Nonalcoholic Fatty Liver Disease (NAFLD): a Comprehensive Review William B

JOURNAL OF INSURANCE MEDICINE Copyright ᮊ 2004 Journal of Insurance Medicine J Insur Med 2004;36:27±41

REVIEW

Nonalcoholic Fatty Liver Disease (NAFLD): A Comprehensive Review William B. Salt II, MD

Nonalcoholic fatty liver disease (NAFLD) is defined as fatty infiltra- Address: The Ohio State University, tion of the liver exceeding 5% to 10% by weight. It is a spectrum of Columbus, Ohio; e-mail: disorders ranging from simple fatty liver (steatosis without liver in- [email protected]. jury), nonalcoholic steatohepatitis (steatosis with inflammation), and Correspondent: William B. Salt II, fibrosis/cirrhosis that resembles alcohol-induced liver disease but MD; Ohio Gastroenterology Group, which develops in individuals who are not heavy drinkers. NAFLD Clinical Associate Professor in Med- is likely the most common cause of chronic liver disease in many icine. countries. NAFLD may also potentiate liver damage induced by other agents, such as alcohol, industrial toxins and hepatatrophic vi- Key words: Nonalcoholic fatty liver ruses. disease (NAFLD), nonalcoholic stea- The lack of specific and sensitive noninvasive tests for NAFLD tohepatitis (NASH), insulin resis- limits reliable detection of the disease. It is often diagnosed on a tance syndrome, liver transaminas- presumptive basis when liver enzyme elevations are noted in over- es, liver biopsy, life insurance. weight or obese individuals without identifiable etiology for liver disease, or when imaging studies suggest hepatic steatosis. NAFLD is now considered to be a component of the insulin resistance syndrome (metabolic syndrome X). Controversy exists relative to optimal recognition, diagnosis and management of these conditions, and treatment recommendations are evolving.

atty liver disease that develops in the ab- THE SPECTRUM OF NAFLD Fsence of alcohol abuse is recognized in- creasingly as a major health burden. Esti- The diagnosis of NAFLD is based upon mates based on imaging and autopsy studies noninvasive and invasive tests, with liver bi- suggest that about 20% to 30% of adults in opsy being the most definitive modality. He- the United States and other Western countries patic steatosis is at the lower end of the spec- have excess fat accumulation in the liver. Ap- trum of NAFLD clinical severity. Predomi- proximately 10% of these individuals, or fully nantly large (macro-) and occasionally small 2% to 3% of adults, are estimated to meet (micro-) vesicles of fat, predominately triglyc- current diagnostic criteria for nonalcoholic erides, accumulate within hepatocytes with- steatohepatitis (NASH). Sustained liver injury out causing appreciable hepatic inflamma- leads to progressive fibrosis and cirrhosis in tion, liver cell death, or scarring. At the mid- up to one third of those with NASH, which range of the severity spectrum is steatohep- may be a cause of cryptogenic cirrhosis. atitis (nonalcoholic steatohepatitis, or NASH), NASH is now a significant health issue for which is an intermediate form of liver dam- obese children, leading to cirrhosis in some. age superimposed upon a background of he-

27 JOURNAL OF INSURANCE MEDICINE

Table 1. Nomenclature of Fatty Disorders of the cleus to the edge of the cell. The original de- Liver scriptions of steatohepatitis included the presence of Mallory bodies, ballooning de- Histologic nomenclature generation, predominant lobular neutrophilic 1. Fatty liver (grade) inflammation, and Rappaport zone III peri- 2. Steatohepatitis sinusoidal fibrosis. Now it is realized that a. Grade: degree of necroin¯ammatory activity b. Stage: degree of ®brosis only some of these features may be present in a given patient. Mallory bodies are less frequently seen in NASH compared with alcoholic steatohepatitis and may even be absent. patic steatosis. Steatohepatitis is characterized Also, in many individuals, atypical features by the appearance of focal hepatic inflam- (eg, predominantly lymphocytic inflamma- mation and hepatocyte necrosis and death. At tion or portal fibrosis) are present. It is not the highest end of the severity spectrum is known whether these histologic patterns rep- cirrhosis. By the time this degree of architec- resent different stages of steatohepatitis or tural distortion develops, hepatic steatosis separate clinicopathologic conditions with has often disappeared. Because all of these overlapping histologic expression. Despite an histologic features also occur in alcohol- or attempt to standardize the histologic diag- drug-induced fatty liver diseases, liver biopsy nostic criteria for steatohepatitis during a cannot reliably distinguish among the vari- workshop at the National Institutes of Health, ous causes of this entity. no consensus exists on this subject. The best definition of NASH remains un- NOMENCLATURE OF FATTY settled, since there is significant diversity of DISORDERS OF THE LIVER opinion among experts regarding the neces- A complete diagnosis of fatty liver disease sity and character of specific findings. A pro- (ideally) should define the histology, includ- posed NAFLD classification system correlates ing the grade (severity) and stage (degree of certain histologic features with the long-term fibrosis) of the disease as well as its clinical prognosis. In 1999, researchers introduced association (Tables 1 and 2). the 4-tiered Matteoni system (Table 3) to de- 1 Traditionally, fatty disorders of the liver scribe the stages of NASH. have been classified as alcoholic or nonalco- Classes 3 and 4 NAFLD in Matteoni’s sys- holic. NAFLD includes both nonalcoholic fat- tem are similar and might be considered as a ty liver and NASH. NAFLD is associated single group constituting NASH. These clas- with numerous etiologies (Table 2). The term ses describe NAFLD that has progressed to ‘‘nonalcoholic fatty liver disease’’ renders the the far more serious NASH, in which varying condition heterogeneous in terms of etiology, degrees of fibrosis and/or cirrhosis develop natural history, as well as response to thera- and balloon cells or Mallory’s bodies appear. py. This makes the condition harder to study Class 2 NAFLD is more controversial; it may and is therefore unsatisfactory. There is cur- be benign and includes relatively more men rently no consensus on the best way to clas- often with a normal body mass index. sify fatty disorders of the liver. In another study, Brunt et al scored a total of 10 findings to develop a grading and staging system for NASH.2 These included he- HISTOLOGIC CRITERIA FOR THE patic macrovesicular steatosis, hepatocellular DIAGNOSIS OF FATTY LIVER AND ballooning, intra-acinar inflammation, portal STEATOHEPATITIS tract inflammation, Mallory’s hyaline, acido- The principal histologic feature of NAFLD phil bodies, glycogen nuclei, lipogranulomas, is the presence of macrovesicular fatty change and hepatocellular iron. Each of these was in hepatocytes with displacement of the nu- scored separately. Three parameters of he-

28 SALT—NONALCOHOLIC FATTY LIVER DISEASE

Table 2. Conditions Associated With Steatohepatitis19

All conditions except alcoholism are usually referred to as nonalcoholic steatohepatitis (NASH). 1. Alcoholism 2. Insulin resistance a. Syndrome X i. Obesity ii. Diabetes iii. Hypertriglyceridemia iv. Hypertension b. Lipoatrophy c. Mauriac syndrome 3. Disorders of lipid metabolism a. Abetalipoproteinemia b. Hypobetalipoproteinemia c. Andersen's disease d. Weber-Christian syndrome 4. Total parenteral nutrition 5. Severe weight loss a. Jejunoileal bypass b. Gastric bypass (much less common than after jejunoileal bypass) c. Severe starvation 6. Iatrogenic/drugs (see Secondary Causes of Steatosis, Table 5) 7. Refeeding syndrome 8. Toxic exposure a. Environmental b. Workplace

Table 3. Matteoni's Classi®cation of the Stages of havior regarding lifestyle) contribute to NAFLD/NASH1 NAFLD. The pathogenesis of NAFLD has been hy- Class 1: Simple steatosis without in¯ammation or ®bro- pothesized to be a 2-stage process. In the first sis stage, fatty acids accumulate in hepatocytes Class 2: Steatosis with lobular in¯ammation but without leading to steatosis. The development of ste- ®brosis Class 3: Additional presence of ballooned hepatocytes atosis is associated with obesity, particularly Class 4: Presence of either Mallory's hyaline or ®brosis central abdominal adiposity, and insulin resistance. Hepatic steatosis predisposes to liver injury in the second stage, which is charac- patic fibrosis were scored: perisinusoidal fi- terized by necrosis, inflammation and fibro- brosis, portal fibrosis, and bridging fibrosis. sis. It has been proposed that the primary Based on these, the necroinflammatory activ- mediator of this second stage is oxidative ity was graded as mild, moderate or severe stress. Pro-oxidants, such as iron, would be (Table 4). expected to worsen liver injury. Conversely, antioxidants, such as certain vitamins and minerals, would be expected to protect PATHOGENESIS OF NAFLD against liver injury. The pathogenesis of NAFLD has not been clearly established. Both ‘‘nature’’ (ie, genetic Stress Response control of inflammatory responses) and ‘‘nurture’’ (ie, epigenetic causes of oxidative Insulin resistance (metabolic syndrome X) stress/inflammation and health-related be- and NAFLD are related to ‘‘stress.’’ Dr. Bruce

29 JOURNAL OF INSURANCE MEDICINE

Table 4. Grading and Staging of NAFLD According to Brunt et al2

Grading Macrovesicular steatosis Grade 0: None Grade 1: Up to 33% Grade 2: 33%±66% Grade 3: Ͼ66% Necroin¯ammatory activity Grade 1: Mild Steatosis up to 66%, occasional ballooned hepatocyte (mainly zone III), scattered intra-acinar neutrophils (PMN) Ϯ lymphocytes, no or mild portal in¯ammation Grade 2: Moderate Steatosis of any degree, obvious zone III ballooning degeneration, intra-acinar PMNs, zone III perisinusoidal ®brosis may be present, mild to moderate, portal and intra-acinar in¯ammation Grade 3: Severe Panacinar steatosis, widespread ballooning, intra-acinar in¯ammation, PMNs associated with ballooned hepatocytes, mild to moderate portal in¯ammation Staging Stage 1: Zone III perisinusoidal/pericellular ®brosis, focally or extensively present Stage 2: Zone III perisinusoidal/pericellular ®brosis with focal or extensive periportal ®brosis Stage 3: Zone III perisinusoidal/pericellular ®brosis and portal ®brosis with focal or extensive bridging ®brosis Stage 4: Cirrhosis

S. McEwen from Rockefeller University has mulate, and overexposure to and/or misman- developed a new way of conceptualizing and agement of mediators of neural, endocrine, understanding the stress response.3 When and immune stress can have adverse effects stressors (triggers) represent a real or per- on various organ systems leading to disease. ceived threat to the neurobiological balance Dysregulation of the central stress re- (homeostasis) of the person, physiological sponse system related to allostatic load con- and behavioral responses are triggered in or- tributes to the development of a variety of der to achieve adaptation and survival in the diseases and illnesses including hyperten- short run (ie, allostasis, or ‘‘good stress’’). The sion, atherosclerosis, central obesity, diabetes, protective and restorative allostatic processes and the insulin resistance syndrome (meta- of the body are mediated through the auto- bolic syndrome X).20 Gastrointestinal diseases nomic nervous system, the hypothalamic-pi- and illnesses adversely impacted by allostatic tuitary-adrenal (HPA) axis and the cardiovas- load include non-alcoholic fatty liver disease cular, metabolic and immune systems. How- (also associated with the insulin resistance ever, over longer periods of time, these same syndrome), gastroesophageal reflux disease response systems that are designed to protect (GERD), peptic ulcer disease, inflammatory and restore can be damaging and can cause bowel disease and functional gastrointestinal or exacerbate symptoms, disease and illness disorders (FGID). (ie, allostatic load, or ‘‘bad stress’’). See Fig- ure 1. NATURAL HISTORY OF NAFLD The perception of stress is influenced by one’s experiences, genetics and behavior. Only limited natural history data are avail- When the brain perceives an experience as able concerning the spectrum of histologic le- stressful, physiologic and behavioral respons- sions seen in macrovesicular fatty disorders es are initiated leading to allostasis and ad- of the liver not associated with the use of al- aptation. Over time, allostatic load can accu- cohol. It is generally believed that there are

30 SALT—NONALCOHOLIC FATTY LIVER DISEASE

Figure 1. The Stress Response and Development of Allostatic Load (McEwen B. N Engl. J Med. 1998;338:171–179. Republished with permission. Copyright ᭧ 1998 Massachusetts Medical Society. All rights reserved.)

not well defined, it seems to be determined by the severity of liver damage. At this time, few patients have been observed prospective- ly in order to document the natural history of NASH, which is generally considered to be a clinically stable disorder with a much better prognosis than alcoholic hepatitis. In 5 series, 54 of 257 patients with NAFLD underwent liver biopsy during an average follow-up of Figure 2. Stages in the Progression of NAFLD. 3.5 to 11 years. Twenty-eight percent had progression of liver damage, 59% had essen- tially no change, and 13% had improvement several distinct histologic states in the natural or resolution of liver injury. Progression from history of these disorders that indicate pro- steatosis to steatohepatitis and to more ad- gression of the lesion (Figure 2). These in- vanced fibrosis or cirrhosis has been recog- clude a fatty liver alone, steatohepatitis, stea- nized in several cases. Some of the few deaths tohepatitis with fibrosis, and eventually cir- that occurred among the 257 patients were rhosis. It has also been noted that fatty liver-related, including one from hepatocel- change may disappear following develop- lular cancer. Thus, many patients with ment of cirrhosis. NAFLD have a relatively benign course, Cross-sectional studies of nonalcoholic fat- whereas in some others, the disease progress- ty liver indicate that most subjects have fatty es to cirrhosis and its complications.9 liver alone. It is currently believed that it is At the time of initial presentation, about rare for such patients to progress to steato- 30%–40% of patients with NASH have ad- hepatitis or steatosis with fibrosis over time. vanced fibrosis, whereas 10%–15% have es- In a recent study, only 2.9% of 546 liver-trans- tablished cirrhosis. By multivariate analysis, plantation procedures performed in a single increasing age, obesity, and diabetes were center were for end-stage steatohepatitis.4 noted to be independent predictors of bridg- Even though NAFLD is common, only a mi- ing fibrosis or cirrhosis. Another recent study nority of patients will require liver transplan- has confirmed the relationship between the tation. degree of obesity and the likelihood of ad- Although the natural history of NAFLD is vanced fibrosis.

31 JOURNAL OF INSURANCE MEDICINE

Patients found to have pure steatosis on liv- and 20 grams per day in women, while others er biopsy seem to have the best prognosis have utilized a cut-off level of 10 grams of within the spectrum of NAFLD. Features of ethanol per day or less. One study has re- steatohepatitis or more advanced fibrosis are ported that limited alcohol intake is protec- associated with a worse prognosis. With the tive against NASH (as well as diabetes), and exception of the histologic features discussed there is evidence for health benefits derived here, there are no clinical or laboratory fea- from modest alcohol ingestion.9 The thresh- tures that can predict progression in any giv- old issue will not be easily resolved. In a re- en patient. cent summary of NASH, a reasonable com- According to Arun J. Sanyal, MD, chairman promise was proposed: 20 grams of ethanol of the Division of Gastroenterology and Nu- per day. This cut-off level is well below the trition at Virginia Commonwealth University traditional threshold for alcohol-induced liver School of Medicine, Classes 1 and 2 NAFLD disease (40 to 60 grams of ethanol per day). appear to be stable over time and even re- Thus, it is generally believed that a fatty liver versible with weight loss, diet modification, does not develop with alcohol consumption and for diabetics sound disease manage- levels Ͻ20 g/day. ment.5 The long-term prognosis for patients Determination of an individual’s alcohol with Classes 1 and 2 NAFLD is usually good, consumption is easier said than done. As- especially if the underlying causes are ad- sessment of the amount of alcohol consumed dressed. Early-stage NAFLD could conceiv- from that reported by patients is notoriously ably progress to NASH, but the transition is inaccurate. Questioning of family members probably unusual, while the speed of liver may be useful in some instances. deterioration remains unknown. Dr. Sanyal The shortcomings of this approach have suspects that once NASH emerges bridged led to the development of direct and surro- with fibrosis or cirrhosis, mortality rates gate markers of alcohol consumption. These reach as high as 20% at 10 to 15 years follow- include serum gamma-glutamyltransferase ing diagnosis; however, he cautions that no levels, mean corpuscular volume, amino- firm data back up this generalized estimate. transferase (AST) levels, AST/ alanine ami- A shortcoming of studies on the natural notransferase (ALT) ratio, mitochondrial AST history of NAFLD is that patients who sub- levels, and desialylated transferrin levels. The sequently underwent liver biopsy and long- first 4 tests are both widely available and rel- term follow-up were highly selected and may atively inexpensive. Unfortunately, these lack have altered health-related behavior. Popula- both sensitivity and specificity, and neither tion-based studies will be necessary to better the negative nor positive predictive values are define the natural history of NAFLD. high enough to be clinically useful. The clinical utility of the other markers has not been DIFFERENTIAL DIAGNOSIS OF NAFLD examined outside of a few limited clinical studies. Alcohol and NAFLD

‘‘Nonalcoholic’’ is inherent to deﬁning Secondary Causes of Steatosis NAFLD and NASH, since it refers to a threshold at which the liver abnormality be- NAFLD is differentiated from steatosis, comes alcohol related. The distinction of al- with or without hepatitis, resulting from sec- coholic vs nonalcoholic is not sharp and is ondary causes, since these conditions have controversial. There are no published and different pathogeneses and clinical outcomes. universally accepted threshold levels of alco- Steatosis may be either macrovesicular (main- hol intake that separate alcoholic fatty liver ly related to imbalance in the hepatic synthe- disease from NAFLD. Many centers accept sis and export of lipids) or microvesicular up to 20–40 grams of ethanol per day in men (mainly related to defects in mitochondrial

32 SALT—NONALCOHOLIC FATTY LIVER DISEASE function). Rarely, phospholipidosis occurs When symptoms occur, they are usually non- (mainly by accumulation of phospholipids in specific. Fatigue is probably the most com- lysosomes). See Table 5. monly reported symptom and does not correlate well with the severity of the histologic COEXISTING DISORDERS lesion. Another common symptom is right upper quadrant discomfort, which is typical- Most liver disorders are associated with ably of a vague, nondescript aching character. normalities of biochemical liver tests and re- A smaller fraction of patients experience quire differentiation from NAFLD. Further- symptoms indicative of more serious liver more, NAFLD is so prevalent that it often co- disease and may develop pruritus, anorexia exists with other liver disorders. It is partic- and nausea. ularly important to identify benign and malignant liver lesions, drug-induced hepa- Signs titis, chronic viral hepatitis B and C, hemochromatosis, and autoimmune hepatitis, since There are no pathognomonic signs of potentially effective therapies are available. NASH. Obesity is the most common abnor- The presence of hepatitis C was originally mality on physical examination and is pre- believed to constitute an exclusion criterion sent in 30%–100% of patients in various cross- for the diagnosis of NASH. However, some sectional studies. The most common finding with hepatitis C clearly have histologic evi- of liver disease is hepatomegaly, which has dence of a classic steatohepatitis rather than been reported in up to 50% of subjects in dif- the predominantly portal lymphocytic infil- ferent studies. A smaller percentage of pa- trate with mild to moderate steatosis seen in tients have stigmata of chronic liver disease. hepatitis C. In such cases, the presence of 2 Of the various stigmata known, the presence diagnoses (ie, hepatitis C and NASH) may be of spider nevi and palmar erythema are most considered. It is important to exclude genetic common. Jaundice, edema, asterixis, and hemochromatosis if iron abnormalities are signs of portal hypertension occur in those identified. (See Liver Enzymes: Ferritin sec- with advanced cirrhosis. Muscle wasting may tion) occur as the liver disease becomes more advanced but is often underestimated due to CLINICAL ASSESSMENT edema and preexisting obesity. Symptoms LIVER ENZYMES As with many other types of chronic liver Aminotransferases disease, most patients with NAFLD in cross- sectional studies are asymptomatic. The liver Clinical suspicion of a diagnosis of NAFLD disease is either discovered incidentally dur- is most commonly triggered by the discovery ing routine laboratory examination or work- of abnormal liver blood test values, particu- up of conditions such as hypertension, dia- larly aminotransferases (AST and ALT). betes, or morbid obesity. Elevated ALT levels However, the presence, degree, and pattern of or sonographic evidence of fatty liver is some- aminotransferase elevation are nonspecific times noted during workup of suspected and do not provide an etiologic diagnosis. gallstone disease. Even when the index of suspicion for NASH Only limited data on symptomatology are is high (eg, in an obese, diabetic individual), available from longitudinal studies, and both the aminotransferase levels do not distin- the likelihood of developing symptoms over guish between fatty liver alone and NASH. time as well as the predictors of future de- Liver enzymes can be normal, at least in- velopment of symptoms are not known. Most termittently, in those with any given histolog- patients with NAFLD are asymptomatic. ic stage of NAFLD. Thus, the presence of nor-

33 JOURNAL OF INSURANCE MEDICINE

Table 5. Secondary Causes of Steatosis19

Nutritional (predominantly mascrovesicular steatosis) Protein-calorie malnutrition Starvation Total parenteral nutrition Rapid weight loss Bariatric surgery Drugs This is a partial list; some drugs also produce in¯ammation. Drug-induced steatosis may have no clinical consequences (eg, corticosteroids) or can result in cirrhosis (eg, methotrexate and amiodarone). Glucocorticoids Synthetic estrogens Aspirin Calcium-channel blockers (weak association) Amiodarone (hepatic phospholipidosis) Tamoxifen Tetracycline Methotrexate Perhexiline maleate (hepatic phospholipidosis) Valproic acid Cocaine Antiviral drugs -Zidovudine -Didanosine -Fialuridine Metabolic or genetic disorders (rare) Lipodystrophy Dysbetalipoproteinemia Weber-Christian disease Wilson's disease Wolman's disease Cholesterol ester storage disease Acute fatty liver of pregnancy Other In¯ammatory bowel disease Small bowel diverticulosis with bacterial overgrowth Human immunode®ciency virus infection Environmental hepatotoxins Phosphorus Petrochemicals Toxic mushrooms Organic solvents Bacillus cereus toxins

mal or near-normal aminotransferase values vation cannot be used as a predictive factor. does not exclude significant underlying liver ALT values do not correlate with the degree damage. of steatosis or fibrosis, and significant liver The degree of aminotransferase enzyme el- disease may exist with liver enzymes in the evation is usually mild and is most typically normal range among NAFLD patients. less than 1.5 times normal, ranging up to 4 Although the ALT levels are higher than times normal. Levels seldom exceed 10 times AST levels in most instances, the AST level normal. The degree of aminotransferase ele- may be higher than the ALT level, especially

34 SALT—NONALCOHOLIC FATTY LIVER DISEASE in the presence of cirrhosis or when under- vated serum ferritin values are found in half lying alcoholism is present. However, the of patients with NAFLD. Increased transfer- AST/ALT ratio is almost never greater than rin saturation is found in 6% to 11% of pa- 2 in NAFLD. In those patients with elevated tients. The hepatic iron index and the hepatic ALT levels, the elevation is usually persistent iron level are usually within the normal although the precise value may fluctuate. Less range. commonly, the serum ALT level remains per- sistently normal. Hepatic Synthetic Function Tests (Albumin, It is not clear if ALT should be used alone Prothrombin Time, Serum Bilirubin) or whether AST and ALT or other combina- tions of liver tests should be used. There is As expected, measures of hepatic function- also debate regarding whether cutoffs should al capacity do not become abnormal until cir- be different for men and women. Further- rhosis and liver failure develop. The serum more, the level of elevation that is considered albumin level and prothrombin time become to be abnormal varies widely and has recent- abnormal before the serum bilirubin level ly been brought into question.6,7 does. In a diabetic subject with NASH, iso- Relative to the controversy over the sensi- lated hypoalbuminemia may also occur due tivity of aminotransferase levels, it is clear to proteinuria related to diabetic nephropa- that more acceptable and accurate methods thy. for the noninvasive recognition and diagnosis of liver disease in general and NAFLD in par- Complete Blood Count (CBC) ticular are needed. While most individuals Hematological parameters are usually nor- with NAFLD have abnormal liver chemis- mal unless cirrhosis and portal hypertension tries, it is clear that significant liver disease lead to hypersplenism. Thrombocytopenia, can exist with liver enzymes in the normal with or without detectable splenomegaly, is a range among NAFLD patients. common clinical clue that cirrhosis (complicated by hypersplenism) has developed. Al- Gamma-Glutamyltransferase (GGT) coholism associated bone marrow suppres- Although GGT levels may be elevated, sion can result in reversible thrombocytope- there are little data on the frequency and de- nia that can be mistaken for hypersplenism. gree of elevation. The degree of elevation tends to be lower than that seen with alco- ABDOMINAL IMAGING holic liver injury. Abdominal imaging studies are often or- dered instead of liver biopsy to confirm the Alkaline Phosphatase clinical suspicion of NAFLD. The rationale The alkaline phosphatase level may also be here is the avoidance of the risks associated variably elevated up to twice the upper limit with an invasive procedure. (See Liver Biopsy of normal. section). The presence of fat in the liver can be diagnosed in many cases using various imag- Ferritin ing modalities. In clinical practice, steatosis is Elevated serum ferritin levels may be a clue commonly detected through noninvasive im- that nonalcoholic steatohepatitis (NASH) is aging with ultrasonography (US), computer- present, because elevated ferritin with or ized axial tomography (CT) and magnetic without increased transferrin saturation has resonance imaging (MRI) when it exceeds been described in many NASH patients who 25% to 30% of liver weight. Liver imaging do not have C282Y or H63D mutations in the techniques are not sensitive for the detection HFe gene (ie, genetic hemochromatosis). Ele- of many individuals with lesser degrees of

35 JOURNAL OF INSURANCE MEDICINE steatosis. The expense of various imaging Thus, liver biopsy remains the only accurate modalities is not trivial, and none of these way to diagnose steatohepatitis. can distinguish simple steatosis from NASH or ‘‘uncomplicated’’ NASH from NASH with LIVER BIOPSY fibrosis. Of these, sonography is the least expensive, and MRI is the most expensive mo- In the absence of steatosis confirmed on dality.8 liver imaging, NAFLD is estimated practical- On ultrasound, steatosis produces a diffuse ly as the percentage of fat-laden hepatocytes increase in echogenicity relative to the kid- observed by light microscopy. The spectrum neys. Fibrosis/cirrhosis has a similar appear- of histologic abnormalities defined by ance regardless of etiology. NAFLD includes simple steatosis (steatosis On CT, steatosis produces a low-density without other liver injury) and NASH as its hepatic parenchyma, which is diffuse in most more extreme form. Liver biopsy is required people who have NAFLD. CT, as well as MR to secure a diagnosis of NAFLD if imaging imaging, can semi-quantitatively estimate liv- is not diagnostic (that is, if the percentage of er fat content. An unenhanced hepatic CT hepatic steatosis is less than 25%–35%), and scan remains the optimal CT method for de- to confirm the diagnosis of NASH, fibrosis, tection of a fatty liver. In those with a fatty and/or cirrhosis. In clinical practice, taking liver, the hepatic attenuation is less than that the risk of liver biopsy is often unnecessary of the blood vessels giving the appearance of if clinical and imaging evidence of advanced a contrast-enhanced scan when no contrast is liver disease complicated by cirrhosis and used. portal hypertension are identified. Differences in the precession frequency be- The value of a liver biopsy for the diagnosis tween water and fat protons can be used to of NAFLD in routine clinical practice is hotly diagnose fatty liver using MRI. T1-weighted debated. Arguments against a liver biopsy in- gradient-echo images obtained by keeping clude the generally good prognosis of most the water and fat spins out of phase show a patients with NAFLD, the lack of an estab- loss of liver hepatic signal intensity relative lished form of effective therapy, and the risks to in-phase images. The fatty liver also has a and costs associated with a biopsy. lower signal intensity compared with adja- Stephen H. Caldwell, MD, coauthor of one cent muscle. Several newer modifications in of the recent reviews on NAFLD,9 recently MRI techniques have resulted in considerable discussed liver biopsy in NAFLD.10 Lack of improvement in the ability to diagnose a fatty knowledge concerning an individual patient’s liver by MRI. extent of the disease may lead to unexplained Occasionally, fatty infiltration is focal. As a fluid retention, inappropriate assignment of consequence, ultrasound and CT scans may blame for liver disease to common drugs like be misinterpreted as reflecting malignant liv- diabetes medications, and to ill-advised die- er masses. MRI can distinguish space-occu- tary recommendations. Patients with fibrosis pying lesions from focal fatty infiltration (iso- and cirrhosis may be at increased risk for ad- lated areas of fat infiltration) or focal fatty verse reactions to drugs prescribed for con- sparing (isolated areas of normal liver). ditions that often coexist with NASH. Most Despite the utility of these imaging mo- patients with type 2 diabetes are taking ACE dalities in the diagnosis of diffuse fatty dis- inhibitors because the drugs have been orders of the liver, none of these modalities shown to preserve renal function. But if they can distinguish between fatty liver vs stea- have bridging fibrosis or cirrhosis on liver bi- tohepatitis. Moreover, diffuse fibrosis is also opsy, the increased risk of fluid retention and associated with a hyper-echogenic ultrasound ascites require consideration. It may not be pattern and cannot be distinguished from fat- advisable to stop these medications immedi- ty liver with accuracy by ultrasonography. ately, but if these problems develop, a secure

36 SALT—NONALCOHOLIC FATTY LIVER DISEASE diagnosis of liver fibrosis/cirrhosis can help nosis in 14% of cases, as well as altered the guide decisions regarding continuation of frequency of monitoring laboratory studies in treatment. 36% of cases and the treatment recommen- A strategy that includes an observation pe- dations in 12 patients. riod of up to 6 months after NAFLD/NASH Although it is essential to include only bi- is suspected is suggested. If signs and symp- opsy-proven cases of NASH in clinical trials, toms persist despite appropriate lifestyle and the decision to perform a liver biopsy in rou- dietary changes, a liver biopsy can be consid- tine clinical practice should take into considered. eration the specific clinical questions that are On the other hand, there is little controver- relevant in a given case (eg, exclusion of al- sy that a liver biopsy is the only accurate ternate causes of liver disease, ascertainment method for the diagnosis of NASH, as shown of degree of fibrosis, and determination of by the poor positive predictive value of clin- long-term prognosis). Thus, both the decision ical and laboratory evaluation for the diag- to perform a liver biopsy in a patient with nosis of NASH using histology as the gold suspected NAFLD and the timing of the bi- standard. opsy must be individualized and should in- Some factors may help to identify patients clude the patient in the decision-making pro- with NAFLD in whom liver biopsy may process. vide the most prognostic information.11,12 An age of 45 years or more, the presence of obesity or type 2 diabetes mellitus, and a ratio NAFLD AND THE INSULIN of AST:ALT of 1 or greater increase the like- RESISTANCE SYNDROME lihood of the presence of advanced fibrosis. In the subgroup of overweight patients with Recent studies confirm the common (al- a body mass index over 25, older age, higher though not invariable) association of NAFLD body-mass index, and higher levels of ALT with both the metabolic syndrome X (insulin and triglycerides are also indicators of more resistance syndrome) and alterations in cen- advanced fibrosis. In severely obese patients tral stress response systems. Management with a body mass index of more than 35, an and treatment of NAFLD/NASH, metabolic index of insulin resistance of more than 5, syndrome, and the adverse effects of harmful systemic hypertension, and an elevated ALT stress response are evolving and may share level correlate strongly with the presence of important therapeutic implications, particu- steatohepatitis; whereas hypertension and larly if both the patient and physician are raised levels of ALT and C-peptide suggest committed to the task. the presence of advanced fibrosis. All of these Individuals at risk for the metabolic syn- predictor factors have not been studied ade- drome can be identified by history, physical quately as predictors of progression over time. Instead, they indicate the likelihood of examination and laboratory evaluation (Table finding more advanced disease on the initial 6). The metabolic syndrome is present in biopsy. more than 20% of the US adult population Sorbi et al13 recently studied the clinical and is associated with several potentially utility of a liver biopsy in routine clinical modifiable lifestyle factors.14,15 There is strong practice. A total of 36 subjects with persis- and increasing evidence that this syndrome is tently elevated ALT levels for which a diag- associated with the development of diabetes nosis could not be established by noninvasive mellitus and an increased risk of cardiovas- methods underwent a liver biopsy. A pre- cular disease. The identification and clinical sumptive diagnosis and plan of management management of this high-risk group is an im- were identified before the biopsies were per- portant aspect of diabetes and coronary heart formed. The liver biopsy changed the diag- disease management and prevention.

37 JOURNAL OF INSURANCE MEDICINE

Table 6. Metabolic Syndrome Criteria Third Report patients with this disease are needed to guide of the National Cholesterol Education Program Adult future decisions. In the meantime, a number Treatment Panel III (ATP III)* of reasonable treatment options are available. The leading Internet-based medical reference The ATP III clinical de®nition of the metabolic syn- (www.UpToDate.com) states,17 ‘‘There is no drome requires the presence of 3 or more of the fol- proven effective therapy for NASH.’’ At- lowing: 1) Abdominal obesity (waist circumference Ͼ40 inch- tempts are made to modify potential risk fac- es in men and Ͼ35 inches in women tors such as obesity, hyperlipidemia, and 2) High triglyceride level (Ն150 mg/dL) poor diabetic control. Weight reduction 3) Low HDL cholesterol level Ͻ40 mg/dL for men should be gradual, since rapid weight loss and Ͻ50 mg/dL for women) has been associated with worsening of liver 4) High blood pressure (systolic greater than or equal disease. Several potential treatments have to 130 mm Hg or diastolic Ն85 mm Hg) 5) High fasting plasma glucose concentration (Ն110 been described, although none is used rou- mg/dL) tinely for clinical practice, with the exception * Expert Panel on Detection, Evaluation, and Treat- of vitamin E. ment of High Blood Cholesterol in Adults. Executive Summary of the Third Report of the National Detection, Stress Reduction and Lifestyle Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA. 2001;285: Modification 2486±2497. In accordance with the concept of ‘‘allostatic load,’’ it is reasonable to recommend that patients deploy a strategy to reduce the INSURANCE DILEMMA harmful effects of the ‘‘bad stress response,’’ Dr. Caldwell commented recently on ‘‘The at least with a strategy of diet, exercise and Insurance Dilemma’’ of NAFLD.16 ‘‘A liver bi- weight management. Other methods of stress opsy may help patients maintain affordable management (eg, relaxation and breathing health insurance. When histology shows only techniques, meditation, yoga) may also be in- Class 1 or 2 NAFLD, the condition is most corporated. likely stable and should not necessitate crush- Results from large, multicenter trials indi- ing rate hikes or denial of coverage. Insurance cate that lifestyle modifications (eg, diet, ex- companies frequently reject the claims of ercise) significantly reduce the risk of devel- these patients or raise the rates astronomical- oping type 2 diabetes. It is reasonable to rec- ly on the basis that high transaminase or ami- ommend this approach for those with notransferase levels signify a progressive dis- NAFLD, since the liver disorder is strongly ease. But for NAFLD (that has not progressed associated with insulin resistance (metabolic to NASH), that isn’t the case. It’s not right to syndrome X). However, it is not yet clear label a whole group at increased risk when whether this strategy will prevent the devel- only a portion really is. We can make some opment of NASH/fibrosis or help individuals useful distinctions based on biopsy. Insurers who have already developed diabetes or who shouldn’t deny coverage to patients with have histologically advanced NAFLD. Small Classes 1 and 2 steatosis. NASH patients pre- studies have demonstrated that weight loss sent a more complicated ethical dilemma.’’ improves liver enzyme elevations in patients with NAFLD and may reduce the degree of steatosis. However, it is equally true that rap- TREATMENT OF NAFLD id, extreme weight loss, such as that induced There is no proven therapy for NAFLD. by bariatric surgery, can accelerate hepatic de- Clearly, more information about the natural compensation in some patients with NAFLD. history of NAFLD and the results of large, For most patients, gradual weight loss is randomized prospective treatment trials for advisable, even though the most effective rate

38 SALT—NONALCOHOLIC FATTY LIVER DISEASE

and degree of weight loss have not been es- and beta-carotene, beta cryptoxanthin, lyco- tablished. More information will be needed pene, lutein, and zeaxanthin. relative to application of weight loss techniques. Little is known about the relative im- Insulin-Sensitizing Drugs portance of changing diet composition as op- posed to general caloric restriction. Mouse models of insulin resistance and fat- Exercise is an important component of ty liver disease have shown that treatment most successful weight loss programs and with metformin or thiazolidinediones im- physical activity enhances insulin sensitivity. proves both conditions. However, there has Most people would benefit from more exer- been some hesitancy about recommending ei- cise; however, it is not clear whether merely ther drug as a general therapy for NAFLD increasing physical activity would provide because of potential treatment-related toxici- the same benefits as dieting for patients with ty. Troglitazone was withdrawn as a first-line NAFLD. therapy for type 2 diabetes because of rare, but potentially fatal, hepatotoxicity. Phenfor- min, the parent compound of metformin, is Antioxidants known to cause life-threatening lactic acido- sis; thus metformin is contraindicated in pa- As discussed earlier, the pathogenesis of tients with liver disease. Regardless, results NAFLD has been hypothesized to be a 2- of the same diabetes prevention trial that stage process with oxidative stress as the pri- demonstrated the efficacy of lifestyle modifi- mary mediator of the second stage. Antioxi- cations also indicated that metformin was dants, such as certain vitamins and minerals, well tolerated in that large population of would be expected to protect against liver in- obese, insulin-resistant subjects and signifi- jury. cantly decreased the incidence of overt dia- In a few trials, treatment with vitamin E betes. The safety and efficacy of troglitazone has been demonstrated to improve liver en- could not be evaluated, because that arm of zyme abnormalities in patients and in certain the study was terminated prematurely be- animal models with fatty liver disease. In cause of concerns about potential hepatotox- clinical practice, gastroenterologists common- icity. It remains plausible that second-gener- ly recommend that patients with NAFLD ation thiazolidinediones, which appear to take vitamin E, since it is cheap and well tol- have less intrinsic hepatotoxicity, might be erated. beneficial in treating NAFLD. Leptin might Other antioxidants, such as betaine, also also have some role as a therapy for selected appear to have some efficacy. Ruhl and Ev- patients with NAFLD. In lipodystrophic mice erhart18 recently evaluated the relationship of that develop leptin deficiency caused by li- elevated serum alanine aminotransferase ac- poatrophy, associated insulin-resistance and tivity with iron and antioxidant levels in the hepatic steatosis have been cured with leptin United States. In this large, national, popula- therapy, suggesting that leptin might be help- tion-based study, the risk for apparent liver ful for lipodystrophic patients with NAFLD. injury was associated with increased iron and decreased antioxidants, particularly caroten- Ursodeoxycholic Acid oids. Nutrients with antioxidant activity include vitamins E (alpha-tocopherol) and C A small study demonstrated that urso- (ascorbic acid), the carotenoids, and the ele- deoxycholic acid improves liver enzyme ab- ment selenium, an essential component of the normalities in NAFLD. Although data sup- antioxidant enzyme, glutathione peroxidase. porting its efficacy in NAFLD is sparse, hep- The carotenoids are a large family of com- atologists are very familiar with this drug, pounds, the most common of which are alpha which is commonly used to treat cholestatic

39 JOURNAL OF INSURANCE MEDICINE liver diseases, and it has become a popular late cellular oxidant production and trigger therapy for NAFLD. The mechanism for the inflammation or taking medications to im- putative beneficial effect of ursodeoxycholic prove their antioxidant/anti-inflammatory acid in NAFLD is uncertain, and the results defenses, whereas others with ‘‘good’’ in- of at least one large ongoing trial should clar- flammation-control genes can be reassured ify whether or not more extensive scrutiny of that they can safely enjoy these pleasures. this agent is justified. The research agenda for the future includes establishing the role of insulin resistance and Lipid-lowering Drugs abnormal lipoprotein metabolism in NASH, determining the pathogenesis of cellular in- It is reasonable to consider lipid-lowering jury, defining predisposing genetic abnor- drugs as potential treatments for NAFLD, be- malities, identifying better noninvasive pre- cause it is a disorder of hepatic lipid homeo- dictors of disease and defining effective ther- stasis. Hypertriglyceridemia and reduced apy. HDL cholesterol level are the types of dysli- pidemia associated with NAFLD. A small tri- Based on a presentation to the American Academy al of gemfibrozil demonstrated no apprecia- of Insurance Medicine, Scottsdale, Ariz, October 15, ble effect on NAFLD. The rationale for using 2003. agents that generally lower cholesterol as therapy for NAFLD is unclear, and these REFERENCES agents have been associated with some inci- dences of liver injury. Nevertheless, a prelim- 1. Matteoni C, Younossi Z, Gramlich T, Boparai N, Liu Y, McCullough A. Nonalcoholic fatty liver dis- inary report of promising effects of atorvas- ease: a spectrum of clinical and pathologic sever- tatin in a small group of patients with ity. Gastroenterology. 1999;116:1413–1419. NAFLD suggests that 3-hydroxy-3-methyl- 2. Brunt E, Janney C, Di Bisceglie A, Neuschewander- glutaryl-coenzyme A reductase inhibitors Tetri B, Bacon B. Nonalcoholic steatohepatitis: a may play some role in the treatment of this proposal for grading and staging the histological disease. However, currently, the routine use of lesions. Am J Gastroenterology. 1999;94:2467–2474. 3. McEwen B. Protective and damaging effects of statins to treat NAFLD cannot be recom- stress mediators. NEnglJMed.1998;338:171–179. mended. 4. Charlton M, Kasparova P, Weston S, et al. Fre- quency of nonalcoholic steatohepatitis as a cause of advanced liver disease. Liver Transpl. 2001;7:608– THE FUTURE 614. The National Institutes of Health is assem- 5. Sanyal A. Gastroenterology & Endoscopy News. Feb- ruary 2003. bling a multicenter network to conduct large 6. Prati D, Taioli E, Zanella A, et al. Updated defi- natural history and treatment trials for pa- nitions of healthy ranges for serum alanine ami- tients with NAFLD, while basic scientific re- notransferase levels. Ann Intern Med. 2002;137:1– search continues to be focused upon patho- 10. genesis. It may soon be possible to counsel 7. Kaplan M. Alanine aminotransferase levels: individuals about their risk for NAFLD, What’s normal? (editorial). Ann Intern Med. 2002; 137:49–51. based upon both genetic and environmental 8. Mofrad P, Sanyal A. Nonalcoholic fatty liver dis- factors. Both ‘‘nature’’ (ie, genetic control of ease. Medscape General Medicine. 2003;5(2). Available inflammatory responses) and ‘‘nurture’’ (ie, at: www.medscape.com/viewarticle/449315. Ac- epigenetic causes of oxidative stress and in- cessed on November 10, 2003. flammation) contribute to this NAFLD. Thus, 9. Neuschwander-Tetri B, Caldwell S. Nonalcoholic individuals who have inherited the ‘‘bad’’ steatohepatitis: summary of an AASLD single top- ic conference. Hepatology. 2003;37:1202–1219. tendency to have sustained inflammatory re- 10. Caldwell S. Gastroenterology and Hepatology News. sponses might be better off minimizing the February 2003. consumption of alcohol or foods that stimu- 11. Angulo P, Keach J, Batts K, Lindor K. Independent

40 SALT—NONALCOHOLIC FATTY LIVER DISEASE

predictors of liver fibrosis in patients with nonal- 15. Hill J. What do we do about the metabolic syn- coholic steatohepatitis. Hepatology. 1999;30:1356– drome (editorial). Arch Intern Med. 2003;163:395– 1362. 397. 12. Angulo P. Nonalcoholic fatty liver disease. NEngl 16. Caldwell S. Gastroenterology and Endoscopy News. J Med. 2002;346:1221–1231. February 2003. 13. Sorbi D, McGill D, Thistle J, Therneau T, Lindor K. 17. Sheth S, Chopra S. Nonalcoholic steatohepatitis. An assessment of the role of liver biopsies in Rose B, Rush J, eds. Available at: www.UptoDate. com. Accessed on August 1, 2003. asymptomatic patients with chronic liver test ab- 18. Ruhl C, Everhart J. Relation of elevated serum al- normalities. Am J Gastroenterology. 2000;95:3206– anine aminotransferase activity with iron and an- 3210. tioxidant levels in the United States. Gastroenterol- 14. Park Y, Shankuan Z, Palaniappan L, Heshka S, ogy. 2003;124:1821–1829. Carnethon M, Heymsfield S. The metabolic syn- 19. Clark J, Brancati F, Diehl A. Nonalcoholic fatty liver drome, prevalence and associated risk factor find- disease. Gastroenterology. 2002;122:1649–1657. ings in the US population from the third National 20. Salt WB II. Stress. In: Johnson L, ed. Encyclopedia Health and Nutritions Examination Survey, 1999– of Gastroenterology. San Diego, Calif: Elsevier/Aca- 94. Arch Intern Med. 2003;163:427–436. demic Press; 2004:462–465.