Study Selection: All studies, including case reports, evaluating the salient features and clinical profile of NASH.

Data Extraction: Data were selected from all articles that fit the study selection criteria.

Data Synthesis: Nonalcoholic steatohepatitis is a distinct clinical entity characterized by elevated plasma liver enzyme levels and liver biopsy findings that are identical to those seen in alcoholic hepatitis; patients with NASH, however, do not consume alcohol in quantities known to cause liver injury. Patients with NASH are typically obese, middle-aged women with asymptomatic hepatomegaly who are diabetic or hyperlipidemic and present with an unrelated medical problem. Analysis of liver biopsy specimens is the cornerstone of diagnosis; hepatic morphologic findings range from mild fatty degeneration and inflammation to cell degeneration, fibrosis, and cirrhosis with or without the presence of Mallory hyaline bodies. Elevated levels of free fatty acids in the liver are thought to be responsible for the development of steatohepatitis. Although NASH is most often a benign disease with an indolent course, patients with this condition occasionally develop cirrhosis, portal hypertension, and hepatic failure. In some cases, NASH may be reversed with weight reduction.

Conclusion: Nonalcoholic steatohepatitis is an important differential diagnosis for asymptomatic patients with chronically elevated plasma liver enzyme levels, especially if obesity, diabetes, or hyperlipidemia are present. Analysis of liver biopsy specimens is necessary for diagnosis and must be done in all patients with unexplained abnormal liver function and negative results on a noninvasive workup. Prognosis is good in most patients. The precise role of weight reduction and ursodeoxycholic acid therapy in the favorable alteration of the natural history of this disorder needs to be addressed in large, well-controlled studies.

Although the prevalence of NASH is not well-defined, it is reported worldwide and is detected in 1.2% to 9% of patients who have liver biopsy. In fact, the prevalence of NASH may be even higher in asymptomatic patients who do not consume substantial quantities of alcohol and in those who have negative results on serologic examinations for viral hepatitis. Thus, many patients with elevated plasma liver enzyme levels and negative results on noninvasive workup may have NASH. Nonalcoholic steatohepatitis is seen most frequently in patients who are female, are morbidly obese, are diabetic, have had jejunal bypass surgery, or have been receiving certain medications. These patients have either no symptoms or vague abdominal discomfort. Levels of plasma aspartate aminotransferase (AST) and alanine aminotransferase (ALT) are usually elevated. The diagnosis of NASH is confirmed by liver biopsy. The histologic features of the condition include macrovesicular fatty changes; parenchymal inflammation, fibrosis, or both; cirrhosis; and the presence of Mallory hyaline bodies [1-3]. The pathogenesis of NASH is unclear, but various theories have implicated free fatty acids. Although NASH is generally considered to be a benign disease, it occasionally progresses to cirrhosis with a fatal outcome.

In this review, we critically examine and discuss the epidemiology, pathogenesis, clinical features, histology, clinical course, diagnosis, and management of NASH and determine the clinical relevance of the condition.

Methods

Top Methods Author & Article Info References

Through a MEDLINE search done using the keywords steatohepatitis, alcoholic, nonalcoholic, fatty liver, hepatitis, and alcohol drinking, we identified and reviewed all articles on NASH published in English-language journals between 1966 and 1996. We also used the bibliographies of relevant articles to augment our search.

What Is Nonalcoholic Steatohepatitis?

"Alcoholic hepatitis" is a well-defined term used to describe the histologic pattern of liver injury that results from long-term consumption of alcohol. The condition is characterized by cell damage that is often distinguished by 1) the presence of Mallory hyaline bodies, an inflammatory infiltrate predominantly comprising neutrophils with frequently associated fatty degeneration [steatosis], and 2) fibrosis ranging from minimal perivascular foci to dense scarring and fibrous septa [4, 5]. Nonalcoholic steatohepatitis is initially a perivenular injury but with severe disease, the portal tracts may be involved, creating necrotic and fibrotic bridges that can lead to cirrhosis [6]. These findings have also been described in patients who do not consume alcohol or who consume it in amounts not known to cause substantial liver injury.

Powell and colleagues [3] proposed the following criteria for the diagnosis of NASH:

1. A liver biopsy specimen that shows moderate-to-gross macrovesicular fatty degeneration with inflammation (lobular or portal) with or without Mallory hyaline bodies, fibrosis, or cirrhosis.

2. Convincing evidence of negligible alcohol consumption (<40 g of ethanol per week) that includes a detailed history taken by three physicians independently and interrogation of family members and local medical practitioners. Results of random blood assays for the estimation of ethanol levels should be negative. (If done, assays for the presence in serum of desialylated transferrin, a marker of alcohol consumption, should also be negative [7]).

3. Absence of serologic evidence of previous infection with the hepatitis B virus or a clinical course that suggests non-A, non-B hepatitis (hepatitis C).

We believe that the diagnosis of NASH should not necessarily be ruled out in patients who have had infection antibodies to the hepatitis B virus. However, it should be ruled out for patients who are positive for hepatitis B surface antigen or antibodies to the hepatitis C virus.

Epidemiology and Associated Conditions

The prevalence of NASH in the general population has not been defined. Among patients who have had liver biopsy, the prevalence is approximately 7% to 9% in western countries [1, 8] and 1.2% in Japan [9]. Data from patients who have had liver biopsies show that alcoholic hepatitis is 10 to 15 times more common than NASH. These figures do not reflect the prevalence of these diseases in the general population [1, 10, 11].

Although NASH has been reported in persons in the second decade of life [12, 13], most cases occur in persons in the fifth and sixth decades of life [1, 9, 10, 14, 15]. Cases occur more frequently in women (65% to 83%) [1, 3, 10, 14], although Bacon and colleagues [15] recently found a high preponderance of NASH in men (58%). It is not clear whether it is a hormonal effect or the higher prevalence of obesity in women that makes NASH more prevalent in women.

Obesity is the condition most often associated with NASH: In most studies, 69% to 100% of patients with NASH were also obese [1, 3, 10, 14, 16]. However, Bacon and colleagues [15] reported a prevalence of obesity of 39%. Most patients with NASH are 10% to 40% heavier than ideal body weight [1, 2, 10, 14, 15]. Nonalcoholic steatohepatitis has also been seen in obese patients who have had surgery for weight reduction. In fact, liver dysfunction occurs in 40% of obese patients who have had jejunal bypass surgery for weight reduction [17]. In 2.2% to 6% of these patients, evidence of liver failure appears during the first 12 to 18 months, when weight loss is most rapid [18].

Non-insulin-dependent diabetes mellitus and elevated blood glucose values are noted in 34% to 75% of patients with NASH [1, 3, 10, 14, 16, 19], although Bacon and colleagues [15] reported a lower incidence (21%) and Baldridge and associates [12] found normal fasting blood glucose levels in 14 (100%) obese peripubertal children with NASH. In an autopsy study, Wanless and Lentz [2] found obesity, diabetes, or both to be present in 20 of 22 patients with steatohepatitis. They also noted a trend toward a higher prevalence of NASH in patients with type II diabetes requiring insulin.

In addition to obesity and non-insulin-dependent diabetes mellitus or hyperglycemia, Wanless and Lentz [2] found a substantially higher incidence of NASH in patients receiving intravenous glucose therapy in the last week of life and in obese patients who lost weight in the last month of life compared with obese patients who did not lose weight.

Hyperlipidemia (hypertriglyceridemia, hypercholesterolemia, or both) is another common abnormality and has been reported in 20% to 81% of patients with NASH [1, 3, 10, 14-16, 19]. In contrast with all other major studies, Bacon and colleagues [15] found that 14 of 33 patients with NASH had normal weight, blood sugar levels, and lipid levels.

Other conditions that are less commonly associated with NASH include extensive small-bowel resection [20]; total parenteral nutrition [1, 21, 22]; jejunal diverticulosis with bacterial overgrowth [23]; gastroplasty for morbid obesity [24]; biliopancreatic diversion [25]; Weber-Christian disease [26]; partial lipodystrophy that does not affect the face [27]; abetalipoproteinemia [28]; and therapy with such drugs as amiodarone [29], tamoxifen [30], perhexiline maleate [31], glucocorticoids [32], and synthetic estrogens [33] (Table 1).

Pathogenesis

The pathogenesis of NASH remains unclear. Steatosis is commonly encountered in protein-energy malnutrition, obesity, acute starvation, carbohydrate overload, and corticosteroid therapy, and there may be a common mechanism for the accumulation of triglycerides in the liver that is associated with these conditions [34]. Accumulation of fat in the liver can occur because of increased delivery of free fatty acids to the liver, increased synthesis of fatty acids in the liver, decreased ß-oxidation of free fatty acids, and decreased synthesis or secretion of very-low-density lipoprotein cholesterol [34, 35]. Increased levels of free fatty acids have been noted in the liver tissue of obese patients [36, 37]. Accumulation of free fatty acids may be responsible for liver dysfunction because fatty acids are highly reactive and can damage biological membranes [3, 38].

The relation between steatosis, steatohepatitis, and fibrogenesis has not been elucidated. Hepatic peroxidation of lipids may result in the generation of potentially toxic intermediates that can induce an inflammatory response in the liver [39, 40]. However, isolated fatty liver without hepatitis occurs more frequently than does steatohepatitis. Thus, it is not clear whether the accumulation of fat in the liver is responsible for inflammation or whether inflammation evoked by some stimulus causes cell dysfunction that results in steatosis. Experimental evidence indicates that the infiltration of fatty acids increases the production of fibrous tissue in the liver [35]; increased fibroblasts have been seen with cholesterol-induced fatty liver in rabbits [41], and an increase in the hepatic content of hydroxyproline and glycosaminoglycan was seen in mice that were fed a protein-calorie balanced diet supplemented with cholesterol or choline [42]. The earliest event in the development of fibrosis appears to be activation of the hepatic lipocyte (also known as the fat-storing or Ito cell) in the subendothelial Disse space by such factors as lipid peroxides. This causes proliferation of lipocytes and initiation of fibrogenic cascade in the liver [43].

Intraperitoneal injection of insulin during long-term peritoneal dialysis results in steatosis and steatohepatitis in the subcapsular portion of the liver in patients with insulin-dependent diabetes [44]. This suggests that insulin may play a role in NASH associated with obesity. Insulin inhibits oxidation of free fatty acids, thereby increasing levels of toxic free fatty acids in the liver [45]. Obese patients not only have greater stores of free fatty acids in adipose tissue but are also resistant to insulin and have higher plasma levels of free fatty acids [46-48]. Insulin levels in obese patients may be normal or elevated [49, 50] and are sufficient to inhibit hepatic disposal of free fatty acids. Thus, elevated insulin levels favor the accumulation of free fatty acids in the liver.

The pathogenesis of NASH after jejunal bypass appears to be multifactorial. Potential mechanisms include absorption of bacterial products or bile acids from the blind loop [51-54], protein malnutrition [55], deficiency of free fatty acids and vitamin E [52], and mobilization of free fatty acids during weight loss [56]. A diet that is low in carbohydrates and fat and high in protein has been reported to prevent acute liver failure after jejunal bypass [57].

In the study by Wanless and Lentz [2], patients often lost weight before death while caloric supplementation was provided in the form of intravenous carbohydrates. This may have stimulated insulin secretion in amounts sufficient to inhibit oxidation of free fatty acids in the liver but not to prevent mobilization of free fatty acids from adipose tissue. Weight loss is associated with marked lipolysis and accumulation of free fatty acids in the liver [58]. When obese patients lose weight through fasting, insulin levels are reduced. Paradoxically, obese patients who have had intestinal bypass surgery for weight reduction have increased insulin levels despite weight loss because they continue to eat, which stimulates insulin secretion. Increased insulin levels in these patients result in the accumulation of free fatty acids in the liver.

Finally, the underlying mechanisms of liver injury in NASH and in alcoholic hepatitis may be similar, as suggested by similar ratios of mitochondrial AST to total AST [59]. Accumulation of free fatty acids in the hepatocytes may cause mitochondrial swelling, increased fragility, and membrane permeability that may, in turn, be responsible for an increased ratio of mitochondrial AST to total AST.

Clinical Features and Laboratory Evaluation

The salient clinical and laboratory features of NASH are outlined in Table 2. A high proportion of patients (48% to 100%) have no symptoms of liver disease, and a small percentage (especially children [12, 13]) have vague abdominal discomfort or pain in the right upper quadrant [13, 15] or fatigue and malaise [10, 15]. Most patients present because of other conditions (such as hypertension, gallstones, coronary artery disease, congestive heart failure, cancer, peripheral vascular disease, hypothyroidism, and gynecologic or psychiatric conditions) [10, 16] and are incidentally found to have abnormal liver function [10, 15]. Significant alcohol consumption must be ruled out in all patients.

The most common finding at initial presentation is asymptomatic hepatomegaly [3, 10, 14, 15] without evidence of chronic liver disease [14]. The most frequently noted abnormality is a two- to threefold elevation of levels of ALT and AST in plasma [3, 10, 12, 14, 15, 19]. In two major studies [3, 15], levels of ALT were noted to be higher than levels of AST, a pattern that contrasts with that seen in alcoholic hepatitis. Van Ness and Diehl [60] found that 19% of patients (17 of 90) who had had liver biopsy for evaluation of chronically elevated plasma levels of ALT and AST in contrast to 7% to 9% of all patients who had had liver biopsies for other reasons, had nonalcoholic steatosis or steatonecrosis [1, 8]. Alkaline phosphatase levels are abnormal in fewer than half of patients, and bilirubin levels are rarely elevated [10, 15]. Serum albumin levels are almost always normal [10, 15], and a prolonged prothrombin time is atypical (although it was noted in 50% of patients in one series [1]).

Bacon and colleagues [15] identified abnormal results of iron tests (elevated levels of serum ferritin and transferrin saturation) in 18 of 31 patients. They reported elevations as great as fivefold in serum ferritin levels. None of the patients had histologic evidence of idiopathic genetic hemochromatosis.

The prevalence of hepatitis C virus has not been systematically studied in patients with NASH. Rogers and colleagues [61] were not able to detect hepatitis C virus by polymerase chain reaction assay in a small series of five patients. Changes in fatty acids similar to those encountered with NASH can be caused by hepatitis C virus [61], and inflammation in NASH may be lymphocytic and periportal, thus mimicking the histologic changes seen in chronic hepatitis C infection. Chronic hepatitis C virus infection and possibly chronic hepatitis G virus infection [62] must be excluded in all patients with steatohepatitis. Ceruloplasmin levels, ₁-antitrypsin levels, renal function test results, and electrolyte levels are usually normal in patients with NASH; tests for hepatitis B surface antigen are negative in these patients [15].

The diagnosis of NASH depends on the absence of a history of excessive alcohol consumption. This may be difficult to verify, and many patients who consume excessive amounts of alcohol underestimate or deny alcohol ingestion [63]. In fact, Pinto and colleagues [19] recently showed that it is difficult to differentiate patients with NASH from asymptomatic ambulatory patients with alcoholic hepatitis on the basis of clinical and biochemical evaluations.

Several markers are currently used as indicators of excessive alcohol consumption. Two of these, the presence of partially desialylated transferrin [64-66] and mitochondrial isoenzyme of AST [67-69], have been reported to be more sensitive and specific than such routinely used markers as -glutamyltransferase, mean corpuscular volume, and levels of AST and ALT. Fletcher and colleagues [59] recently showed that a ratio of desialylated transferrin to total transferrin greater than 0.013 had 81% sensitivity and 98% specificity in diagnosing ongoing alcohol consumption. Abstinence from alcohol for more than 7 days resulted in a mean decline in the ratio of 28%. They also found that the ratio of mitochondrial AST to total AST had a specificity of only 50% in differentiating patients with nonalcoholic steatohepatitis from alcoholic persons. Such conventional markers as mean corpuscular volume, -glutamyltransferase levels, and an AST-to-ALT ratio greater than 2 were not useful in differentiating patients with NASH from alcoholic persons in general [59] and patients with asymptomatic alcoholic hepatitis in particular [19]. Thus, it seems that the ratio of desialylated transferrin to total transferrin is the best single marker for long-term excessive alcohol consumption [59] and would be a useful aid in identifying alcoholic persons. However, this marker has not been used in most studies. Moreover, this test is not readily available, forcing physicians to rely on a thorough history to distinguish alcoholic from nonalcoholic patients.

Histology

In patients who do not consume alcohol, liver biopsy specimens with histologic features similar to those seen in alcoholic hepatitis are the cornerstone of the diagnosis of NASH. The typical histologic features of NASH are shown in Figure 1; various histologic features have been described. In many studies [1, 3, 15], NASH has been diagnosed by the presence of fatty degeneration (steatosis) and lobular inflammation (hepatitis) only; hepatocyte degeneration, necrosis, fibrosis, cirrhosis, and Mallory hyaline bodies may not be present. Other studies [2, 10] have used a stricter definition for the histologic diagnosis of NASH, including steatohepatitis and hepatocyte ballooning or degeneration or hepatic fibrosis. We agree that in addition to steatosis and inflammation, either hepatocyte ballooning or degeneration or hepatic fibrosis should be present-the presence of these features is the key to diagnosing NASH.

Moderate-to-severe steatosis (macrovesicular steatosis more often than microvesicular steatosis) is seen in all patients who have NASH. Fatty degeneration may be diffuse or located primarily in the central zone and accompanied by fat cysts [1]. Inflammation is usually centrilobular [10], although mild involvement of portal and periportal areas may also be seen [1]. The cellular response may be predominantly neutrophilic [10], predominantly lymphocytic [1], or mixed [15].

The presence of Mallory hyaline bodies varies (they are present in 9% to 90% of cases [1-39, 10, 12, 15, 19, 61]). These bodies are usually sparse, small, and inconspicuous compared with those of patients with alcoholic hepatitis and are frequently found in zone 3 of Rappaport [1]. They are large and antler-shaped in patients with alcoholic hepatitis [1]. Itoh and colleagues [70] studied the ultrastructure of Mallory hyaline bodies in patients with NASH and found these bodies to be similar to those seen in patients with alcoholic hepatitis. Powell and colleagues [3] detected Mallory hyaline bodies in 4 of 42 patients with NASH. All 4 of these patients were obese, 2 were diabetic, and 2 were hyperlipidemic. Mallory hyaline bodies have also been described in cases of NASH associated with Weber-Christian disease [26], gastroplasty [24], amiodarone use [29], and biliopancreatic diversion [25]. The bodies were not detected in NASH that was secondary to jejunal diverticulosis [23] or limb lipodystrophy [27].

Fibrosis in association with NASH ranges from absent to severe and may be perisinusoidal, centrilobular, or septate. The prevalence of mild to moderate fibrosis in patients with NASH varies from 76% to 100% [1, 10, 14, 16, 19]; that of severe fibrosis ranges from 15% to 50% [1, 3, 10, 14-16]. Cirrhosis, however, occurs less frequently in adults (7% to 16%) [1, 3, 10, 15] and is absent in children [12].

Two major studies [14, 19] have compared the histologic features of alcoholic hepatitis with those of NASH. These studies found a higher prevalence of nuclear vacuolation (reflecting the presence of diabetes) and steatosis in NASH than in alcoholic hepatitis. Periportal and pericellular fibrosis and proliferation of the bile duct was more common in patients with alcoholic hepatitis. Itoh and colleagues [14] found no difference in the extent of centrilobular necrosis, extent of centrilobular fibrosis, and presence of Mallory hyaline bodies between the two groups. Recently, Pinto and colleagues [19] compared histologic findings in patients with NASH, ambulatory patients with asymptomatic alcoholic hepatitis, and hospitalized patients with alcoholic hepatitis. They reported that the degree of severity of hepatocellular damage, Mallory hyaline bodies, neutrophilic and mononuclear inflammation, and fibrosis increased from patients with NASH to ambulatory patients with asymptomatic alcoholic hepatitis to hospitalized patients with alcoholic hepatitis.

Wanless and Lentz [2] studied liver biopsy specimens obtained from obese patients and controls. Using the so-called ballooning of hepatocytes with clearing of hepatocellular cytoplasm accompanied by large-droplet steatosis as the minimal criteria for steatohepatitis, they found that the prevalence of steatohepatitis was 18.5% in obese patients and 2.7% in lean patients. Other studies [71, 72] have shown the prevalence of NASH to be lower (1% to 9%) among obese patients. The reason for this discrepancy was thought to be that Wanless and Lentz [2] had larger amounts of tissue available for examination, suggesting sampling variability. In addition, most of the patients studied by Wanless and Lentz were catabolic in their preterminal state, and the possibility of recent alcohol consumption could not be excluded.

Wanless and Lentz [2] also found that the amount of steatosis was proportional to the degree of obesity and was not related to non-insulin-dependent diabetes mellitus. Weight loss before death did not affect the prevalence of steatosis. Wanless and Lentz also showed the severity of steatohepatitis to be proportional to the degree of obesity. The presence of non-insulin-dependent diabetes mellitus increased the prevalence of steatohepatitis 2.6-fold, with a trend toward increased prevalence in patients with non-insulin-dependent diabetes mellitus requiring insulin. Patients receiving intravenous glucose therapy in the last week of life and obese patients who lost weight in the last week of life had a higher incidence of steatohepatitis than did obese patients without weight loss. The prevalence of steatohepatitis was not altered by the duration of the hospital stay. In contrast, Powell and colleagues [3] found no correlation between the degree of obesity or the presence of diabetes or hyperglycemia and the severity of histologic features.

Wanless and Lentz [2] showed that the prevalence of fibrosis was higher in severely obese patients than in lean patients. However, this effect was explained by the presence of non-insulin-dependent diabetes mellitus in the obese patients. At any degree of obesity, men and women were equally likely to have steatosis, steatohepatitis, and fibrosis.

The histologic features of NASH encompass a wide spectrum, from mild steatohepatitis to bridging fibrosis and cirrhosis. Lee [73] pointed out that the inclusion criteria for NASH in many recent studies (including that by Bacon and colleagues [15]) required the presence of macrovesicular steatosis with parenchymal inflammation only. Although this literally means steatohepatitis, on the one hand, it includes cases with benign steatosis and nonspecific reactive inflammation; on the other hand, it could encompass such unrelated conditions as Wilson disease, galactosemia, and methotrexate toxicity. Thus, the milder forms of NASH bear no similarity to alcoholic hepatitis, which, according to the diagnostic criteria proposed by an international group of pathologists [5], must include hepatocyte damage, fibrosis, and neutrophilic inflammation with or without Mallory hyaline bodies. Therefore, to narrow the clinical and histologic spectrum of this disease, hepatocyte degeneration and ballooning (with or without Mallory hyaline bodies) or fibrosis in zone 3 must be present for a diagnosis of NASH. In fact, most of the patients studied by Bacon and colleagues [15] who had more than trivial fibrosis and met the stricter diagnostic criteria were women who were obese, had diabetes or hyperlipidemia, and otherwise showed the typical profile for patients with NASH. We agree with Lee [73] about the need for strict inclusion criteria [5] for the histologic diagnosis of NASH to avoid confusion with milder forms of steatohepatitis that result from other causes.

Clinical Course

Nonalcoholic steatohepatitis has generally been thought of as a stable disease. Lee [10] followed 39 patients clinically for an average of 3.8 years. During this period, only 1 patient developed hepatic failure. The patient was an obese, diabetic woman who had cirrhosis on biopsy. She developed increasing jaundice, ascites, and encephalopathy and died of gastrointestinal bleeding 1.6 years after diagnosis. Ten of 39 patients died because of an underlying medical condition that was not related to liver disease. Three of the 5 patients with cirrhosis died, and 2 remained stable. Follow-up biopsies done in 13 patients over an average of 3.5 years showed no increase in fibrosis or substantial change in morphology in 8 patients. Five patients had a progression of fibrosis during 1.7 to 6.1 years, and 2 developed cirrhosis. No progression of fibrosis was reported by Falchuk and colleagues [74] in 2 of 5 patients who had biopsies repeated after 1 year.

Powell and colleagues [3] repeated biopsies in 13 of 42 patients over a 1- to 9-year period. In 6 patients, no morphologic changes were noted. Two patients who had had active cirrhosis or marked fibrosis showed inactive cirrhosis or marked fibrosis. Three patients who had initially had fatty inflammation developed fibrosis during 1-, 2-, and 6-year periods. One patient had loss of inflammation and fibrosis at 6 years, and another progressed from fibrosis to cirrhosis during the same period. The patient who had had cirrhosis on initial biopsy developed hepatocellular cancer after 6 years. Bacon and colleagues [15] documented no histologic progression in 1 patient at 4 years and progression from bridging fibrosis to cirrhosis in 1 patient after 7 years.

Combining the results of Powell and colleagues [3], Lee [10], and Bacon and colleagues [15] gives a total of 28 patients. Of these, 1 (3%) improved, 15 (54%) remained unchanged, and 12 (43%) had histologic progression during 1- to 7-year periods. Approximately 8% to 17% of patients with NASH [3, 10] compared with 38% to 50% of patients with alcoholic hepatitis [75, 76] progress to cirrhosis after a similar follow-up period.

Cirrhosis related to obesity is present in 0.3% of the general population at autopsy and 1.8% of obese patients at autopsy. Cirrhosis related to obesity comprises approximately 12% of all cases of cirrhosis [2]. In addition to cirrhosis related to obesity, evolution to cirrhosis after jejunal bypass has also been clearly established [77, 78].

Propst and colleagues [8] compared survival rates of patients with alcoholic hepatitis with those of patients with NASH using Kaplan-Meier survival curves. They found that the 5- and 10-year probabilities of survival were 38% and 15% for alcoholic hepatitis and 67% and 59% for NASH. They also reported that patients with NASH did not have a lower life expectancy than age- and sex-matched controls from the normal population.

The effect of weight loss is variable, and it is interesting to note that although the histologic lesion of steatohepatitis deteriorates after rapid weight loss [79], improvement occurs after gradual weight reduction [80].

The clinical status and parameters of liver disease of most patients with NASH remain stable. Lee [10] showed that no statistically significant change in liver function occurred for an average of 3.8 years. The degree of obesity did not substantially alter the course [10].

Although patients with NASH generally have an indolent course, nearly half develop progressive fibrosis and as many as one sixth develop cirrhosis. No clinical, laboratory, or histologic features can predict progression or distinguish patients with or without worsening liver disease.

Diagnosis

The clinician's ability to predict the nature of the pathologic process that causes chronic elevations of plasma levels of ALT and AST without examining liver biopsy specimens remains imperfect. Van Ness and Diehl [60] showed that the predictive value of a diagnosis of nonalcoholic fatty liver before biopsy was only 56% for NASH compared with 86% for alcoholic liver disease.

In patients with abnormal liver function, extensive evaluations should be done to determine the cause of the abnormality. An accurate history often suggests alcoholic liver disease or drug-related abnormality. Laboratory testing, including serologic tests for viral hepatitis; iron studies; and measurements of ceruloplasmin levels, levels and phenotype of -antitrypsin, and antimitochondrial and antinuclear antibodies, should be done for potentially treatable causes of chronic liver diseases. Such radiologic studies as liver ultrasonography and Doppler ultrasonography of hepatic and portal vessels are also appropriate. Ultrasonography in NASH often shows a hyperechoic texture or a bright liver [81] because of diffuse fatty infiltration. However, this finding is not specific and cannot be used to diagnose NASH. If none of the above tests are definitive, then a liver biopsy is recommended to confirm or rule out NASH, especially if the patient has the typical clinical features described earlier.

Management

No proven therapy for NASH exists. Because obesity is the condition most commonly associated with NASH, weight loss is frequently advocated. However, weight loss is difficult to evaluate because obese patients with NASH rarely achieve or maintain sustained reductions in weight. Moreover, the effect of weight loss on liver disease is not consistent [11, 79, 80, 82, 83]. Eriksson and colleagues [11] studied three patients with histologic features of NASH who were 50% to 60% overweight. The patients all responded with normalization of biochemical values, and two had normalization of histologic changes (liver biopsy was not repeated in the third patient) after weight reduction of 9% to 28%, although the patients still exceeded ideal body weight. Abdelmalek and colleagues [83] reported a patient with abnormal liver function and mild steatohepatitis. The patient was empirically treated with ursodeoxycholic acid for 1 year and had complete biochemical normalization of liver test results despite a weight gain of 2.72 kg (6 pounds). A repeated liver biopsy showed only mild, patchy portal hepatitis and periportal fibrosis without necrosis or fatty changes. The patient discontinued therapy with ursodeoxycholic acid on her own and developed abnormal liver function after 1 year; however, normal function returned after therapy with ursodeoxycholic acid was reinstituted.

Ursodeoxycholic acid has both lipid-altering properties and direct cytoprotective effects [84] that may have caused the improvement in this patient. Laurin and colleagues [85] recently studied the role of ursodeoxycholic acid and clofibrate therapy in the treatment of patients with NASH. Twenty-four patients received ursodeoxycholic acid, 13 to 15 mg/kg of body weight per day, and 16 patients with hypertriglyceridemia received clofibrate, 2 g/d. Both groups received treatment for 1 year. Despite its lipid-lowering properties, clofibrate was not found to be beneficial in the treatment of NASH. In contrast, ursodeoxycholic acid therapy significantly reduced plasma levels of ALT, alkaline phosphatase levels, -glutamyltransferase levels, and hepatic steatosis. Randomized, controlled trials are required to validate the effect of ursodeoxycholic acid therapy in patients with NASH.

Conclusions

Nonalcoholic steatohepatitis is a distinct clinical entity that is occasionally encountered in adults and children. It forms an important differential diagnosis for asymptomatic patients with chronically elevated plasma levels of AST and ALT, especially if obesity, diabetes, or hyperlipidemia are present. Diagnosis is confirmed by liver biopsy that shows variable histologic features, ranging from mild steatohepatitis to severe fibrosis and cirrhosis. Although NASH is usually a benign condition with an indolent course, nearly half of cases progress, and as many as one sixth of patients with NASH develop cirrhosis. No definite therapy for NASH exists, although weight reduction or ursodeoxycholic acid therapy may result in improvement.

Dr. Gordon: Division of Gastroenterology, Beth Israel Deaconess Medical Center, 110 Francis Street, Suite 8E, Boston, MA 02215.

Dr. Chopra: Division of Gastroenterology, Beth Israel Deaconess Medical Center, Dana 501, 330 Brookline Avenue, Boston, MA 02215.