifiers of NAFLD pathogenesis. Carbohydrate intake and especially excessive fructose consumption have been linked to an increased risk of NASH and NASH-related fibrosis. Fructose bypasses the phosphofructokinase regulatory mechanism that limits the extent to which other carbohydrates can stimulate de novo lipogenesis, and high doses of fructose deplete hepatic ATP supplies. The nature of dietary fatty acids and the risk of developing NASH have also been examined, although generally using imperfect dietary recall tools. The n-3 polyunsaturated fatty acids (PUFAs) are thought to be anti-inflammatory, whereas n-6 PUFAs that have been increasing in the Western diet may be proinflammatory and may play a role in NASH. Trans fats are associated with increased cardiovascular risks, and limited data suggest also a role in liver disease. Other nondietary factors linked with disease progression include the presence of obstructive sleep apnoea (OSA).

In addition, the presence of NAFLD potentially makes the liver more sensitive to other injurious processes such as alcohol consumption. In a UK study of obese drinkers consuming ethanol at levels exceeding 150 g/week, the interaction between obesity and alcohol consumption conferred an additional 5.58-fold increase in the relative risk of liver-related death above that attributed directly to either alcohol consumption or obesity. This implies that the distinction between NAFLD and ALD becomes subjective, and that these apparently mutually exclusive conditions may coexist in some patients: ‘dual-etiology fatty liver disease’. Similarly, the coexistence of NAFLD with other causes of liver disease leads to accelerated fibrosis progression (e.g. CHC, haemochromatosis ).

Genetic modifiers

A heritable component to NAFLD risk is evidenced by both familial aggregation studies and twin studies. These show a greater concordance between identical monozygotic twins than dizygotic twins for both degree of steatosis and stage of hepatic fibrosis, amounting to almost 50% heritable contribution to variability. Ethnicity also influences susceptibility to NAFLD, with Hispanics more prone to advanced disease than Caucasians or African Americans.

A number of genes have been implicated as modifiers of NAFLD pathogenesis and progression of fibrosis. However, NAFLD is clearly not a monogenic condition. No single gene is sufficient to determine disease outcome. Instead, susceptibility is determined by the combined effects of multiple, relatively common genetic variants, each variant making a small contribution to net disease risk. Genetic variants such as single nucleotide polymorphisms (SNPs) that associate with disease severity are identified in candidate-gene studies based on a priori hypotheses, or through hypothesis generating genome-wide association studies (GWAS). The genes associated with NAFLD fall into four broad categories, influencing (1) insulin sensitivity, (2) lipid handling and metabolism (triglyceride accumulation within the liver), (3) progression to NASH (modifiers of oxidative stress, endotoxin response or cytokine/adipokine activity) and (4) modifiers of fibrosis development. Of the numerous genes associated with NAFLD, two genes stand out: patatin-like phospholipase domain-containing 3 (PNPLA3) and transmembrane 6 superfamily 2 (TM6SF2) .

PNPLA3 encodes a 481-amino acid protein (adiponutrin) that is structurally similar to the major adipose tissue triglyceride hydrolase. First identified by GWAS and subsequently validated in numerous candidate-gene association studies, a nonsynonymous SNP in PNPLA3 (rs738409) causes an isoleucine-to-methionine switch (I148M) that has been reproducibly associated with increased hepatic fat content, more severe NASH and approximately a 1.5-fold increased risk of advanced fibrosis or cirrhosis. This variant has also been linked to an increased risk of NAFLD-associated HCC, independent of the confounding effects of age, gender, diabetes, BMI and cirrhosis, and in a meta-analysis of several smaller studies. However, the precise physiological role of PNPLA3 remains poorly defined. Studies employing transgenic murine models overexpressing the wild-type or I148M variants in hepatocytes suggest that PNPLA3 plays a role in remodelling triglyceride in lipid droplets through a combination of increased formation of fatty acids and triglyceride, impaired triglyceride hydrolysis and relative depletion of triglyceride long-chain PUFAs.

The role of the TM6SF2 gene as a modifier of NAFLD has been discovered more recently. Similar to PNPLA3 , the nonsynonymous TM6SF2 (rs58542926) E167K variant is associated with increased hepatic fat content, more severe NASH and almost twofold increase in the chance of developing advanced fibrosis. At present, little is known about the functional role of the encoded protein, which is believed to be a lipid transporter localized to the intracellular endoplasmic reticulum and ER–Golgi intermediate compartments. Interestingly, whereas carriage of the minor allele has been linked to NAFLD severity, carriage of the major allele is associated with dyslipidaemia (increased serum LDL cholesterol and triglyceride) and risk of cardiovascular disease. This observation has been termed the ‘ TM6SF2 Catch-22’ paradigm.


Insulin resistance, defined as an impaired ability of insulin to evoke normal cellular responses at physiological concentrations, is frequently found in patients with NAFLD. Insulin resistance may be a major underlying abnormality in adult patients with the metabolic syndrome, and thus NASH has been called the ‘hepatic manifestation’ of the metabolic syndrome. The metabolic syndrome, although not a distinct disease but rather a clustering of abnormalities that together confer increased cardiovascular risk, is currently defined by the presence of centripetal obesity, dyslipidaemia, hypertension and elevated fasting glucose levels. Since insulin resistance is typically accompanied by compensatory hyperinsulinaemia (or pharmacological insulin replacement) to achieve glycaemic control, certain insulin signalling pathways may be overactivated while other pathways are impaired at the tissue and cellular levels.

Steatosis (nonalcoholic fatty liver)

The accumulation of triglyceride in the liver was once thought to contribute to impaired cellular function and cellular injury. However, data published over the past decade have suggested that triglyceride accumulation is relatively inert, although it does serve as an indicator of an increased flux of fatty acids, a defect in the secretion of triglyceride, or both. Therefore from the standpoint of pathogenesis, the accumulation of triglyceride as steatosis and the development of cellular injury should be considered separately. As a generalization, nonalcoholic fatty liver (NAFL) can be considered a consequence of an oversupply of fatty acids coupled with a contribution from defects in their metabolism ( Fig. 5.33 ).

Figure 5.33

Central role of lipotoxicity in nonalcoholic steatohepatitis (NASH). The emerging lipotoxicity model of NASH pathogenesis is based on the generation of nontriglyceride fatty acid metabolites primarily responsible for the hepatocellular injury and death that characterize NASH. The primary sources of hepatocellular fatty acids are uptake of circulating free fatty acids and synthesis of new fatty acids ( de novo lipogenesis). Fatty acids are normally eliminated by oxidative and secretory pathways. When these pathways are impaired or overwhelmed, lipotoxic species can be formed. The accumulation of triglyceride in lipid droplets may be an adaptive protective response that occurs when the formation and secretion of triglyceride as very-low-density lipoprotein (VLDL) is insufficient to handle the amount of triglyceride synthesized. Supply-side factors that predispose to NASH include insulin resistance at the level of adipose tissue resulting in failure to suppress lipolysis and excessive de novo lipogenesis in the liver caused by an overabundance of substrate, typically carbohydrates. A minor contributor to the hepatocellular free fatty acid burden is the uptake of circulating lipoprotein remnants (e.g. chylomicron remnants, low-density lipoprotein remnants). A potentially important source of hepatocellular free fatty acids not shown is the turnover of lipid droplets, either through lipolytic enzymes such as adipose triglyceride lipase or lysosomal breakdown of autophagosome contents. Thus, formation of lipid droplets is only a temporizing measure, and the liver must be able to handle the fatty acids released from this pool at some point. Oxidative pathways generate reactive oxygen species that can be associated with oxidant stress, but a universal role of this process in the pathogenesis of NASH has not been demonstrated. ER, Endoplasmic reticulum; SER, smooth endoplasmic reticulum.

Hepatocyte micro- and macrovesicular steatosis represent the accumulation of triglyceride in droplets surrounded by a phospholipid monolayer membrane. Hepatocytes synthesize triglyceride from free fatty acids, a property shared only by small bowel enterocytes and mammary gland epithelia. The fatty acids used to make triglyceride are mostly derived from the uptake of circulating fatty acids, the latter produced by adipocyte lipolysis. A small fraction of fatty acids in hepatocytes are synthesized through the process of de novo lipogenesis (DNL) using excessive carbohydrate and amino acids as precursors for the acetyl-CoA that feeds into the DNL pathway. An elegant study using stable isotope tracers demonstrated that this synthesis of new fatty acids normally contributes only about 5% of the fatty acids used for hepatocyte triglyceride production, but that this fraction is increased to about 25% in NASH patients. Liver biopsies with NAFLD associated with the metabolic syndrome typically show hepatocytes containing a mix of large-droplet fat that pushes the nucleus to the periphery and small-droplet fat. In contrast to this variable mix of large- and small-droplet fat, the presence of a foamy appearance caused by accumulation of microvesicular fat is often associated with mitochondrial dysfunction. Mitochondrial dysfunction may contribute to the pathogenesis of NASH and clearly plays a major role in ALD (see earlier), perhaps explaining why microvesicular fat is more common in ALD than NAFLD.

Steatohepatitis (nonalcoholic steatohepatitis)

In NASH the criteria for establishing the diagnosis of steatohepatitis are based on the presence of abnormalities associated with significant hepatocellular injury in the setting of steatosis. An emerging viewpoint based on in vivo studies and human data is that steatohepatitis is caused by fatty acid metabolites other than triglyceride, and that the accumulation of triglyceride should be considered as a visible marker of an increased metabolic burden as the liver handles excessive fatty acids, either from peripheral lipolysis or excessive DNL (see Fig. 5.33 ). Diversion of fatty acids into the triglyceride lipid-droplet pool may actually represent an adaptive protective pathway that prevents fatty acids from being used by metabolic pathways that produce lipotoxic intermediates. Establishing the identities of the molecular species responsible for lipotoxic liver injury remains an area of ongoing investigation. One possible candidate is lysophosphatidylcholine, the product of removing a fatty acid acyl group from phosphatidylcholine (lecithin). Additional candidates include free fatty acids themselves, ceramides, phosphatidic acids, diacylglycerols and the many downstream metabolites of these molecules. The number of double bonds in fatty acids, their relative positions and whether they are in the natural cis configuration or the synthetically created trans configuration also appear to be important. Fully saturated fatty acids (i.e. those without any double bonds) are lipotoxic in cell culture systems, and trans fats have been shown to cause steatohepatitis in mice, whereas PUFAs such as those found in fish oil are currently being evaluated in clinical trials as possible therapeutic agents for NASH.

Even when lipotoxic fatty acid metabolites capable of causing the phenotype of NASH are found, this mechanism is unlikely to be the sole cause of steatohepatitis in all patients with NASH. The phenotype currently identified as steatohepatitis in the absence of alcohol abuse likely represents multiple underlying disease mechanisms, either present singly or in combination. Pathogenetic factors in addition to, or perhaps caused by, the accumulation of lipotoxic species include ER stress, oxidative stress, mitochondrial dysfunction, membrane cholesterol accumulation, excessive exposure to gut-derived endotoxin and dysregulated adipokinesis. Additional recognized environmental factors that could play a role in some patients include intermittent hypoxia from OSA and changes in gut flora.


A variety of drugs have been implicated as causing NAFLD (see Tables 5.1 and 5.2 ). The most compelling data for drug-induced NASH are provided by studies of patients treated with tamoxifen who clearly developed imaging evidence of NAFLD after beginning the drug and on further evaluation were found to have NASH. The causative role of tamoxifen has been further established by the demonstration of improvement when tamoxifen was discontinued. However, several studies have shown that components of the metabolic syndrome are also risk factors for the development of NASH during tamoxifen treatment, suggesting that tamoxifen accentuates pre-existing metabolic abnormalities to cause NASH in some women.

Drug- and toxin-induced fatty liver disease is discussed in more detail later in the section on ‘NAFL and NASH in other clinical settings’.

Clinical features

No specific physical signs confirm a diagnosis of NAFLD or distinguish NAFL from NASH. Central obesity is the most common examination finding, present in over 50% of patients. However, clinical examination is frequently unremarkable, and patients with NAFLD are typically asymptomatic. For these reasons, patients with NAFLD typically present late in the course of disease, with 25–33% of patients already having progressed to advanced fibrosis or cirrhosis before diagnosis.

The two most common modes of detection of NAFLD are incidental findings of unexplained transaminase elevations and imaging evidence of increased lipid accumulation. Such abnormalities are often ignored, but if investigated appropriately, may lead to a diagnosis of NAFL or NASH. Other associated symptoms and findings, all nonspecific, include right upper quadrant abdominal pain that can vary in intensity but is usually a mild ache, fatigue and manifestations of insulin resistance, such as acanthosis nigricans and polycystic ovarian syndrome (PCOS) ( Table 5.3 ).

Table 5.3

Common associations with NASH

  • Obesity

  • Type 2 diabetes

  • Hypertension

  • Hyperlipidaemia

  • Polycystic ovarian syndrome

  • Obstructive sleep apnoea

  • Acanthosis nigricans


Because there are no pathognomonic symptoms or signs to establish a diagnosis of NAFLD, it is essential to timely diagnosis that clinicians recognize patients exhibiting features of the metabolic syndrome who may have NAFLD. Unexplained elevations of serum transaminases are a common reason for further evaluation that can lead to a diagnosis of NAFLD. Typically, these biochemical abnormalities are relatively mild, usually being less than twice the upper limit of normal. Although many NAFLD cases are first identified after an incidental finding of a mild increase in serum transaminases, it should be remembered that these tests are insensitive and that approximately 80% of NAFLD patients have normal-range alanine transaminase (ALT) levels (men <40 IU/L; women <31 IU/L), even in the presence of active NASH with advanced fibrosis. Thus in clinical practice, it is important to appreciate that absolute transaminase values do not accurately reflect histological disease severity. Aspartate transaminase (AST) levels are characteristically lower than ALT levels in the setting of NAFLD, a feature that helps distinguish NASH from ALD. However, ALT falls and AST rises as NAFLD progresses to cirrhosis, so this ratio reverses in advanced liver disease regardless of etiology. Because type 2 diabetes is common in the NAFLD population, affecting about 25% of patients, those with NAFLD frequently have increased blood glucose levels, and about 70% will have dyslipidaemia (hypertriglyceridaemia, low HDL cholesterol).

In addition, a range of other nonspecific laboratory abnormalities may be encountered. These include increased γ-glutamyltransferase (GGT), either in isolation or with mildly raised alkaline phosphatase (ALP) levels; low-titre positive antinuclear antibodies (ANA), usually at a titre ≤1:160; and anti-smooth muscle antibody (ASMA) at a titre ≤1:40, which are present in 20–30% of NAFLD cases, although with normal-range IgG levels. Elevated serum ferritin levels in the presence of a normal transferrin saturation (<45%) reflect an acute-phase response rather than iron overload. Detection of these abnormalities in patients may distract attention from the diagnosis of NAFLD and lead to its misclassification as ALD, autoimmune hepatitis or haemochromatosis. Increased ferritin levels and IgA titres appear to develop as fibrosis progresses and therefore are suggestive but certainly not diagnostic of more advanced disease. If there is uncertainty about the diagnosis of NAFLD or a desire to determine prognosis more accurately, a liver biopsy remains the gold standard test for confirming the diagnosis, as well as diagnosing NASH and staging fibrosis.


The three most common imaging modalities—ultrasound (US), computed tomography (CT) and magnetic resonance imaging (MRI)—can each detect steatosis with increasing sensitivity and cost, in that order. US is relatively insensitive, requiring at least 30% fat for consistently detectable increased echogenicity that characterizes steatosis and is qualitative rather than quantitative. Non-contrast-enhanced CT demonstrates steatosis as having a lower liver density than normal, but also lacks sensitivity for lesser amounts of fat. MRI and H-MRS are the most accurate modalities for detection of steatosis and can provide quantitative information, allowing detection of changes in hepatic lipid content. Currently, no routinely available imaging modality has been robustly demonstrated to distinguish NAFL from NASH accurately or to detect early stages of fibrosis.

Noninvasive markers of steatohepatitis and NASH-related fibrosis

Performing liver biopsies on every patient suspected of having NAFLD is neither possible in terms of health care resources nor recommended from a risk/benefit standpoint. A more pragmatic approach initially employs noninvasive strategies, including both serological tests and imaging techniques, reserving liver biopsy for patients in whom it offers clinically relevant additional information. To date, no noninvasive test to reliably distinguish NAFL from NASH has been developed, although this remains an area of active research. Tests for fibrosis stage, both commercial and noncommercial, have been more promising. Serological tests may be divided into ‘indirect’ markers (reflecting alterations in hepatic function, e.g. AST/ALT ratio) and ‘direct’ markers (biologically related to fibrogenesis and extracellular matrix turnover). These noninvasive tests typically perform best at the ends of the spectrum of disease, discriminating normal liver or cirrhosis. The NAFLD Fibrosis Score and the FIB-4 Score are calculated using standard clinical parameters such as age, presence of type 2 diabetes, BMI, AST/ALT ratio, albumin level and platelet count. NAFLD Fibrosis Score values <−1.455 effectively exclude advanced fibrosis (negative predictive value [NPV] 93%) and values >0.676 predict presence of advanced fibrosis (positive predictive value [PPV] 90%). The FIB-4 Score appears to be one of the best simple noninvasive tests for advanced fibrosis and narrowly outperforms the NAFLD Fibrosis Score. In general, simple panels have relatively robust NPV and thus can reliably exclude advanced fibrosis, but have poor PPV. The NAFLD Fibrosis Score has been shown to be predictive of long-term disease outcome and mortality. Noninvasive testing has been recently reviewed.


Liver fibrosis reduces tissue elasticity, making measurement of liver stiffness an attractive surrogate for fibrosis severity. Transient elastography (Fibroscan), an imaging technique employing US monitoring of the passage of a low-frequency pressure wave through liver tissue, has been found to be a promising noninvasive technique for the detection of advanced fibrosis caused by chronic viral hepatitis and NASH. Although abdominal obesity may compromise its utility in NAFLD, this technique has an Area Under Receiver Operating Characteristic (AUROC) of 0.84 and 0.93 for the detection of ≥F2 and ≥F3 fibrosis, respectively, and when low liver stiffness measures are obtained, elastography does reliably exclude advanced fibrosis. Novel alternatives, including acoustic radiation force impulse imaging (ARFI) and magnetic resonance elastography (MRE), are emerging but currently are not widely adopted.

Role of liver biopsy

The decision to obtain a liver biopsy weighs the small but not negligible morbidity and mortality risks associated with this invasive procedure against the benefits of (1) identifying liver disease that could progress to advanced fibrosis and cirrhosis, (2) diagnosing cirrhosis so that appropriate surveillance can be undertaken for varices and HCC and (3) highlighting other, unsuspected disease processes or coexisting processes such as iron overload, autoimmune hepatitis, diabetic hepatosclerosis and α1-AT abnormalities. Liver biopsy can also provide important prognostic information for long-term outcome in NAFLD patients. In clinical trials, liver biopsy is essential for confirming the presence of NASH, assessing and semiquantitating individual features and ultimately, evaluating the effects of the therapeutic intervention. In a research setting, in-depth study of liver tissue by IHC and molecular methods is invaluable for clarifying aspects of the pathogenesis and natural history of NAFLD.

The implementation of new, noninvasive methods may change the role of liver biopsy in NAFLD in the future. Currently, however, only liver biopsy can provide simultaneous information on steatosis, inflammation, hepatocellular injury, fibrosis and concurrent liver disease. Clinical practice guidelines issued by international hepatology societies and expert groups recommend liver biopsy for NAFLD patients at high risk for NASH and/or advanced fibrosis. The majority of these guidelines also support the need for a histological diagnosis in suspected NAFLD patients in whom other aetiology of steatosis or chronic liver disease cannot be excluded on clinical grounds.

Lifestyle modification remains the primary recommendation for patients with NASH. Having definitive biopsy evidence of aggressive disease can provide additional incentive for some patients to engage in such changes. Drug treatments are likely to be available, with varying degrees of efficacy, over the next decade, and identifying patients who might benefit from drug therapy will increasingly be a rationale for obtaining a biopsy.

Sampling and interobserver variability are known limitations of liver biopsy. These may partly be overcome, however, by taking tissue cores of adequate length (at least 16 mm) and diameter (preferable using a 16-gauge or greater diameter needle) in similar fashion from the same liver lobe, using standardized scoring systems for the interpretation of histopathological findings and/or using digital image analysis for the quantitative assessment of histological features.

There is no consensus on when a biopsy should be obtained with respect to the stage of disease. No therapeutic intervention has yet been shown to improve fibrosis, and thus it seems reasonable to perform liver biopsy at an early stage of disease rather than waiting until progression to cirrhosis with signs of portal hypertension. A common approach is to select patients with a higher likelihood of having advanced disease based on the presence of obesity, diabetes and age >40 years. However, a decision analysis accounting for risks and benefits of obtaining a biopsy versus management without a tissue diagnosis favours an early-biopsy approach to reduce long-term adverse outcomes.

Pathology of nonalcoholic fatty liver disease

Adult NAFL and NASH

In contrast to many other forms of chronic necroinflammatory liver disease, the lesions of activity in NAFL and NASH in adults are more often accentuated in acinar zone 3 ( Fig. 5.34 ). This accentuation may or may not be as readily identified in paediatric NASH; in fact, portal predominance of steatosis, inflammation and fibrosis may be seen. As the progressive remodelling of cirrhosis occurs, the necroinflammatory lesions of steatohepatitis may or may not persist. The histological features of NAFL and NASH, as well as minimum criteria for diagnosing steatohepatitis, are discussed next.

Figure 5.34

Noncirrhotic nonalcoholic steatohepatitis is characterized by zone 3 accentuation of lesions, as shown in this low-power view (H&E stain).


In noncirrhotic NAFLD, steatosis is macrovesicular and primarily located in acinar zone 3 or zones 3 and 2 in adults (i.e. sparing zone 1), in contrast to periportal or nonzonal steatosis in children or chronic hepatitis C. Steatosis was present in 100% of 433 cases of NASH from 13 series between 1979 and 2000. Studies that use Oil Red O stain for evaluation of the presence and amount of steatosis often report much greater amounts than those with standard H&E staining, and most pathologists agree that special stains for macrovesicular fat are not required. This type of steatosis is characterized by a single large droplet that occupies all (or nearly all) the hepatocytic cytoplasm, leaving the nucleus eccentrically placed (large-droplet macrovesicular steatosis). There may also be hepatocytes with combinations of single large droplets and multiple smaller droplets that are often clustered around the larger droplet (small-droplet macrovesicular steatosis). Usually, a combination of small- and large-droplet macrovesicular steatosis is seen in NAFLD.

In NAFLD, as in ALD, clusters of hepatocytes with delicate intracytoplasmic septations caused by microvesicular steatosis may occur and are often nonzonally located ( Fig. 5.35 ). Mixed steatosis, when both macro- and microvesicular types coexist, is common in NAFLD, but pure microvesicular steatosis has not been reported in NAFLD to date. The significance of microvesicular steatosis patches in NAFLD has not yet been elucidated, but their presence correlates with increased steatosis severity and progressive disease.

Figure 5.35

Nonalcoholic fatty liver disease. A portion of the parenchyma shows macrovesicular steatosis, while another area is predominantly microvesicular. (H&E stain.)

Lipid droplets are dynamic and metabolically active organelles with a central core of triacylglycerols and cholesterol esters and a peripheral layer of phospholipids with associated proteins of the perilipin/PAT family. These include perilipin, adipophilin, TIP47, S3-12 and MLDP, newly named as perilipins 1–5 to unify nomenclature and expand for newly identified molecules. Perilipins are sequentially expressed during the formation of intracellular lipid droplets, with TIP47- and MLDP-positive microvesicular steatosis evolving over time to adipophilin- and perilipin 1-positive macro­vesicular steatosis. The differences in lipid-droplet PAT–protein composition may be responsible for the different clinical significance of micro- and macrovesicular steatosis and could aid the distinction between acute and chronic steatosis. The rs738409, I148M sequence polymorphism in patatin-like phospholipid domain-containing protein (PNPLA3) has been correlated with NAFLD development. The inactive PNPLA3 accumulates on the surface of lipid droplets and is associated with an increase in macrovesicular steatosis. In NAFLD, accumulation of triacylglycerols within intracytoplasmic droplets may actually protect hepatocytes from the detrimental effect of non-droplet-bound, lipotoxic, saturated free fatty acids.

To date, the amount of steatosis required to be considered ‘pathological’ is not truly known; sophisticated imaging quantification studies have calculated that 5.6% hepatic lipid content is ‘normal’ in the American population, which approximates a commonly accepted normal value for liver steatosis of 5%, based on lipid content measurements. Steatosis >5% is therefore the minimum criterion for a histopathological diagnosis of NAFLD. Common semiquantitative assessments for steatosis are based on the percentage surface area of a histological section involved or the percentage area of uninvolved parenchyma. Zonal architecture can be a useful guide: zone 3 versus zone 1 predominance. On light microscopy, subdivisions of thirds is relatively straightforward because this incorporates the acinar divisions of the liver into zones 3, 2 and 1; thus the assessments are ≤33% (or 0–5%, 5–33%), 34–66% and >66%. The severity of steatosis may be converted into mild, moderate or severe, respectively. Although a comparison of this approach with stereological point counting suggested that semiquantitative analysis overestimates the area/volume of fat content, especially in severe steatosis, other studies assessing biochemical measurements of triglyceride have shown close association and concluded that the histopathological estimation remains useful. The accuracy of estimating the extent of steatosis with the microscope can be increased with the use of guideline images. Recently, digital image analysis (DIA) has been proposed as an objective method to quantify steatosis in liver tissue.

In patients with extensive steatosis, reticulin loss may be observed independent of the severity of inflammation or stage of fibrosis. Steatosis may ‘burn out’ during NAFLD progression and is frequently absent in NAFLD-related cirrhosis.

Steatosis and inflammation

Steatosis in NAFLD is usually accompanied by a mixed or chronic mononuclear cell inflammatory infiltrate of variable intensity located in the lobules and composed of lymphocytes (mainly T cells), rare plasmacytes and monocytes. Mixed inflammation with neutrophils is less common than chronic; eosinophils are usually seen in relation to lipogranulomas. Mild chronic or mixed portal inflammation may also be present. Single ceroid-laden PAS-diastase-positive Kupffer cells and groups of hypertrophic Kupffer cells (microgranulomas) indicating previous inflammatory activity may be seen diffusely in the lobules. A recent IHC study assessing the composition of the portal and lobular inflammatory infiltrates in NAFLD showed that CD4+ and CD8+ T lymphocytes, CD20+ B lymphocytes and neutrophil elastase-positive cells could be identified throughout the lobules and in portal tracts. However, the portal tracts were more densely populated by CD68+ macrophages and CD8+ T lymphocytes at all disease stages. Indeed, the earliest inflammatory change detected in patients with steatosis was an increase in CD68+ macrophages accompanied by an increase in proinflammatory cytokine messenger RNA expression, supporting an initiating role for the innate immune system in NAFLD. Lipogranulomas, composed of a central steatotic hepatocyte or fat droplet, an occasional eosinophil and peripheral collections of mononuclear cells and macrophages, are a frequent finding in nonsteatohepatitis NAFLD and are usually located in perivenular areas. Parenchymal lipogranulomas do not indicate active inflammation and should not be included in the evaluation of necroinflammatory activity in NAFLD. Portal lipogranulomas may also be present.

Studies on NAFLD pathogenesis have shown that inflamma­tion may be a driver for the development and progression of disease. Until recently, steatosis and steatosis with inflammation were considered ‘innocent’ and nonprogressive. However, recent data from studies with paired liver biopsies have shown that both can rarely progress to steatohepatitis with clinically significant fibrosis.

An unresolved yet practical concern remains that of ‘surgical hepatitis’. This is a well-recognized consequence of surgery, anaesthesia, organ manipulation or other causes that may result in lobular or perivenular clusters of neutrophils. The lesion itself is not considered significant for liver disease, but the polymorph clusters complicate histopathological semiquantitative analysis for lobular inflammation in liver biopsy taken during surgical procedures (e.g. bariatric). The recommended approach to prevention of surgical hepatitis is to obtain the intraoperative liver biopsy early in the procedure and before instrumentation of the liver.

Steatosis and fibrosis

In adult NAFLD, mild fibrosis, either portal or sinusoidal in zone 3, without hepatocellular injury may occasionally be encountered in cases of steatosis or steatosis with inflammation. This histological pattern probably represents an intermediate stage in the dynamic process of NAFLD development. According to traditional definitions of NAFLD, hepatocellular injury and fibrosis are not observed in cases of simple NAFL (i.e. steatosis or steatosis with inflammation) and are considered features of progression to steatohepatitis. Indeed, the presence of fibrosis in simple NAFL could indicate prior episodes of active steatohepatitis. In treated NASH patients, steatosis and fibrosis, with or without inflammation and in the absence of hepatocyte ballooning, are indicative of resolution of steatohepatitis. Sampling variability, as discussed later, may also be responsible for the absence of hallmarks of hepatocellular injury in certain patients with fibrotic NAFLD.


Hepatocyte ballooning is considered a manifestation of significant cell injury ( Fig. 5.36 ). This finding remains a ‘required’ lesion of NASH in adults for many investigators. In the seminal study that proposed four types of NAFLD based on histology, clinically significant outcomes (cirrhosis and liver death) correlated with progressive NAFLD, and ballooning was an identified characteristic of the progressive forms of NAFLD types 3 and 4. The underlying pathogenetic mechanisms that result in ballooning in NASH are not fully understood. Recent studies have shown that ballooned hepatocytes in NASH produce the ligand sonic hedgehog (Shh) that can initiate Hedgehog pathway signalling promoting fibrogenesis and can also serve as an autocrine survival factor during lipotoxic stress. Shh-positive ballooned hepatocytes are found in zone 3 and are frequently surrounded by collagen fibres (pericellular fibrosis) ( Fig. 5.36 B and D ). Their number increases with NASH severity and disease progression. Interestingly, in the PIVENS clinical trial, therapy of aggressive NASH with the antioxidant vitamin E, but not the insulin sensitizer pioglitazone, improved steatosis and lobular inflammation and reduced the number of Shh-positive hepatocytes. This reduction also correlated with an improvement in serum ALT and AST values reflecting a decrease in liver injury. In NASH, loss of cytoplasmic K8/18 immunostain and Shh-positive cytoplasmic immunostain may serve as objective markers of ballooned hepatocytes ( Fig. 5.36 C and D ).

Figure 5.36

Nonalcoholic steatohepatitis. A, Ballooned hepatocytes are more prominent in zone 3 admixed with steatosis; the involved hepatocytes are enlarged and swollen and may or may not contain Mallory–Denk bodies. B, Ballooned hepatocytes are surrounded by collagen fibres (CAB stain). C, Loss of cytoplasmic K8/18 immunostaining in ballooned hepatocytes. D, Ballooned hepatocytes are decorated by antibodies specific for sonic hedgehog (Shh) (immunostain for Shh).

Courtesy of Dr. Carolin Lackner.

Lobular inflammation is common in NASH, usually mild and composed of mixed infiltrates ( Fig. 5.37 ), in which polymorphonuclear leukocytes may be a minor (or even absent) component. In the early reported series of NASH, polymorphonuclear leukocytes were noted in 56–100% of cases in series where the feature had been specifically examined. Satellitosis is less common in NASH than in ALD. However, pathologists disagree about the requirement of polymorphs in NASH.

Figure 5.37

Nonalcoholic steatohepatitis in metabolic syndrome. There is mixed acute and chronic inflammation in this field, where steatosis and ballooned hepatocytes are seen. (H&E stain.)

Kupffer cells may be aggregated intensely in foci of ballooning and sinusoidal fibrosis and result in potential confusion for a portal tract. On IHC, Kupffer cell enlargement and aggregation in zone 3 have been shown in cases of steatohepatitis (alcoholic or nonalcoholic), whereas in non-NASH steatosis and in normal livers, CD68+ Kupffer cells were diffusely present in the lobules as well as in portal tracts. Kupffer cells play a significant role in NAFLD pathogenesis and progression by regulating hepatic triglyceride storage, mediating inflammation, contributing to hepatocyte injury and initiating fibrosis. Their numbers in NAFL/NASH have been correlated with necroinflammatory activity, progressive injury and stage of fibrosis.

Both a diet-induced steatohepatitis animal model and a study of human liver biopsies have shown a progressive increase in natural killer T cells in liver tissue samples in the progressive forms of NAFLD, although neither described microscopic localization. Both studies indicated that these cells are involved in fibrogenesis.

Portal inflammation is relatively common in NASH, is mixed lymphocytic and mononuclear and typically mild (see Fig. 5.9 ). Neutrophils and a prominent ductular reaction are suggestive of alcoholic hepatitis; numerous plasma cells and eosinophils should suggest another, possibly concurrent disease process. To date, there are four documented settings in which portal inflammation can be noted in NAFL/NASH: (1) paediatric NAFLD, (2) severe NASH in adults and children, (3) post-treatment for NASH and (4) concurrent liver disease in NAFLD. A study of 728 adults and 205 children from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)-sponsored Nonalcoholic Steatohepatitis Clinical Research Network (NASH CRN) documented mild or greater portal inflammation in 84% of adult biopsies and 90% of paediatric biopsies. The study further showed significant correlations of portal chronic inflammation with steatosis, ballooning and advanced fibrosis in adults and portal/periportal fibrosis in children. There were no correlations with lobular inflammation scores, laboratory markers of liver injury or presence of autoantibodies. In adults, greater than mild portal inflammation correlated with BMI, female gender, increased age and markers of insulin resistance. In children, correlations were only with younger age. However, in a recent study of 430 children with biopsy-proven NAFLD, portal inflammation was independently associated with more advanced disease and features of the metabolic syndrome. In adult NAFLD, portal macrophage accumulation has been associated with hepatocyte ballooning and lobular inflammation but not with the extent of steatosis. Portal inflammation may contribute directly to fibrogenesis because it is strongly correlated with fibrosis stage and the ductular reaction. The ductular reaction represents expansion of the hepatic progenitor cell compartment in response to injury and is composed of poorly formed ductular structures within an acutely inflamed fibrous stroma. Portal inflammation may also influence the fate of hepatic progenitor cells related to the ductular reaction, regulating the balance between liver repair and fibrosis.

Increased chronic portal inflammation in NASH has also been shown in studies of treatment efficacy with paired pre- and post­intervention biopsies. Portal inflammation either increased or did not decrease in concert with the other lesions of steatohepatitis after clinical and histopathological reversal of NASH and fibrosis. Other examples include bariatric surgery and medical therapy. Another consideration for increased portal inflammation in fatty liver is in the setting of concurrent liver injury of another origin, such as chronic viral hepatitis or autoimmune liver disease, as discussed later.

Mallory–Denk bodies

In NASH, Mallory–Denk bodies (MDBs) are noted in ballooned hepatocytes, usually in zone 3, and are often described as ‘poorly formed’. In contrast, abundant and well-formed MDBs are more suggestive of alcoholic hepatitis. MDBs may be chemotactic, as described earlier, and affected hepatocytes may be surrounded by polymorphs (satellitosis) ( Fig. 5.38 ). Immunostains for p62 and ubiquitin are useful for the detection of early forms not visible in routinely stained sections (see Fig. 5.5 ). In their prognostic study of NAFLD, Matteoni et al. used MDBs as a lesion to delineate the progressive type 4 from types 1, 2 and 3. Other studies of NASH have concurred that MDBs are more often found in conjunction with higher necroinflammatory grades of steatohepatitis and other markers of progression and fibrosis in NAFLD.

Sep 29, 2019 | Posted by in NEPHROLOGY | Comments Off on Fatty Liver Disease
Premium Wordpress Themes by UFO Themes