nonalcoholic fatty liver disease - semantic scholar€¦ · nonalcoholic fatty liver disease jeanne...

9
GASTROENTEROLOGY 2002;122:1649-1657 Nonalcoholic Fatty Liver Disease JEANNE M. CLARK, FREDERICK L. BRANCATI, and ANNA MAE DIEHL Departments of Medicine and Epldemiology, The Johns Hopkins University, Baltimore, Maryland N 'onalcoholic fatty liver disease (NAFLD) includes a spectrum of hepatic pathology that resembles al- cohol-induced liver disease but develops in individuals who are not heavy drinkers. NAFLD is likely to be the most common cause of chronic liver disease in many countries and may also potentiate liver damage induced by other agents, such as alcohol, industrial toxins, and hepatotrophic viruses. The lack of specific and sensitive noninvasive tests for NAFLD limit reliable detection of the disease. Often, NAFLD is diagnosed presumptively when imaging studies suggest hepatic steatosis or when liver enzyme elevations are noted in overweight or obese individuals with no other identifiable reason for liver disease. NAFLD is strongly associated with insulin re- sistance and dyslipidemia; however, whether it is a cause or a consequence of these other conditions in not clear. The natural history of NAFLD is also uncertain. Never- theless, discrepancies between the apparently high prev- alence of NAFLD in the population and the generally low prevalence of clinically significant liver disease, as well as the relatively low representation of NAFLD pa- tients among liver transplant populations, have gener- ated considerable skepticism about the clinical impor- tance of NAFLD. The latter has reduced enthusiasm for aggressive diagnosis, treatment, and follow-up of pa- tients with NAFLD. This review of the clinical literature and selected basic information about NAFLD challenges some of these assumptions. NAFLD--What Is It? The diagnosis of NAFLD requires a combination of invasive and noninvasive tests. Liver biopsy is the most sensitive test for detecting and staging fatty liver disease. Fatty liver (hepatic steatosis) is the lesion at the most clinically benign end of the spectrum. Large (macro-) and small (micro-) vesicles of fat, predominately triglycer- ides, accumulate within hepatocytes without causing appreciable hepatic inflammation, liver cell death, or scarring. Cirrhosis is the lesion at the opposite end of the spectrum. By the time this degree of architectural dis- tortion develops, hepatic steatosis has often disappeared. Steatohepatitis is an intermediate form of liver damage characterized by the appearance of focal hepatic inflam- mation and hepatocyte death on a background of hepatic steatosis. The inflammatory infiltrate includes polymor- phonuclear leukocytes as well as mononuclear cells and is often prominent around ballooned hepatocytes, which sometimes contain Mallory's hyalin. These foci of injury tend to predominate in acinar zone 3 and are variably associated with perisinusoidal, perivenular, or bridging fibrosis. Because all of these histologic features also occur in alcohol- or drug-induced fatty liver diseases, liver biopsy cannot reliably distinguish among the various causes of this entity. 1 Abnormal liver blood test values, particularly ami- notransferases and gamma glutamyl transpeptidase, often trigger the evaluation that leads to the diagnosis of NAFLD. However, no single blood test is specific for NAFLD. Moreover, because liver enzymes can be normal, at least intermittently, in patients with any given histo- logic stage of NAFLD, the presence of normal or near- normal aminotransferase values does not exclude signif- icant underlying liver damage. 2 Classically, however, patients with NAFLD have slightly elevated liver en- zyme values, deny excessive alcohol consumption, and have negative serologic tests for viral hepatitis, autoim- mune liver disease, and congenital causes of chronic hepatitis. The utility of this "operational" definition has been validated by several investigators who performed liver biopsies in otherwise-unselected groups of asymp- tomatic adults with mildly elevated liver enzyme values of uncertain etiology. Fatty liver was demonstrated in 30%-40% and steatohepatitis, with varying degrees of fibrosis, was seen in an additional 15%-30%. 3-6 Thus NAFLD accounts for about 70% of the cases of "crypto- genic" chronic hepatitis in the general population. The prevalence of NAFLD is even higher in select obese or diabetic populations, in which liver biopsy demonstrates Abbreviations used in this paper: BMI, body mass index; CT, com- puted tomography; NASH, nonalcoholic steatohepatitis; NAFLD, non- alcoholic fatty liver disease; NHANES, National Health and Nutrition Examination Survey; NMR, nuclear magnetic resonance; TNF, tumor necrosis factor. © 2002 by the American Gastroenterological Association 0016-5085/02/$35.00 doi:lO.1053/gast.2002.33573

Upload: others

Post on 23-Jun-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Nonalcoholic Fatty Liver Disease - Semantic Scholar€¦ · Nonalcoholic Fatty Liver Disease JEANNE M. CLARK, FREDERICK L. BRANCATI, and ANNA MAE DIEHL Departments of Medicine and

GASTROENTEROLOGY 2002;122:1649-1657

Nonalcoholic Fatty Liver Disease

JEANNE M. CLARK, FREDERICK L. BRANCATI, and ANNA MAE DIEHL Departments of Medicine and Epldemiology, The Johns Hopkins University, Baltimore, Maryland

N 'onalcoholic fatty liver disease (NAFLD) includes a spectrum of hepatic pathology that resembles al-

cohol-induced liver disease but develops in individuals who are not heavy drinkers. NAFLD is likely to be the most common cause of chronic liver disease in many countries and may also potentiate liver damage induced by other agents, such as alcohol, industrial toxins, and hepatotrophic viruses. The lack of specific and sensitive noninvasive tests for NAFLD limit reliable detection of the disease. Often, NAFLD is diagnosed presumptively when imaging studies suggest hepatic steatosis or when liver enzyme elevations are noted in overweight or obese individuals with no other identifiable reason for liver disease. NAFLD is strongly associated with insulin re- sistance and dyslipidemia; however, whether it is a cause or a consequence of these other conditions in not clear. The natural history of NAFLD is also uncertain. Never- theless, discrepancies between the apparently high prev- alence of NAFLD in the population and the generally low prevalence of clinically significant liver disease, as well as the relatively low representation of NAFLD pa- tients among liver transplant populations, have gener- ated considerable skepticism about the clinical impor- tance of NAFLD. The latter has reduced enthusiasm for aggressive diagnosis, treatment, and follow-up of pa- tients with NAFLD. This review of the clinical literature and selected basic information about NAFLD challenges some of these assumptions.

NAFLD--What Is It?

The diagnosis of NAFLD requires a combination of invasive and noninvasive tests. Liver biopsy is the most sensitive test for detecting and staging fatty liver disease. Fatty liver (hepatic steatosis) is the lesion at the most clinically benign end of the spectrum. Large (macro-) and small (micro-) vesicles of fat, predominately triglycer- ides, accumulate within hepatocytes without causing appreciable hepatic inflammation, liver cell death, or scarring. Cirrhosis is the lesion at the opposite end of the spectrum. By the time this degree of architectural dis- tortion develops, hepatic steatosis has often disappeared. Steatohepatitis is an intermediate form of liver damage characterized by the appearance of focal hepatic inflam-

mation and hepatocyte death on a background of hepatic steatosis. The inflammatory infiltrate includes polymor- phonuclear leukocytes as well as mononuclear cells and is often prominent around ballooned hepatocytes, which sometimes contain Mallory's hyalin. These foci of injury tend to predominate in acinar zone 3 and are variably associated with perisinusoidal, perivenular, or bridging fibrosis. Because all of these histologic features also occur in alcohol- or drug-induced fatty liver diseases, liver biopsy cannot reliably distinguish among the various causes of this entity. 1

Abnormal liver blood test values, particularly ami- notransferases and gamma glutamyl transpeptidase, often trigger the evaluation that leads to the diagnosis of NAFLD. However, no single blood test is specific for NAFLD. Moreover, because liver enzymes can be normal, at least intermittently, in patients with any given histo- logic stage of NAFLD, the presence of normal or near- normal aminotransferase values does not exclude signif- icant underlying liver damage. 2 Classically, however, patients with NAFLD have slightly elevated liver en- zyme values, deny excessive alcohol consumption, and have negative serologic tests for viral hepatitis, autoim- mune liver disease, and congenital causes of chronic hepatitis. The utility of this "operational" definition has been validated by several investigators who performed liver biopsies in otherwise-unselected groups of asymp- tomatic adults with mildly elevated liver enzyme values of uncertain etiology. Fatty liver was demonstrated in 30%-40% and steatohepatitis, with varying degrees of fibrosis, was seen in an additional 15%-30%. 3-6 Thus NAFLD accounts for about 70% of the cases of "crypto- genic" chronic hepatitis in the general population. The prevalence of NAFLD is even higher in select obese or diabetic populations, in which liver biopsy demonstrates

Abbreviations used in this paper: BMI, body mass index; CT, com- puted tomography; NASH, nonalcoholic steatohepatitis; NAFLD, non- alcoholic fatty liver disease; NHANES, National Health and Nutrition Examination Survey; NMR, nuclear magnetic resonance; TNF, tumor necrosis factor.

© 2002 by the American Gastroenterological Association 0016-5085/02/$35.00

doi:lO.1053/gast.2002.33573

Page 2: Nonalcoholic Fatty Liver Disease - Semantic Scholar€¦ · Nonalcoholic Fatty Liver Disease JEANNE M. CLARK, FREDERICK L. BRANCATI, and ANNA MAE DIEHL Departments of Medicine and

1650 CLARK ET AL. GASTROENTEROLOGY Vol. 122, No. 6

NAFLD in up to 90% of individuals with cryptogenic hepatitis. 7-9 However, these diagnostic criteria probably

underestimate the true prevalence of NAFLD. Not only do some patients with NAFLD have normal aminotrans-

ferase levels, but at least some patients with other types of liver disease also have NAFLD. The latter point is important, because NAFLD may influence the outcome

of those other diseases. For example, recent evidence suggests that steatosis and steatohepatitis are major in- dependent risk factors for cirrhosis in patients infected

with hepatitis C. 1°,~ Conversely, other liver diseases can influence the outcome of NAFLD. This has been dem- onstrated for NAFLD patients who also have genetic

hemochromatosis. Such individuals are significantly more likely to develop cirrhosis than NAFLD patients with a normal HFe genotype. 12 Therefore, positive tests

for viral hepatitis or hemochromatosis do not entirely exclude a diagnosis of NAFLD. Indeed, elevated serum

ferritin levels may be a clue that nonalcoholic steatohepa- titis (NASH) is present, because hyperferritinemia with

or without increased transferrin saturation has been de- scribed in many NASH patients who do not have C282Y or H63D mutations in the HFe gene) 3,~4

Abdominal imaging studies are often ordered instead of liver biopsy to confirm the clinical suspicion of NAFLD. This approach is rationalized because it avoids

the risks associated with an invasive procedure. However, the risk of significant bleeding or death from "blind"

percutaneous liver biopsy in patients with incidentally detected liver enzyme elevations is exceedingly rare, most likely far less than the figures derived from liver

biopsy populations that included patients with condi- tions that increase biopsy-related morbidity and mortal- ity, such as coagulopathy or liver tumor. In addition, the

potential drawbacks of limiting diagnostic procedures to noninvasive tests must be considered. Ultrasonography is commonly used to screen for fatty liver disease. A recent

study that correlated radiologic and histologic diagnoses in 24 healthy volunteers and 28 patients with elevated

liver enzyme values demonstrated that ultrasound detec- tion of fatty infiltration had a sensitivity of 67%, a specificity of 77%, a positive predictive value of 77%,

and a negative predictive value of 67%. Thus, relying on ultrasound to diagnose fatty liver disease gives an incor- rect diagnosis in 25% to 33% of patients. 15 One study

found computed tomography (CT) to be inferior to ul- trasound in diagnosing fatty liver, mostly because asso-

ciated hepatic iron overload produced a masking effect that decreased the sensitivity of CT scan) 6 However, in another study, when test objects containing variable

amounts of fat were scanned to generate a CT scan

Table 1. Comparison of the Expense and Limitations of Various Tests for Fatty Liver Disease

T e s t Approximate Cost Dm~tations

Ultrasound $420 Insensitive, nonspeciftc CT scan $650 Insensitive, nonspecific MRI $970 Nonspectftc Liver biopsy $415 Invastve

density calibration curve before patients with fatty livers were evaluated, an excellent correlation was seen between the hepatic fat content and liver-to-spleen density ratio. Thus calibrated CT scans might be useful in monitoring hepatic fat content, t7 Proton nuclear magnetic resonance (NMR) spectroscopy has also been validated as a reliable test for quantifying liver fat. Hepatic triglyceride con- tent assessed by proton NMR spectroscopy and by liver biopsy correlate almost perfectly. 18 Thus, the latter ap- proach seems to be the best noninvasive way for diag- nosing and quantifying liver fat. However, the expense of various imaging modalities is not trivial (Table 1), and none of these can distinguish simple steatosis from NASH or "uncomplicated" NASH from NASH with fibrosis. Therefore, liver biopsy remains the most cost- effective and sensitive test for diagnosing and staging NAFLD. t 9

NAFLD--Who Gets It?

The absence of blood tests, imaging modalities, or histologic findings that can distinguish NAFLD from alcohol- or drug-induced fatty liver disease forces clini- cians to rely heavily on other historical information, physical findings, and laboratory data to reach a final diagnosis of NAFLD. However, the success of this strat- egy rests on the validity of the assumptions made about "safe" levels of alcohol consumption, as well as the strength of the associations between NAFLD and gender, race, age, body mass index (BMI), and certain other medical conditions, such as dyslipidemia and insulin

resistance.

Prof i l ing for N A F L D

Much of the NASH literature emphasizes the prevalence of the disease in obese, middle-aged women who do not drink alcohol, although a few series call attention to the fact that the disease also occurs in mildly overweight men who abstain from alcohol. 2°,21 Because most of the reported cases of NASH have been white, there has been some speculation that nonwhite races might be somewhat protected from NAFLD despite having an increased risk for other obesity-related health problems. Given this background information, clinicians

Page 3: Nonalcoholic Fatty Liver Disease - Semantic Scholar€¦ · Nonalcoholic Fatty Liver Disease JEANNE M. CLARK, FREDERICK L. BRANCATI, and ANNA MAE DIEHL Departments of Medicine and

May 2002 NONALCOHOLIC FATTY LIVER DISEASE 1651

have been conditioned to favor NAFLD as the cause of abnormal liver enzyme values in patients who match the profile of the typical NASH patient and to look harder for other causes of liver disease in those who do not. However, it is important to acknowledge that this ap- proach might be flawed, because despite the ethnic di- versity of the general U.S. population, studies have often been limited to small, largely white, clinic-based samples influenced by selection bias related to referral patterns for abnormal liver tests. Given ethnic differences in obesity, type 2 diabetes, and dyslipidemias, all thought to be risk factors for NAFLD, it is conceivable that the previous small series have generated a somewhat distorted per- spective of the distribution of NAFLD in the general U.S. population.

To address this issue, we used the National Health and Nutrition Examination Survey (NHANES) III (con- ducted from 1988 to 1994 by the Centers for Disease Control) to assess the prevalence of NAFLD in the gen- eral U.S. population. = Data on 12,241 adults (69% of the entire study sample) who had complete information on liver test values and other causes of liver disease were analyzed. 2~

Elevated liver enzyme values were defined as any above-normal value of alanine aminotransferase (ALT), aspartate aminotransferase (AST), or ~/-glutamyltrans- ferase (GGT) (>30 U/L), which is similar to definitions used in other studies and based on generally accepted laboratory normal values. 24 A diagnosis of NAFLD was presumed if an individual had 1 or more elevated liver enzyme values and no evidence of an alternative expla- nation (e.g., negative testing for hepatitis B surface an- tigen and hepatitis C antibody, transferrin saturation < 50%, consumption of fewer than 2 alcoholic drinks per day for men and less than 1 drink per day for women).

Surprisingly, we found that in every age group, men were significantly more likely to have NAFLD than women. The possibility that female hormones protect against NAFLD was further supported by evidence that NAFLD is twice as common in postmenopausal women as in premenopausal women (odds ratio [OR], 2.05; 95 % confidence interval [CI], 1.43-2.94; P < 0.05) and post- menopausal women who receive hormonal replacement therapy are significantly less likely to have NAFLD than postmenopausal women who do not (OR, 0.69; 95% CI, 0 .48-0.99; P < 0.05). Also, contrary to what we ex- pected, even after adjusting for age and BMI, both non-Hispanic blacks and Mexican-Americans were sig- nificantly more likely to have NAFLD than non-His- panic whites. However, as expected, NAFLD increased with increasing BMI, even after adjusting for age and

Table 2. Adjusted Relative Odds of Cryptogenic Liver Enzyme Elevat ions (i.e., Presumed NAFLD) in NHANES III

Men Women BMI categories

(kg/m 2) OR 95% Cl OR 95% Cl

<18.5 0.81 0.38-1.72 1.16 0.62-2.16 18.5-24.9 1.00 Reference b 1.00 Reference b 25.0-29.9 2.12 a 1.69-2.66 2.07 a 1.61-2.68

30-34.9 3.71 a 2.62-5.25 2.67 a 1.94-3.66 35-39.9 5.02 a 3.04-8.30 3.92 a 2.83-5.44 >40.0 6.65 a 3.23-13.69 5.32 a 3.45-8.21

NOTE. Results are adjusted for age and ethnicity. OR, odds ratio; CI, confidence interval. ap < 0.05. ~Standard considered to be normal.

ethnicity (Table 2). Consistent with other evidence that central adiposity is a major risk factor for NAFLD, we found that both men and women with greater waist circumferences are more likely to have NAFLD. Type 2 diabetes mellitus is associated with a 2-5-fold increased risk of NAFLD. Based on these results, it appears that we must modify our perception about the "typical" NAFLD patient--"she" is actually a "he," and he might not be white.

Associated Condit ions

Obes i ty . As indicated earlier, obesity is strongly associated with NAFLD. However, it is equally clear that some morbidly obese individuals do not have NAFLD, because the prevalence of NAFLD reportedly ranges from 50% to 90% in various obese cohorts. 7,8 In the NHANES III population, only about 30% of obese men and 40% of obese women have NAFLD. 23 Moreover, other studies have already shown that NAFLD definitely occurs in nonobese individuals 24 and is particularly com- mon in patients with congenital or acquired lipodystro- phy, 26 marked by a generalized paucity of adipose tissue. These findings suggest that obesity and NAFLD are common consequences of some other underlying disor- der, or that obesity increases the risk of developing NAFLD after exposure to certain r'6ther insults.

With reference to the latter possibility, it is becoming apparent that obesity potentiates alcohol-related liver problems. For example, Bellentani et al. 4 found ultra- sound evidence of fatty liver in 46% of nonobese and 95% of obese heavy drinkers, demonstrating that obesity doubles the prevalence of alcohol-induced fatty liver disease. Another group has shown that obesity signifi- cantly increases the risk of alcohol-related cirrhosis. 27 Thus, obesity conceivably could lower the threshold for alcohol-related liver disease. If this is true, then levels of

Page 4: Nonalcoholic Fatty Liver Disease - Semantic Scholar€¦ · Nonalcoholic Fatty Liver Disease JEANNE M. CLARK, FREDERICK L. BRANCATI, and ANNA MAE DIEHL Departments of Medicine and

1652 CLARK ET AL. GASTROENTEROLOGY Vol. 122, No. 6

alcohol consumption considered "safe" for lean individ- uals might be dangerous for obese people.

One might even argue that some level of exogenous alcohol exposure is required to develop NAFLD. To evaluate this possibility, we reanalyzed the NHANES III data and looked at the prevalence of NAFLD in non- drinkers (i.e., those reporting fewer than 12 alcoholic drinks in their lifetime). 23 This subgroup, representing 14% of the NHANES III population, was more likely to be female, nonwhite, older, and to have type 2 diabetes. Nevertheless, among these teetotalers, the overall prev- alence of NAFLD is 21% (28% in men and 18% in women, P < 0.001). The pattern of associations with race, age, BMI, diabetes, and menopausal status is also nearly identical to that in the general population. There- fore, active exposure to alcoholic beverages is not re- quired to develop NAFLD.

Might obesity so sensitize individuals to alcohol-re- lated liver damage that they are injured by the small amounts of ethanol that are produced by their own intestinal flora? This is an interesting question given evidence that endogenous ethanol production is signifi- cantly increased in genetically obese, ob/ob mice that develop NAFLD spontaneously28 We found a positive correlation between breath ethanol concentrations and obesity in a small group of individuals who denied recent consumption of alcoholic beverages. 29 A group of Chil- ean investigators also presented preliminary evidence that malondialdehyde-acetaldehyde adducts, which form as a result of ethanol metabolism, 3° are increased in the livers of morbidly obese patients undergoing gastric bypass surgery. Thus some cases of NASH might really be alcoholic steatohepatitis (ASH).

If NAFLD can be induced by exposure to small amounts of alcohol, then it is conceivable that other toxins might also contribute to the pathogenesis of NAFLD. This possibility is supported by evidence that obesity increases the risk of liver enzyme elevations after occupational exposure to certain industrial chemi- cals. 31,32 Prescribed and over-the-counter medications might also promote NAFLD in susceptible individuals given that NALFD-related comorbidities (e.g., diabetes, dyslipidemia, hypertension) often require medical treat- ment. Indeed, our analysis of NHANES III demonstrates that both men and women with NAFLD were more likely to have visited physicians during the past year than were individuals who had normal liver enzymes (men, OR 1.02; 95% CI 1.01-1.04; women, OR 1.03; 95% CI 1.01-1.04; P < 0.05). 23 Women with NAFLD were also more likely to report their health as either fair or poor compared to women with normal liver enzymes (OR,

1.47; 95% CI, 1.10-1.96; P < 0.05). Therefore, it is likely that individuals with NAFLD might be taking more medications than individuals without NAFLD. To evaluate the role of potential hepatotoxins in NAFLD pathogenesis, we looked at the prevalence of NAFLD in the 35% of the NHANES III population who took no prescription medication or acetaminophen in the past month. NAFLD was present in 21% of this subgroup (28% in men, 10% in women). Again, NAFLD was found to be more common in persons who are older, nonwhite, and postmenopausal and have higher BMI, higher waist circumference, and type 2 diabetes. There- fore, although obesity might potentiate the hepatotoxic- ity of alcohol and other chemicals, the latter are not required for NAFLD to develop.

Dyslipidemia. By definition, hepatic lipid ho- meostasis is disordered in NAFLD. However, whether dyslipidemia is a cause or a consequence of NAFLD is not clear. Assy et a l ) 3 evaluated patients in the lipid clinic at their hospital and found that almost 2 of 3 of the patients had elevated liver enzyme values and that ultrasound findings revealed fatty liver disease in about half of the patients. Most of the patients with hypercholesterolemia had normal ultrasounds, whereas severe hypertriglyceri- demia and mixed hyperlipidemia increased the risk of fatty liver disease by 5-6- fo ld) 3 After adjusting for age, ethnicity, BMI, and type 2 diabetes, we found that both male and female NHANES III participants with a serum triglyceride level >200 mg/dL had about a 3-fold greater risk of having NAFLD than those with more normal triglyceride levels. In our analysis, a high-density li- poprotein (HDL) cholesterol level <35 mg/dL also al- most doubled the risk of NAFLD. 23 Taken together, these results confirm earlier reports documenting an association between certain dyslipidemias and NAFLD.

Insulin resistance and type 2 diabetes. Insulin resistance is strongly associated with conditions consid- ered to be risk factors for NAFLD, including obesity and certain dyslipidemias. Therefore, insulin resistance may play a fundamental role in the pathogenesis of these disorders. 3<35 Consistent with the possibility that insulin resistance plays a primary role in NAFLD pathogenesis, mild insulin resistance is very common at the earliest stages of NAFLD, and more severe insulin resistance (i.e., type 2 diabetes) correlates with more advanced stages of NAFLD. 35,36 Moreover, in mice that have been geneti- cally manipulated to overexpress lipoprotein lipase in the liver, hepatic insulin signaling is selectively inhibited and NAFLD develops. 37 This finding has major clinical implications. For example, it helps us understand why blood glucose levels, glycosylated hemoglobin concen-

Page 5: Nonalcoholic Fatty Liver Disease - Semantic Scholar€¦ · Nonalcoholic Fatty Liver Disease JEANNE M. CLARK, FREDERICK L. BRANCATI, and ANNA MAE DIEHL Departments of Medicine and

May 2002 NONALCOHOLIC FATTY LIVER DISEASE 1653

trations, and insulin-tolerance test results (all of which are heavily influenced by muscle insulin resistance) might be normal in some patients with NAFLD. This justifies our practice of using "guilt by association" with other insulin-resistant states to diagnose NAFLD.

Even more exciting, the molecular basis for insulin resistance has been discovered recently. Two groups have reported that IKK-beta, the kinase that activates nuclear factor beta, is chronically activated in ob/ob mice and several other murine models of insulin resistance) 8,39

The investigators used pharmacologic and genetic strat- egies to inhibit IKK-beta in these animals and proved that insulin resistance was abolished. The next challenge is to identify the genetic and environmental factors that promote the sustained activation of this enzyme. At least 1 clinically relevant IKK-beta activator is well recog- nized: tumor necrosis factor (TNF)-oL. Moreover, not only does TNF-oL activate IKK-beta, but also the result- ant induction of NF-KB promotes the transcription of TNF-o~, suggesting a self-reinforcing, positive-feedback mechanism that could maintain chronic insulin resis- tance) 8

NAFLD--Who Cares?

Prevalence

Diseases that are endemic or become epidemic generally attract our attention. For example, there is no doubt that hepatitis B is an important disease in China (with a prevalence of 8% t o 1 5 % ) . 40 Screening for dia- betes has become routine in the United States, because this disease develops in 5% of American adults. 41 Most general practioners are on the alert for hepatitis C, which affects about 2% of the U.S. population. 42 To judge

whether NAFLD should be placed on our national health-alert "radar screen," it is important to define its prevalence here. The prevalence of NAFLD in European and Japanese population-based studies is estimated to be 14%-21%)'43 However, no U.S. population-based study of NAFLD has been reported. Such studies are difficult because, as discussed previously, no single blood test, imaging study, or histologic parameter is 100% sensitive or specific for NAFLD.

Using the biopsy-validated operational criteria that define NAFLD as the most likely cause of abnormal liver enzymes in adults who have no other obvious cause of liver disease, our analysis of NHANES III suggests that NAFLD is by far the most common cause of elevated liver enzyme values in American adults. 23 We found that 27 % of adults have an elevated AST, ALT, or G G T level, and that 79% of these cases cannot be explained by other common causes of liver disease. Thus, NAFLD is the

most likely explanation for the elevated liver enzyme values observed in 23% of the general U.S. adult popu- lation. This represents approximately 31 million people. According to our definition, 31% of men and 16% of women have NAFLD.

Further studies will be required to determine whether or not 1 out of every 4 - 5 American adults actually has NAFLD. It is possible that this incidence rate is inflated because the cut-off values that we used to diagnose elevated liver enzyme values were unrealistically low. An analysis of liver enzyme values in NHANES III partici- pants demonstrates that none of these values are distrib- uted normally. 23 For example, the median value for AST

was 19 (range, 6-517) , the median ALT value was 13 (range, 1-486), and the median G G T value was 20 (range, 1-1342). Therefore, using a cut-off value of 30 identifies an individual whose liver enzyme values defi- nitely exceed the median. When the same data are rean- alyzed using more stringent criteria for elevated liver enzymes (i.e., --1.5 times normal), the overall prevalence of NAFLD decreases to 11% (14% in men, 8% in women), but the pattern and magnitude of the associa- tions discussed previously persist. Because liver biopsies were not performed in NAHNES III, confirmatory his- topathology is lacking. Moreover, tests to exclude un- usual causes of chronic hepatitis, such as oq-antitrypsin deficiency, Wilson's disease, and autoimmune hepatitis, were not performed. However, it is unlikely that these rare conditions account for most of the liver enzyme elevations in the general population. It also seems that alcohol or medication use cannot be blamed, because our subsidiary analyses in subgroups that did not drink or take medications showed findings similar to those in the entire population. Given that several studies have dem- onstrated NAFLD histology in most of their cases with elevated cryptogenic liver enzyme values and that the prevalence of NAFLD in the NHANES III population is very similar to the prevalence of NAFLD in population- based studies from Western Europe and Japan, it seems safe to conclude that NAFLD is the most common cause of liver enzyme elevations in American adults.

Clinical Impact

Debate persists over whether someone with inciden- tally detected liver enzyme elevations has a disease. Gener- ally, this is believed to be true for patients with alcoholic liver disease, chronic viral hepatitis, sclerosing cholangitis, or primary biliary cirrhosis. Studies of patients with these disorders suggest that 10%-30% develop cirrhosis) °,44-48 Virtually identical findings have been reported in the small series of NAFLD patients who have had a b iopsy . 25,49-51

Moreover, given the high incidence of NAFLD in the

Page 6: Nonalcoholic Fatty Liver Disease - Semantic Scholar€¦ · Nonalcoholic Fatty Liver Disease JEANNE M. CLARK, FREDERICK L. BRANCATI, and ANNA MAE DIEHL Departments of Medicine and

1654 CLARK ET AL. GASTROENTEROLOGY Vol. 122, No. 6

general population, it is not surprising that this disease is

now considered the most common cause of cryptogenic

cirrhosis. 52 In chronic liver diseases, liver-specific morbidity

and mortality is generally restricted to patients with cirrho-

sis. However, it is important to recall that the risk of

developing clinically significant complications of cirrhosis is

relatively low (about 25%-30% per decade), and liver-

related mortality rarely occurs in the absence of advanced

portal hypertension. Patients with well-compensated alco-

hol-induced cirrhosis have only an approximate 10% risk of

dying from liver disease in 10 years. 46 Matteoni et al. 51

reported an 11% 8-year liver-related mortality rate in their

small series of patients with NAFLD-induced cirrhosis,

suggesting that the natural history of NAFLD is likely to be

similar to that of other causes of chronic hepatitis.

Sometimes liver transplant registries are used to gauge

the clinical impact of a particular type of liver disease.

Using these criteria, NAFLD does not seem to be a very

serious type of liver disease; only 1%-2% of liver trans-

plant recipients carry a pretransplant diagnosis of

NAFLD. However, this conclusion probably should be

revised, because patients with cryptogenic cirrhosis con-

stitute a large subset of transplant recipients and many

patients with cryptogenic cirrhosis have NAFLD. Also,

NAFLD-related comorbidities, such as diabetes and dys-

lipidemias, predispose patient with NAFLD to cardio-

vascular and renal diseases that are generally contraindi-

cations for liver transplantation. Thus many patients

with advanced NAFLD may die without being listed for

liver transplantation. This possibility is supported by

evidence that cirrhosis is the 10th-leading cause of death

in middle-aged adults and that half of these liver-related deaths are attributed to cryptogenic cirrhosis. 47

There is also growing evidence that NAFLD contrib-

utes to the progression of other liver diseases. For exam-

ple, a recent study identified hepatic steatosis related to

visceral adiposity as the major independent risk factor for

fibrogenesis related to chronic hepatitis C infection,

whereas viral burden had absolutely no relevance to

disease progression. 1° Moreover, NAFLD is likely to

interfere with the therapy for other liver diseases given

the high prevalence of NAFLD in the general popula-

tion, the increasing reliance on living donor liver trans-

plantation as a treatment for end-stage liver disease, and

the fact that fatty liver grafts are at high risk for primary nonfunction after transplantation. 53 Finally, even in the

absence of liver-specific complications, individuals with

NAFLD were more likely to visit doctors, take medica-

tions, and to report poor health status than other indi-

viduals surveyed in NHANES III, suggesting that pa-

tients with incidentally noted liver enzyme elevations are not truly asymptomatic. 23

N A F L D - - W h a t Can W e Do?

The fact that no proven therapy for NAFLD exits is one of the most often-voiced rationalizations for avoid- ing aggressive diagnostic testing of patients with pre- sumed NAFLD. Without a doubt, more information about the natural history of NAFLD and the results of large, randomized prospective treatment trials for pa- tients with this disease is needed to guide future deci- sions. In the meantime, a number of reasonable treat- ment options are available.

Putat ive T r e a t m e n t s

Lifestyle modification. The preliminary results from a large, multicenter trial indicate that lifestyle modi- fications (e.g., diet and exercise) significantly reduce the risk of developing type 2 diabetes. 54 Given the strong associa- tion between insulin resistance and NAFLD, it is reasonable to recommend such lifestyle modifications as a treatment for NAFLD, although at this point it is not clear whether this approach will help individuals who have already developed diabetes or who have histologically advanced NAFLD. Sev- eral small studies have demonstrated that weight loss im- proves liver enzyme elevations in patient with NAFLD55,56; however, it is equally true that rapid, extreme weight loss, such as that induced by weight-reduction surgery, acceler- ates hepatic decompensation in some patients with NAFLD. 5v A more careful evaluation of the techniques used to induce weight loss might be informative because little is known about the relative importance of changing diet com- position as opposed to general caloric restriction. At least in rats and mice, evidence clearly indicates that diets with high sucrose or fat content are more likely to cause insulin resistance and hepatic steatosis than are equicaloric diets enriched with glucose or protein. 58-6° Also, because exercise is an important component of most successful weight loss programs and physical activity enhances muscle insulin sensitivity, 56,6' it is not clear whether merely increasing physical activity would provide the same benefits as dieting for patients with NAFLD.

Insulin-sensitizing medications. Mouse models of insulin resistance and fatty liver disease have shown that treatment with metformin or thiazolidinediones im- proves both conditions. 62,63 Caldwell et al. 64 reported

that treatment with troglitazone significantly improved liver enzyme abnormalities and somewhat improved he- patic histology in a small series of patients with NAFLD. Recently, metformin therapy was shown to significantly improve liver enzyme values and hepatic steatosis in another small group of patients with NAFLD. 65 How-

Page 7: Nonalcoholic Fatty Liver Disease - Semantic Scholar€¦ · Nonalcoholic Fatty Liver Disease JEANNE M. CLARK, FREDERICK L. BRANCATI, and ANNA MAE DIEHL Departments of Medicine and

May 2002 NONALCOHOLIC FATTY LIVER DISEASE 1655

ever, there has been some hesitancy about recommending either drug as a general therapy for NAFLD because of potential treatment-related toxicity. Troglitazone was withdrawn as a first-line therapy for type 2 diabetes because of rare, but potentially fatal, hepatotoxicity. 66 Phenformin, the parent compound of metformin, is known to cause life-threatening lactic acidosis; thus met- formin is contraindicated in patients with liver disease. 67 Nonetheless, results of the same diabetes prevention trial that demonstrated the efficacy of lifestyle modifications also indicated that metformin was well tolerated in that large population of obese, insulin-resistant subjects and significantly decreased the incidence of overt diabetes. 54 The safety and efficacy of troglitazone could not be evaluated, because that arm of the study was terminated prematurely because of concerns about potential hepato- toxicity. It remains plausible that second-generation thiazolidinediones, which appear to have less intrinsic hepatotoxicity, might be beneficial in treating NAFLD. Leptin might also have some role as a therapy for selected patients with NAFLD. In lipodystrophic mice that de- velop leptin deficiency caused by lipoatrophy, associated insulin-resistance and hepatic steatosis have been cured with leptin therapy, 68 suggesting that leptin might be helpful for lipodystrophic patients with NAFLD.

Antioxidants. A few trials have demonstrated that treatment with vitamin E improves liver enzyme abnormalities in patients and certain animal models with fatty liver disease. 69-72 Vitamin E is also well tolerated and cheap, making it a particularly attractive potential treatment for this common and indolently progressive disease. Other antioxidants, such as betaine, also appear to have some efficacy, v3,74 Perhaps the benefits of anti- oxidants are not so surprising now that the links between NAFLD and insulin-resistance 36 and between insulin resistance and the IKK-beta pathway 38,39 have been es- tablished. Oxidant--and inflammatory--stresses are known to activate the IKK-beta pathway38; therefore, treatments such as vitamin E that reduce oxidative stress and inflammation should inhibit IKK-beta activation. Interestingly, this may be a common mechanism of action for several agents thought to improve NAFLD. For example, studies in mice and cultured hepatocytes demonstrate that both metformin and thiazolidinediones inhibit the proinflammatory cytokine TNF-o~, 62,75,76 a potent activator of IKK-beta. 38

Ursodeoxycholic acid. A small study demon- strated that ursodeoxycholic acid improves liver enzyme abnormalities in NAFLD. 77 Although data supporting its efficacy in NAFLD is sparse, hepatologists are very familiar with this drug, which is commonly used to treat

Figure 1, Polymorphisms of genes that regulate inflammatory re- sponses and exposure to certain environmental factors might influ- ence the extent of IKKbeta activation and hence, the development of hepatic insulin resistance and NAFLD.

cholestatic liver diseases, and it has become a popular therapy for NAFLD. The mechanism for the putative beneficial effect of ursodeoxycholic acid in NAFLD is uncertain, and the results of at least one large ongoing trial should clarify whether or not more extensive scru- tiny of this agent is justified.

Lipid-lowering drugs. Given that NAFLD is a disorder of hepatic lipid homeostasis, it is certainly rea- sonable to consider lipid-lowering drugs as possible treatments for NAFLD. As discussed previously, hyper- triglyceridemia and reduced HDL cholesterol level are the types of dyslipidemias associated with NAFLD. A small trial of gemfibrizol demonstrated no appreciable effect on NAFLD. 78 The rationale for using agents that generally lower cholesterol as therapy for NAFLD is unclear, and these agents have been associated with some incidences of liver injury. Nevertheless, a preliminary report of promising effects of atorvastatin in a small group of patients with NAFLD suggests that 3-hydroxy- 3-methylglutaryl-coenzyme A reductase inhibitors may play some role in the treatment of this disease. 79 How- ever, currently, the routine use of statins to treat NAFLD cannot be recommended.

N A F L D ~ W h a t ' s N e x t ?

This is an exciting time for investigators inter- ested in NAFLD. The National Institutes of Health is assembling a multicenter network to conduct large nat- ural history and treatment trials for patients with NAFLD. Meanwhile, basic scientists from a number of disciplines are chipping away the mysteries of NAFLD pathogenesis. Soon, we may know enough about the

Page 8: Nonalcoholic Fatty Liver Disease - Semantic Scholar€¦ · Nonalcoholic Fatty Liver Disease JEANNE M. CLARK, FREDERICK L. BRANCATI, and ANNA MAE DIEHL Departments of Medicine and

1656 CLARK ET AL. GASTROENTEROLOGY Vol. 122, No. 6

genetic and environmental factors involved to be able to counsel individuals about their risk for NAFLD. Current data suggest that both "nature" (i.e., genetic control of inflammatory responses) and "nurture" (i.e., epigenetic causes of oxidative stress/inflammation) contribute to this type of liver disease (Figure 1). Hence, individuals who have inherited the "bad" tendency to have sustained inflammatory responses might be better off minimizing the consumption of alcohol or foods that stimulate cel- lular oxidant production and trigger inflammation or taking medications to improve their antioxidant/anti- inflammatory defenses, whereas others with "good" in- flammation-control genes can be reassured that they can safely enjoy these pleasures.

References 1. Ludwig J, wgg~ano RT, McGdl DB. Nonalcohohc steatohepat~tis:

Mayo Clinm expenences w~th a h~therto unnamed d~sease. Mayo Clin Proc 1980;55:342-348.

2. Yano E, Tagawa K, Yamaoka K, Mori M. Test validity of periodic liver function tests in a population of Japanese male bank em- ployees. J Clin Epidemiol 2001;54:945-951.

3. Bellentani S, Tinbelh C, Saccoccm G, et al. Prevalence of chronic liver disease in the general population of northern Italy: the Dyonisos study. Hepatology 1994;20:1442-1449.

4. Bellentani S, Saccoccio G, Masutti F, et al. Prevalence of and risk factors for hepatic steatosls in Northern Italy. Ann Intern Med 2000;132:112-117.

5. Ratziu V, Giral P, Charlotte F, et al. Liver fibrosis in overweight patients. Gastroenterology 2000;118:1117-1123.

6. Skelly MM, James PD, Ryder SD. Findings on liver biopsy to investigate abnormal hver function tests m the absence of d~ag- nostic serology. J Hepatol 2001;35:195-199.

7. Braillon A, Capron JP, Herve MA, Degott C, Quenum C. Liver in obesity. Gut 1985;26:133-139.

8. Nomura F, Ohnlshl K, Satomura Y, Ohtsuki T, Fukunaga K, Honda M, et al. Liver function in moderate obesity: study in 534 mod- erately obese subjects among 4613 male company employees. Int J Obes 1986;10 :349-354.

9. Silverman JF, O'Brien KF, Long S, et al. Liver pathology in mor- bidly obese patients with and without diabetes. Am J Gastroen- terol 1990;85:1345-1355.

10. Adinolfl LE, Gambardella M, Andreana A, Trlpodl MF, Utlll R, Rugglero G. Steatosls accelerates the progression of liver dam- age of chronm hepatitis C patients and correlates with specific HCV genotype and v,sceral obesity. Hepatology 2001 ;33 :1358 - 1364.

11. Ong J, Yuounossi ZM, Speer C, Olano A, Gramhch T, Boparal N. Chronic hepatitis C and superimposed nonalcoholic fatty liver disease. Liver 2001;21 :266-271.

12. George DK, Goldwurm S, MacDonald GA, et al. Increased hepatm iron concentration m nonalcohohc steatohepat~t~s ~s associated with increased fibrosis. Gastroenterology 1998;114:311-318.

13. Mendler MH, Turlin B, Moirand R, et al. Insulin resistance-associated hepatic iron overload. Gastroenterology 1999;117:1155-1163.

14. Fargion S, Mattioh M, Fracanzanl AL, et al. Hyperferritlnemla, iron overload, and multiple metabolic alterations identify patients at risk for nonalcoholic steatohepatitis. Am J Gastroenterol 2001; 96 :2448-2455 .

15. Gra~f M, Yanuka M, Baraz M, et al. Quantitative estimation of attenuation in ultrasound video images: correlation with histology in diffuse liver disease. Invest Radiol 2000;35 :319-324.

16. Mendler MH, Bouillet P, LeSidaner A, Lavoine E, Labrousse F, Sautereau D, Pdlegand B. Dual-energy CT in the diagnosis and quantification of fatty liver limited clinical value in comparison to ultrasound scan and single-energy CT, with special reference to iron overload. J Hepatol 1998;28:785-794.

17. Riccl C, Longo R, Giouhs E, et al. Nonlnvasive in vivo quantitative assessment of fat content in human liver. J Hepatol 1997;27: 108-113.

18. Szczepanlak LS, Babcock EE, Schlck FD, RL, Garg A, Burns T, McGarry JD, Setin DT. Measurement of intracellular trlglycende stores by H spectroscopy and validation in vivo. Am J Physiol 1999;276:E977-E989.

19. Bravo AA, Sheth SG, Chopra S. L~ver bmpsy. N Engl J Med 2001;344:495-500.

20. Sheth SG, Gordon FD, Chopar S. Nonalcoholic steatohepatltls. Ann Intern Med 1997;126:137-145.

21. Falck-Ytter Y, Younassl ZM, Marcheslni G, McCullough AJ. Clinical features and natural h~story of nonalcohohc steatos~s syn- dromes. Semm Liver D~s 2001;21:71-26.

22. National Center for Statistics. Plan and operation of the Third National Health and Nutrition Examination Survey, 1988-1994, vol 1, 1994 vital health statistics.

23. Clark JM, Brancati FL, Diehl AM. Nonalcoholic fatty liver disease and its correlates in the United States. In review.

24. Pratt DS, Kaplan MM. Evaluation of abnormal liver enzyme results in asymptomatic patients. N Engl J Med 2000;342:1266-1271.

25. Bacon B, Faravash M J, Janney CG, et al. Nonalcohohc steato- hepatitis: an expanded chnmal entity. Gastroenterology 1994; 107:1103-1106.

26. Powell EE, Searle J, Mortimer R. Steatohepat~bs associated with limb llpodystrophy. Gastroenterology 1989;97:1022-1024.

27. Giraud NS, Borotto E, Aubert A, Capron F, Chaput JC. Excess weight risk factor for alcoholic liver disease. Hepatology 1997; 25:108 -111.

28. Cope K, Risby T, Diehl AM. Increased gastrointestinal ethanol production in obese mice: implications for fatty liver disease pathogenesls. Gastroenterology 2000;119 :1340-1347.

29. Nair S, Cope K, Risby T, Diehl AM. Obesity and female gender increase breath ethanol concentration: potential implications for the pathogenesls of nonaicohohc steatohepatlt,s. Am J Gastro- enterol 2001;96 :1200-1204.

30. Rolla R, Vay D, Mottaran E, et al. Detection of circulating anb- bodies against malondialdehyde-acetaldehyde adducts in pa- tients with alcohol-reduced liver disease. Hepatology 2000;31: 878 -884 .

31. Hodgson M, Van Th~el DH, Goodman-Klein B. ObesW and hepa- totoxln risk factors for fatty liver disease. Br J Industr Med 1991 ;48 :690-695 .

32. Cotrim HP, Andrade ZA, Parana R, Portugal M, Lyra LG, Freitas LA. Nonalcoholic steatohepatitis: a toxic liver disease in industrial workers. Liver 1999;19:299--304.

33. Assy N, Kaita K, Mymm D, Levy C, Rosser B, Minuk G. Fatty infiltration of liver mhyperhpldemlc patients. Dig Dis Sci 2000; 45 :1929-1934 .

34. Cortez-Pinto H, Camllo ME, Baptlsta A, De Oliveira AG, De Moura MC. Nonalcoholic fatty hver: another feature of the metabolic syndrome? Chn Nutr 1999;18.

35. Marceau P, Iron S, Hould FS, Marceau S, Slmard S, Thung SN, Kral JG. Liver pathology and the metabolic syndrome X in severe obesity. J Clin Endocrinol Metab 1999;84:1513-1517.

36. Marchesm~ G, Bnzi M, Morselli-Labate AM, et al. Assoc~atmn of nonalcoholic fatty liver disease with insulin resistance. Am J Med 1999 ;107 :450-455 .

37. Kim JK, Fillmore J J, Chen Y, et al. Tissue-specific overexpression of lipoproteln lipase causes tissue-specific insulin resistance. Proc Natl Acad Sci U S A 2001;98:7522-7527.

38. Yuan M, Konstantopoulos N, Lee J, Hansen L, Li Z-W, Karin M,

Page 9: Nonalcoholic Fatty Liver Disease - Semantic Scholar€¦ · Nonalcoholic Fatty Liver Disease JEANNE M. CLARK, FREDERICK L. BRANCATI, and ANNA MAE DIEHL Departments of Medicine and

May 2002 NONALCOHOLIC FATTY LIVER DISEASE 1657

Shoelson SE. Reversal of obesity- and diet-reduced insulin resis- tance with salicylates or targeted disruption of IKKbeta. Science 2001;293:1673-1677.

39. Kim JK, Kim Y.I, Fdlmore J J, et al. Prevention of fat-reduced msuhn resistance by salicylate. J Chn Invest 2001;108 :437-446.

40. Ni YH, Chang MH, Huang LM, et al. Hepatitis B virus infection in children and adolescents in a hyperendemic area: 15 years after mass hepatitis B vaccination. Ann Intern Med 2001;135:796-800.

41. Moss SE, Klein R, Klein BE. Cause-specific mortality in a popu- lation-based study of diabetes. Am J Public Health 1991;81: 1158-1162 .

42. Alter M J, Kruszon-Moran D, Nainan OV, et al. The prevalence of hepatitis C virus infection in the United States: 1988-1994 . N Engl J Med 1999;341:556-562.

43. Nomura H, Kashiwagl S, Hayashi J, Kajiyama W, Tani S, Goto M. Prevalence of fatty liver in a general population of Okinawa, Japan. ,lpn J Med 1988;27:142-149.

44. Galambos J. Natural history of alcohohc hepatitis II1: histolog~c changes. Gastroenterology 1972;56:515-522.

45. Pares A, Caballena J, Brugera M. Histologic course of alcoholic hepatitis: influence of abstinence, sex, and extent of hepatic damage. J Hepatol 1986;2:33-38.

46. Gines P, Quintero E, Arroyo V, Teres J, Bruguera M, Rimola A, et al. Compensated c~rrhosis: natural history and prognostic fac- tors. Hepatology 1987;7:122-128.

47. Deaths and hospitalizations from chronic liver d~sease and c~r- rhosis--United States, 1980-1989 . MMWR Weekly 1993;41: 969-973 .

48. Poupon RY, Serfaty LD, Amonm M, Galulal G, Chretien Y, Cha- zouilleres O. Combination of steatosis and alcohol intake ~s the main determinant of fibrosis progression in patients w~th hepati- tis C. Hepatology 1999;30:406A.

49. Diehl AM, Goodman Z, Ishak KG. Alcohol-like liver disease in non- alcoholics: a chn~cal and h~stopathological comparison with alcohol- induced hver inJury. Gastroenterology 1989;95:1056-1060.

50. Powell EE, Cooksley GE, Hanson R, Searle J, Halliday RW, Powell LW. The natural history of nonalcoholic steatohepat~tls: a fol- low-up study of 42 patients followed for up to 21 years. Hepatol- ogy 1990 ;11 :74 -80 .

51. Matteoni C, Younoss~ ZM, McCullough A. Nonalcoholic fatty hver disease: a spectrum of clinical pathological severity. Gastroen- terology 1999;116:1413-1419.

52. Caldwell SH, Oelsner DH, lezzon~ JC, Hespenheide EE, Battle EH, Driscoll CJ. Cryptogen~c cirrhosis: clinical characterization and risk factors for underlying disease. Hepatology 1999;29:664-669.

53. Trotter JF. Thin chance for fat people (to become living donors). Liver Transpl 2001;7 :415-417.

54. Chamberlain J, DeMouy J. D~et and exercise dramatically delay type 2 diabetes; diabetes medication, metformin, also effective. U.S. Depart- ment of Health and Human Services Press Release 2001 8-8-01; http.%\www.hhs.go~newskpressk2001pressk20010808a.htmL

55. Palmer M, Schaffner F. Effect of weight reduction on hepatic abnormaht~es in overweight patients. Gastroenterology 1990;99: 1408-1411 .

56. Ueno T, Sugawara H, Sujaku K, et al. Therapeutic effects of restricted diet and exercise in obese patients w~th fatty liver. J Hepatol 1997;27:103-107.

57. Capron JP, Delamarre M, Dupas JL, Bradlon A, Degott C. Fasting in obesity: another cause of liver inJUry with alcoholic hyaline? Dig Dis Sci 1982;54 :374-377.

58. Nowkoff PM. Fatty liver induced in Zucker "fatty" (if) rats by a semisynthetic diet rich in sucrose. Proc Natl Acad Sci U S A 1977;74 :3550-3554.

59. Poulsom R. Morphological changes of organs after sucrose or fructose feeding. In: Progress m biochemical pharmacology, vol. 21. Basel: Karger, 1986 :104-134 .

60. Akiyama T, Tachlbana I, Shirohara H, Watanabe N, Otsukl M.

High-fat hypercaloric diet induces obesity, glucose intolerance and hyperlipidemia in normal adult male W!star rat. Diabetes Res Clin Pract 1996;31:27-35.

61. Moiler DE, Flier JS. Insulin resistance: mechanisms, syndromes, and implications. N Engl J Med 1991;325:938-948.

62. Lm HZ, Yang SQ, Kujhada F, Ronnet G, Diehl AM. Metformin reverses nonalcoholic fatty liver disease in obese leptm-defic~ent mice. Nat Med 2000;6 :998-1003.

63. Kakuma T, Lee Y, Higa M, Wang ZW, Pan W, Shimomura I, Unger R. Leptin, troglitazone, and the expression of sterol regulatory element binding proteins in liver and pancreatic islets. Proc Natl Acad Sci U S A 2000 ;97 :8536 - -8541 .

64. Caldwell SH, Hespenheide EE, Redlck JA, et al. A pilot study of a thiazolidinedione, troglitazone, in nonalcoholic steatohepatitis. Am J Gastroenterol 2001;96 :519-525.

65. Marches~ni G, Bnzl M, Bianchi G, Tomasselli S, Zoll M, Mel- chlonda N. Metformin in non-alcoholic steatohepatit~s. Lancet 2001;358 :893-894.

66. Kohlroser J, Mathai J, Reichheld J, Banner BF, Bonkovsky HL. Hepatotoxicity due to troghtazone: report of two cases and rewew of adverse events reported to the United States Food and Drug Administration. Am J Gastroenterol 2000;95:272-276.

67. Stang M, Wysowski DK, Butler-Jones D. Incidence of lactic aci- dosis in metformm users. D~abetes Care 1999;22:925-927.

68. Shimomura I, Hammer RE, Ikemoto S, Brown MS, Goldstem JL. Leptin reverses insuhn resistance and diabetes melhtus m mice with congenital lipodystrophy. Nature 1999;401:73-76.

69. Newbeme PM. Effect of two synthetic antioxidants, vitamin E and ascorbic acid, on the choline-deficient rat. J Nutr 1969;97:219-231.

70. de la Maza MP, Petermann M, Bunout D, Hirsch S. Effects of long-term vitamin E supplementation in alcohohc cirrhot~cs. ,l Am Cell Nutr 1995;14:192-196.

71. Nanji AA, Yang EK, Fogt F, Sadrzadeh SM, Dannenberg AJ. Me- dium chain triglycerides and vitamin E reduce the severity of established expenmental alcoholic liver disease. J Pharmacol Exp Ther 1996 ;277 :1694-1700.

72. Lavine JE. V~tamin E treatment of non-alcohohc steatohepatitis in children: a pilot study. J Pediatr 2000;136:734-738.

73. Barak A J, Beckenhauer HC, Badakhsh S, Tuma DJ. The effect of betaine in reversing alcoholic steatosis. Alcohol Chn Exp Res 1997;21 :1100-1102.

74. Abdelmalek MF, Angulo P, Jorgensen RA, et al. Beta~ne, a promising new agent for patients with nonalcohohc steatohepatitis: results of a pilot study. Am J Gastroenterol 2001;96:2711-2717.

75. Purrello F, Gullo D, Buscema M, Pezzino V, Vignen R, Goldfine ID. Metformm enhances certain msuhn actions in cultured rat hepa- toma cells. Diabetologia 1988;31:385-389.

76. Solomon SS, Mishra SK, Cwik C, Rajanna B, Postlethwaite AE. Pioglitazone and metformin reverse insuhn resistance induced by tumor necrosis factor-alpha in hver cells. Horm Metab Res 1997; 29 :379-382 .

77. Laurin J, Lindor KD, Cnppin JS, Gossard A, Gores G J, Ludwig J, et al. Ursodeoxychohc acid or clofibrate in the treatment of non- alcohol-induced steatohepatitis: a pilot study. Hepatology 1996; 23 :1464-1467 .

78. Basaranoglu M, Acbay O, Sonsuz A. A controlled trial of gemfi- brozil in the treatment of patients with noalcoholic steatohepati- t~s. J Hepatol 1999;31:384.

79. Harlander JC, Kwo PY, Cummings OW. Atorvastin for the treat- ment of NASH. Gastroenterology 2001;120A:544:2767.

Received January 3, 2002. Accepted January 21, 2002. Address requests for reprints to: Anna Mae Diehl, M.D., 912 Ross

Building, The Johns Hopkins University, 720 Rutland Street, Baltimore, Maryland 21205. e-mail: [email protected]; fax: (410) 955-9677.