D

D

TtiTntehttdwovalwCpfdshmpmcvtc

Bcfc(ti

GASTROENTEROLOGY 2008;134:1682–1698

iagnosis and Therapy of Nonalcoholic Steatohepatitis

Dawn M. Torres Stephen A. Harrison

epartment of Gastroenterology, Fort Sam Houston, Brooke Army Medical Center, San Antonio, Texas

tshcfhaapatenpo

ipombircNNmc

idnlnuoa

he increasing prevalence of obesity, insulin resis-ance, and the metabolic syndrome has significantmplications for the future of chronic liver disease.he resultant increase in the number of patients withonalcoholic fatty liver disease (NAFLD) is expected

o translate into increased numbers of patients withnd-stage liver disease (cirrhosis), liver failure, andepatocellular carcinoma. It is particularly important

o identify the patients who are at greatest risk ofhese aforementioned complications of chronic liverisease, those nonalcoholic fatty liver disease patientsith nonalcoholic steatohepatitis. Currently liver bi-psy is the gold standard for diagnosis, but less in-asive, highly accurate, and affordable screening toolsre required. These tools may include radiologic oraboratory studies to identify patients noninvasivelyho may benefit from therapeutic interventions.linical scoring systems that may be used in generalractice as initial screening tools also may prove use-ul. Most therapeutic modalities available or underevelopment target the major pathways thought es-ential in the pathogenesis of nonalcoholic steato-epatitis and often are directed at reducing bodyass index and improving insulin resistance via

harmacologic, surgical, dietary, or exercise regi-ens. Other potential therapeutic agents directed at

ytoprotection or reduction of fibrosis are under in-estigation. This article focuses on diagnosis andherapy available and under development for thishronic liver disease.

ecause obesity has become commonplace in today’smodern society, the face of chronic liver disease is

hanging to reflect a significant increase in patients witheatures of the metabolic syndrome, leading to an in-reased prevalence of nonalcoholic fatty liver diseaseNAFLD). It is estimated that 30% of the adult popula-ion in the United States now has NAFLD, with an

ncreasing prevalence to 90% in morbidly obese popula-

ions presenting for bariatric surgery.1–3 This is worri-ome because patients with NAFLD appear to have aigher all-cause mortality in addition to liver-relatedause of death and recent data points to an increased riskor cardiovascular disease.4 – 8 However, in clinical andistopathologic studies, the course of isolated fatty liverppears more indolent and less likely to progress todvanced liver disease.9 Alternatively, a subset of NAFLDatients have a more aggressive form of fatty liver knowns nonalcoholic steatohepatitis (NASH), in which pa-ients are at greater risk for progression to cirrhosis,nd-stage liver disease, and likely hepatocellular carci-oma. It is estimated that 3%– 6% of the United Statesopulation has NASH, with upwards of 30% of morbidlybese patients showing features of NASH.3,10

Because of the inordinately high prevalence of NAFLD,t is important to identify those patients with NASH,articularly those at risk for advanced disease. Liver bi-psy traditionally has been the gold standard used toake the diagnosis of NASH. However, without adequate

iopsy tissue, there is a high degree of sampling variabil-ty. Additional problems include the small but inherentisk of complications, potential patient discomfort, andost. Subsequently, noninvasive tools to correctly identifyAFLD patients who have histopathologic evidence ofASH with or without advanced fibrosis are in develop-ent and include a variety of techniques that vary from

omposite laboratory and biomarker data to tests that

Abbreviations used in this paper: ADP, adiponectin; BMI, body massndex; CRP, C-reactive protein; CT, computed tomography; HDL, high-ensity lipoprotein; LDL, low-density lipoprotein; MR, magnetic reso-ance; NAFLD, nonalcoholic fatty liver disease; NAS, nonalcoholic fatty

iver disease activity score; NASH, nonalcoholic steatohepatitis; NPV,egative predictive value; PPV, positive predictive value; PUFA, poly-nsaturated fatty acid; RCT, randomized controlled trial; ROC, receiveperating curve; ROS, reactive oxygen species; SFA, saturated fattycid; TZD, thiazolidinediones; US, ultrasound.

© 2008 by the AGA Institute0016-5085/08/$34.00

doi:10.1053/j.gastro.2008.02.077

m(

acTstsmaicrf

ccll(3pafi

orbac(rsioHoUa

Hpoosdip

Ugaaust

bsLubMlvtan

igOdqsmtprl

T

UL

CC

MMMU

May 2008 DIAGNOSIS AND THERAPY OF NASH 1683

easure hepatic tissue elasticity such as the FibroscanEchosens, Paris, France).

Establishing the diagnosis of NASH is helpful only ifn appropriate care plan can be developed and this in-ludes providing an effective treatment for the disease.hese therapeutic modalities should be readily available,

afe, effective, and relatively inexpensive. Therapies thatarget the underlying metabolic syndrome through life-tyle modification, dietary adjustments, exercise regi-

ens, bariatric surgery, or medications are all potentialvenues of treatment. Other potential therapeutic targetsnclude cytoprotective and antifibrotic agents. This arti-le focuses on the diagnostic and therapeutic tools cur-ently available as well as those under investigation foruture clinical use.

DiagnosisPatients with NAFLD often are asymptomatic and

ome to attention secondary to mild to moderate in-reases in hepatic aminotransferase levels or abnormaliver appearance on abdominal imaging. Liver enzymeevel increases tend to show an alanine aminotransferaseALT) predominance and rarely are increased more than

times the upper limits of normal.11,12 The alkalinehosphatase level occasionally may be increased mildly,nd rarely is the only liver enzyme abnormality identi-ed.13

Conventional radiology studies used in the diagnosisf fatty liver include ultrasound (US), computed tomog-aphy (CT), and magnetic resonance (MR) imaging (Ta-le 1). US can identify hepatic steatosis with reasonableccuracy. Typical criteria used to assess for steatosis in-lude hepatorenal echo contrast, liver brightness, deepposterior beam) attenuation, and vascular blurring. Aecent Italian study of 235 patients showed an excellentpecificity of 97%, but a lower sensitivity of 64%. Thisncreased to 100% specificity and 89.7% sensitivity whennly including patients with at least 30% steatosis.14

amaguchi et al15 used a 6-point scoring system basedn liver brightness, attenuation, and vascular blurring onS to evaluate for NAFLD and showed 100% specificity

able 1. Radiology Studies in the Diagnosis of NAFLD

Modality Sensitivity Specificity

S14–18 49%–89% 75%–100%evovist US34 100% 100%

T19,20 54%–93% 95%T with liver-spleenattenuation ratios21

82% 100%

R imaging22 NA NAR spectroscopy24 NA NAR elastography38 98% 99%S elastography36 87% 91%

nd 91.7% sensitivity when compared with liver biopsy. t

owever, this study was performed in relatively thinatients. When performed in morbidly obese patients,thers have shown the sensitivity and specificity to benly 49.1% and 75%, respectively, for identifying steato-is.16 Other studies have emphasized the difficulty inistinguishing steatosis from fibrosis on US and the

nability of US to correlate with the other essential he-atic histology required for the diagnosis of NASH.17,18

CT has shown a similar diagnostic yield similar to US.nenhanced CT is accurate in predicting steatosis ofreater than 30%, but has been shown to be much lessccurate in predicting lower-grade steatosis.19 Diagnosticccuracy has been improved with unenhanced CT scansing liver:spleen attenuation ratios with up to 100%pecificity and 82% sensitivity for hepatic steatosis greaterhan 30%.19 –21

Hepatic steatosis may be quantified from MR imagingased on the signal differences between fat and water andhows good correlation with microscopic fat content.22

imitations of this modality include expense, inability tose in patients with implantable devices or claustropho-ia, and altered values in patient with iron overload.23

agnetic reasonance spectroscopy is a specialized radio-ogic study that can measure triglyceride content nonin-asively. This technique was used on 2349 patients fromhe Dallas Heart Study with highly reproducible andccurate results, but widespread application of this tech-ology is limited by cost and availability.24

As a group, radiologic studies have been shown todentify hepatic steatosis accurately and often bring aroup of asymptomatic patients to clinical attention.ccasionally, patients with NAFLD are symptomatic andescribe a fullness or discomfort in their right upperuadrant that does not appear to be associated witheverity of disease. Physical examination generally is nor-

al with the exception of the common finding of hepa-omegaly. A novel finding of increased dorsocervical li-ohypertrophy in patients with NAFLD was describedecently by Cheung et al.25 Interestingly, dorsocervicalipohypertrophy was found to be the single greatest con-

Comments

Identifies steatosis onlySmall preliminary study of 64 patients—potentially useful in

NASH patients�30% steatosis, cannot differentiate NASHBetter than standard CT for hepatic steatosis �30%; no

NASH distinctionHighly accurate in determining steatosis �30%Highly accurate for steatosisGood for both steatosis and fibrosis; early studies onlyFor stage 4 fibrosis, cannot differentiate NASH (accuracy in

obese patients is unknown)

ributor to the severity of histologic findings of steato-

hn

aAs7sfiDtofiwhowmimeg

dtslwasvNaAsswrs

amUawioct

ioad

Nscloa

pnsd5NacUd

ibsaeupssspvsfttitbt

bhpsarftmctei

1684 TORRES AND HARRISON GASTROENTEROLOGY Vol. 134, No. 6

epatitis and may prove to be a useful tool in the diag-osis of NASH.NAFLD patients typically meet criteria for the met-

bolic syndrome as defined by the American Heartssociation and National Heart, Lung, and Blood In-

titute, with prevalence rates reported from 47% to1%.26 –28 In addition, the presence of the metabolicyndrome has been linked to NASH as well as severebrosis in NAFLD patients on multivariate analysis.29

efining criteria for the metabolic syndrome (applyinghe earlier-described society guidelines) include any 3f the following 5 laboratory and physical examinationndings: increased waist circumference (men, �40 in;omen, �35 in), increased triglyceride levels of 150 origher, reduced high-density lipoprotein (HDL) levelsf less than 40 mg/dL in men or less than 50 mg/dL inomen, increased blood pressure of greater than 130m Hg systolic or greater than 85 mm Hg diastolic, or

ncreased fasting glucose level of greater than 100g/dL.30 Lipid and glucose measurements are consid-

red positive if the patient is already on cholesterol- orlucose-lowering medications.

Percutaneous liver biopsy remains the standard foristinguishing isolated fatty liver from NASH. Key his-opathologic criteria seen with NASH include macrove-icular steatosis, lobular inflammation, hepatocyte bal-ooning, and often perisinusoidal/perivenular fibrosis asell as Mallory–Denk bodies.31 However, agreementmong pathologists is not universal. The NAFLD activitycore (NAS) is a histologic scoring system that was de-eloped by the National Institutes of Health–sponsoredASH Clinical Research Network, which focuses on ste-

tosis, lobular inflammation, and hepatocyte ballooning.NAS of 5 or greater is consistent with NASH and a

core of 2 or less is not associated with NASH, withcores of 3 or 4 falling somewhere in between. The NASas not intended to be used as a diagnostic tool, but

ather to provide a uniform tool for assessing diseaseeverity, ideally in clinical trials.32

Recent data have shown that significant sampling vari-bility exists for fibrosis and inflammation, although thisay be attenuated with good core biopsy samples.33

nfortunately, adequate biopsy samples remain an issuend given the significant prevalence of NAFLD coupledith the previously described limitations of biopsy, fewer

nvasive tests are being developed to identify NASH withr without advanced fibrosis. These diagnostic modalitiesan be divided into tests using image analysis and testshat combine biomarker and clinical data.

Imaging for NASHAn important limitation of all imaging modalities

ncluding CT scan, US, and MR imaging is the inabilityf these techniques to differentiate isolated hepatic ste-tosis from steatohepatitis. The definitive study by Saa-

eh et al18 of 90 consecutive patients with biopsy-proven p

AFLD who underwent CT, US, and MR imaging clearlyhowed that these radiologic studies were unable to ac-urately discriminate between NASH and isolated fattyiver. All 3 radiologic studies were best able to predict 33%r greater steatosis, but were not accurate in assessing forny features of steatohepatitis.

Novel techniques are under development that mayrove to be useful radiologic tools to diagnose NASHoninvasively. One such idea involves contrast ultra-ound using Levovist (Shering, Berlin, Germany), withelayed images taken at 5- to 10-minute increments for0 minutes, and has shown great accuracy in diagnosingASH with a receiver operating curve (ROC) of 100%

mong the 21 NASH, 33 NAFLD, and 10 healthy patientontrols.34 This minimally invasive novel application ofS requires further study but shows early promise as aiagnostic tool.

Imaging for NASH FibrosisAlthough efforts to use noninvasive imaging stud-

es to accurately grade NAFLD in terms of NASH haveeen unsuccessful, the investigations that use imaging totage fibrosis in patients, and thus identify those withdvanced disease, have shown more promise. Transientlastography is an ultrasound-based technology that issed to measure liver stiffness. In a recent study of 68atients with NAFLD, a stepwise increase in elasticity washown with severity of hepatic fibrosis.35 This studyhowed good correlation between liver biopsy fibrosistages 1– 4 and liver stiffness using ROCs with positiveredictive values (PPVs) between 64% and 93.5% for thearying stages of fibrosis. A recent meta-analysis of 9tudies using this technology showed pooled estimatesor sensitivity and specificity of 87% and 91%, respec-ively, for patients with stage IV fibrosis and 70% sensi-ivity and 84% specificity for stages II–IV fibrosis.36 Onemportant caveat that requires further investigation ishat steatosis and high body mass index (BMI; �25) haseen shown to reduce the reproducibility of this newechnology.37

An MR equivalent to transient elastography also haseen investigated as a noninvasive means to evaluate forepatic fibrosis. A recent study by Yin et al38 showedromising results yielding 98% and 99% sensitivity andpecificity, respectively, for identifying the presence ofny fibrosis. Hepatic steatosis also was identified cor-ectly and the degree of steatosis did not seem to con-ound fibrosis measurements. One limitation of this ini-ial study was that the liver biopsies were on average 6.5

onths apart from the imaging study, but the signifi-ance of this separation is uncertain. Further study intohis novel imaging technique is warranted. Both of theselastography techniques require large multicenter trialsn conjunction with liver biopsy before they can be ap-

lied to regular clinical practice.

vatNolib(

pmcmltpocsobtsapsrR

oses

cstnqts

ifsNiipacsclwarN

aibpsCNaip

aeaoopa(aap8mwNca

T

R

L

A

H

C

May 2008 DIAGNOSIS AND THERAPY OF NASH 1685

Differentiating Isolated Fatty LiverFrom NASHBiomarkersSerum biomarkers that could be used as nonin-

asive tests to distinguish NASH from isolated fatty liverlso have been investigated. One or several laboratoryests that could accurately identify, grade, and stageASH would allow for screening of the large population

f patients with NAFLD and avoid invasive and expensiveiver biopsies. Several classes of biomarkers are in thenitial stages of study that are directed at the pathwayselieved to be involved in the pathogenesis of NASHTable 2).

Oxidative stress is presumed to be central to theathogenesis of NASH and therefore biomarkers thateasure oxidative stress through reactive oxygen spe-

ies (ROS) have been looked at as potential surrogatearkers of NASH.39 Chalasani et al40 compared serum

evels of oxidized low-density lipoprotein (LDL) andhiobarbituric acid-reacting substances in 21 NASHatients with 19 control patients without liver diseasef similar age, BMI, and nutritional intake. Signifi-antly higher levels of lipid peroxidation products wereeen in NASH patients, but limitations included a lackf liver biopsy in the control patients and small num-ers overall. In contrast, no correlation between oxida-ive stress and steatosis or fibrosis was seen in anothertudy of 64 patients with either NAFLD or viral hep-titis.41 These discordant results may reflect that he-atic oxidative stress is not reflected accurately byerum measurements, but further study is required toender a final verdict on the utility of measuring serumOS in NAFLD/NASH patients.Another area of interest has been serum measurements

f specific adipocytokines (produced by adipose tissue)uch as leptin and adiponectin (ADP).42 Obesity andxcess caloric intake enhance up-regulation of leptin re-

able 2. Noninvasive Biomarkers in NASH

Serum biomarker Marker of Findings

OS40,41 Oxidative stress Conflicting results: somecorrelation betweenNASH and increasedlevel of ROS

eptin44–47 Insulinresistance

Conflicting results: higherlevels in some studies

diponectin48–54 Insulinsensitivity

ADP lower in NASHpatients

igh-sensitivity CRP55–57 Systemicinflammation

Conflicting results: someincreased high-sensitivity CRP withNASH compared withNAFLD

ytokeratin 1859–61 Hepaticapoptosis

Significantly higher inNASH

ulting in increased free fatty acid delivery to the liver, t

ontributing to hepatic steatosis.43 Initial investigationshowed an association between higher leptin levels andhe presence of NASH as well as steatosis grade, althoughot inflammation or fibrosis.44,45 However, several subse-uent studies have failed to show any associations be-ween leptin levels and the presence or absence of hepaticteatosis, inflammation, or fibrosis.46,47

In contrast to leptin, ADP is thought to promotensulin sensitivity and increase glucose utilization andree fatty acid oxidation in the liver. Hui et al48 foundignificantly lower serum levels of ADP in patients withASH when compared with both normal controls and

ndividuals with simple steatosis. Numerous other stud-es have confirmed that ADP is generally lower in NAFLDatients compared with healthy controls.49 –53 Targher etl54 recently showed that ADP levels were associatedlosely with severity of liver histology to include hepaticteatosis, necroinflammation, and fibrosis. Although all 3omponents of hepatic histology were associated withow ADP levels, only steatosis and necroinflammationere found to be independent predictors in multivariatenalysis. Further study is required to fully delineate theelationship between ADP and severity of liver disease inAFLD patients.Other markers of systemic inflammation such as C-re-

ctive protein (CRP) also have been studied with conflict-ng results. An Australian study found no differenceetween high-sensitivity CRP levels in patients with sim-le steatosis and NASH.55 This is in contrast to othertudies showing significant increases in high-sensitivityRP levels in patients with NASH compared withAFLD controls.56,57 Given these conflicting results, CRP

s a marker for hepatic inflammation requires furthernvestigation, although its association with other diseaserocesses may limit its clinical usefulness.Hepatic apoptosis is a prominent feature of NASH,

nd, as such, biomarkers of hepatic apoptosis have beenxplored as a potential diagnostic tool.58 Wieckowska etl59 showed a striking relationship between plasma levelsf caspase-generated cytokeratin-18, a protein involved inne of the final steps of apoptosis, and the histologicresence of NASH on liver biopsy. Similarly, Yilmaz etl60 measured serum levels of extracellular cytokeratin 18M30 antigen and M65 antigen) in patients with NAFLDnd varying degrees of severity of NASH. Both of thesentigens were found to be significantly higher in NASHatients with a sensitivity of 60%– 69% and a specificity of7%–97% in predicting NASH if the antigen levels werearkedly abnormal. The utility of cytokeratin 18 recentlyas validated independently in a study of 21 pediatricAFLD patients, which directly correlated the level of

ytokeratin-18 to severity of liver disease.61 This modalityppears to offer significant promise as a future diagnostic

ool to differentiate isolated fatty liver from NASH.

aNbvafTt3hms0it

NiipcsssNdi

pbrf

a3ttefitssomipPTwlil

to9rtdp

ccgAsla

T

SM

P

P

T

SPF

T

1686 TORRES AND HARRISON GASTROENTEROLOGY Vol. 134, No. 6

Scoring Systems for Identifying NASH FromNAFLDScoring systems using one or several clinical

nd/or laboratory parameters to identify patients withASH from the larger pool of NAFLD patients also have

een assessed. Several studies have looked at a variety ofariables that might prove suitable (Table 3). Palekar etl62 developed a clinical model that sums 5 risks factorsor NASH identified on multivariate logistic regression.hese factors include age 50 or older, female sex, aspar-

ate aminotransferase (AST) level of 45 or greater, BMI of0 or greater, AST/ALT ratio of 0.80 or greater, andyaluronic acid level of 55 or greater. Combining 3 orore of these factors yielded a sensitivity of 73.7% and a

pecificity of 65.7% for detecting NASH with an ROC of.763. This study provided a relatively simple means of

dentifying NASH patients but requires further valida-ion in a larger study population.

A larger European study assessed the utility of theASH Test (Biopredictive) to distinguish NASH from

solated fatty liver.63 This complex test including 13 clin-cal and laboratory parameters was developed using 160atients in the training group, 97 patients in a multi-enter validation group, and 383 control patients. Overallpecificity was shown to be 94%, with a much lowerensitivity of 33% and an ROC of 0.79. Further scoringystems are under development that may better identifyAFLD patients with NASH, but will require indepen-ent, prospective validation before being applied broadly

n clinical practice.Distinguishing Advanced Fibrosis in NASHThe ability of one or several screening tests to

redict hepatic fibrosis in patients with NAFLD also haseen an area of intense interest. Study of potential sur-ogate markers of fibrosis have shown success in identi-

able 3. Macronutrients in NAFLD

Macronutrient Found In

aturated fatty acids Lard, butter, coconut oil, palm oil Inonounsaturated fattyacids

Olive oil, nuts, avocados, peanutbutter, peanut oil

H

UFAs (n-6) Sunflower, corn oil D

UFAs (n-3) Fish oil, walnuts, salmon,shellfish

D

rans fatty acids Fast foods, baked goods, deepfried foods

Dairy products (�5%)Ruminants (cows/sheep)Margarine

In

(N

ucrose/fructose Sweetened drinks/sodas, candy Wrotein Meat, fish, eggs, dairy Uiber Whole grains, fruits, and

vegetablesIm

NF, tumor necrosis factor; ?, conflicting data.

ying those with advanced fibrosis but have been less 8

ccurate in predicting mild to moderate fibrosis (Table).64 One of the early scoring systems developed calledhe body mass index; age at liver biopsy; alanine amino-ransferase; and serum triglycerides (BAAT) score used 4asily determined clinical variables to assess for hepaticbrosis.65 This scoring system showed good PPV in de-ermining advanced fibrosis but fell short of an idealcreening test to identify mild to moderate disease. Theame group expanded their efforts with the developmentf the FibroTest-FibroSURE (Biopredictive) based on �2-acroglobulin, apolipoprotein A1, haptoglobin, total bil-

rubin, and �-glutamyltransferase levels.66 A 90% negativeredictive value (NPV) for advanced fibrosis and a 70%PV for advanced fibrosis were noted in final analysis.he diagnosis of borderline NASH was harder to makeith a 72% NPV and 74% PPV. This is an important

imitation of the FibroTest-FibroSURE, which translatesnto one third of total patients enrolled still requiring aiver biopsy.

The European Liver Fibrosis study evaluated 1021 pa-ients using 9 surrogate markers of fibrosis and devel-ped an algorithm using 4 of these markers that showed0% sensitivity in detecting fibrosis as well as 92% NPV inuling out fibrosis.67 This study was limited in the facthat it combined patients with all types of chronic liverisease and requires further study in NAFLD-specificatient populations.The NAFLD fibrosis scoring system, developed in

ollaboration between multiple liver centers, uses 6ommonly measured parameters including age, hyper-lycemia, BMI, platelet count, albumin level, and AST/LT ratio.68 This scoring system uses a 2-point cut-off

ystem similar to the FibroTest-FibroSURE. By using aow cut-off score (�1.455), the NPV for excludingdvanced fibrosis was 93% in the estimation group and

Action Findings in NASH

ses LDL Higher intake in NASH patientsintake decreases LDL,ases HDL

Uncertain

ses HDL, ? increaseative stress

High n-6/n-3 ratio in NASH patients

ses FFA, glucose, insulin,-�

Higher intake decreases hepaticsteatosis (small studies)

ses inflammatory markersLDLally derived trans FAs dohave above actions)

Uncertain

ns insulin resistance Higher intake in NASH patientsain Uncertaine insulin resistance Lower intake in NASH

creaigherincre

ecreaoxid

ecreaTNFcreaandaturnot

orsencertprov

8% in the validation group. When a high cut-off score

(fiNassfflbfm

tiTftmdtfiiNs

nogtpcettbt

oehoc

Fsht reas.

May 2008 DIAGNOSIS AND THERAPY OF NASH 1687

0.676) was used, the PPV for predicting advancedbrosis was 90% and 82%, respectively. Overall, theAFLD fibrosis score predicted advanced fibrosis with

n ROC of 0.82. The authors68 suggest that liver biop-ies could therefore be avoided in 75% of patients. Aeparate validation study of 79 patients found the NPVor excluding advanced fibrosis to be 93%, but the PPVor predicting advanced fibrosis was only 42%. Thisower PPV translates into fewer patients avoiding liveriopsy.69 Therefore, further validation is necessary be-ore this or any other scoring system can be recom-

ended broadly.A recent review by Guha et al70 tabulated 29 studies

hat looked at noninvasive markers of hepatic fibrosisn NAFLD either as primary or secondary end points.his comprehensive review identified the key variables

ound in the majority of these studies, which includedhe presence of diabetes, increasing age, increased ho-

eostatic insulin resistance, increased AST/ALT ratio,ecreased platelets, hyaluronic acid, and BMI. Al-hough these models are good in predicting advancedbrosis at any one time point, they fall short in defin-

ng an excellent screening test to identify patients withASH noninvasively who are at risk of progressing to

igure 1. Potential pathophysiologic effects of therapies that are unteatohepatitis is multifaceted. Several therapeutic modalities under inveave pleiotrophic effects in improving NAFLD. *Endocannabinoid recepo include adipocytes, liver, and possibly skeletal muscle and the panc

evere fibrosis. h

Treatment for NASHMuch like the biomarkers being studied to diag-

ose NASH noninvasively, current therapies are focusedn the various pathways thought central in the patho-enesis of this disease (Figure 1). Treatment regimensargeting insulin resistance, oxidative stress, diabetes, hy-erlipidemia, obesity, and hepatic fibrosis all warrantritical appraisal. Multiple modalities to include diet,xercise, surgical interventions, and finally pharmaco-herapy require evaluation to determine the most effec-ive treatment algorithms. Although one panacea has noteen found and is unlikely to exist, a multimodalityreatment regimen may prove effective.

Weight LossCalorie Reduction

Patients with NAFLD typically are overweight orbese, insulin resistant, and have a consistently highernergy intake when compared with individuals withoutepatic steatosis.71 Data have shown that in the setting ofbesity, moderate weight loss of approximately 6% viaaloric restriction improves insulin resistance and intra-

nvestigation. The development of hepatic steatosis and subsequention such as the TZDs, diet, and exercise, and possibly rimonabant, mayave been identified in multiple organs involved in energy homeostasis

der istigattors h

epatic lipid content.72 Furthermore, caloric restriction

it

swostwowafplpoHsadntahssteps

hu(etmcah

bbhvislha

Iw

whvnNicmt

mdpauosbbe1sm

stnihsArcPbifsad

fbitasi

eNNsss

1688 TORRES AND HARRISON GASTROENTEROLOGY Vol. 134, No. 6

mproves serum aminotransferase levels and hepatic his-ology.73–77

The degree of caloric restriction has been questioned. Amall study by Anderson et al74 showed that extremeeight loss via starvation leads to worsening liver histol-gy, including fibrosis. In addition, 2 bariatric surgerytudies have shown mild worsening of lobular inflamma-ion and fibrosis in a subset of patients with a meaneight loss of 32 and 38 kg, respectively.78,79 In contrast,thers have shown significant histologic improvementith weight loss through either caloric reduction or bari-tric surgery.80 – 84 Huang et al85 counseled patients toollow a 1400 kcal/day diet for 12 months in 15 biopsy-roven NASH patients with a subsequent mean weight

oss of 2.9 kg (3% of body weight). Among the 9 of 15atients (60%) who had a histopathologic improvementn repeat liver biopsy, the average weight loss was 7%.epatic fibrosis did not change significantly in this small

tudy. No significant change in fibrosis also was seen inmeta-analysis of 15 studies that evaluated weight re-

uction without bariatric surgery in the treatment ofonalcoholic fatty liver.86 Although it would seem thathe preponderance of the evidence supports that moder-te caloric restriction resulting in modest weight loss iselpful in improving biochemical parameters, insulinensitivity, and hepatic steatosis, its role in hepatic fibro-is is less clear. In keeping with the vast majority ofoday’s clinical practices, the recommendations for mod-rate caloric restriction as an initial therapy in NAFLDatients is supported by data, but well-controlled, largertudies are needed to conclusively show a fibrosis benefit.

Altering Macronutrient ContentEvidence suggests that NASH patients have diets

igher in simple carbohydrates and lower ratios of poly-nsaturated fatty acids (PUFAs) to saturated fatty acids

SFAs).87,88 Thus, the concept of altering the macronutri-nt content rather that the total calories of NASH pa-ients also has been investigated. Several areas of diet

acronutrient manipulation have been studied to in-lude overall fat vs carbohydrate vs protein intake as wells specific regimens of fatty acids designed to reduceepatic steatosis and potentially steatohepatitis.The ideal diet composition for NASH patients has not

een established, although the general consensus haseen that excessive intake of either fat or carbohydrate isarmful. A randomized controlled trial (RCT) of low-fats low-carbohydrate diet showed greater weight loss andmproved lipid panels in the low-carbohydrate group.89 Amall pilot trial of 5 NAFLD patients who followed aow-carbohydrate (�20 g/day) diet for 6 months showedistologic improvements in 4 of the 5 patients who hadmean weight loss of 12.8 kg.90

Other studies have shown benefit with lower-fat diets.mprovements in hepatic steatosis via MR spectroscopy

as shown in 2 recent trials: 1 of 10 obese nondiabetic g

omen91 and 1 of 8 NAFLD patients.92 A recent compre-ensive review by Zivkovic et al93 examined the role ofarious popular weight-loss diets to include Atkins, Or-ish, Zone, and South Beach diet in the treatment ofAFLD. This intriguing article suggested that diets that

mprove insulin sensitivity, promote foods with low gly-emic indices, and yield sustainable weight loss offer theost potential benefit, although well-designed clinical

rials are lacking.Another concept in the arena of dietary compositionodification is the difference in the various types of

ietary fat (Table 4). Excess SFAs have been shown toromote endoplasmic reticulum stress, hepatic steatosis,nd inflammation in animal models, although it remainsncertain if a minimum intake of SFAs are required forptimal metabolism.87,93–95 On the other hand, monoun-aturated fatty acids such as found in olive oil, peanututter, nuts, and avocados are thought to be generallyeneficial because they decrease total cholesterol, triglyc-rides, serum LDL, and maintain HDL. A small study of1 insulin-resistant patients showed improvement in in-ulin resistance and increased adiponectin levels after 1

onth of a monounsaturated fatty acid– enriched diet.96

Altering the amount of PUFAs in diets also has beentudied. When PUFAs have been used to replace SFAs inhe diet, significant cardiovascular benefits have beenoted.97 Hepatic steatosis in obese mice also has been

mproved with dietary PUFAs via negative regulation ofepatic lipogenesis.98 The ratio of n-6 to n-3 PUFAs alsoeems to be important in predicting insulin resistance.n excessive amount of n-6 PUFAs with a high n-6/n-3

atio was shown in a recent study of 45 NASH patientsompared with 856 age- and sex-matched controls.99 N-3UFAs such as alpha-linoleic acid (fish oil) appeared to beeneficial in 2 small studies in patients with NAFLD with

mprovement in serum triacylglycerol concentrations,asting glucose, liver enzyme levels, and hepatic steato-is.100,101 Walnuts are another source of alpha-linoleiccid that has been shown to improve lipid profiles iniabetes.102

On the other hand, trans fatty acids occurring in dairyoods as well as hydrogenated margarine products haveeen less well studied in patients with NAFLD. However,

nformation in healthy subjects suggests that intake ofhese hydrogenated oils increases inflammatory markersnd increases the LDL:HDL ratio.103,104 Based on thesetudies, general recommendations to avoid or reduce thentake of trans fatty acids seems reasonable.

The increased intake of sucrose and fructose in West-rn diets also may play a role in the development ofAFLD. A recent population-based study showed thatAFLD patients consumed almost twice the amount of

oft drinks than control counterparts.105 It has beenpeculated that the high fructose content of nondietodas is associated with increased hepatic de novo lipo-

enesis, hypertriglyceridemia, and hepatic insulin resis-

tmmtc

aNbdwmagc

lttptt

frwsepuwmat

mewPsmwmit

T

O

R

I

T

M

V

B

UPH

E

A

23

May 2008 DIAGNOSIS AND THERAPY OF NASH 1689

ance.106 This corresponds to a recent study in a mouseodel showing that rapidly absorbed carbohydrates pro-otes hepatic steatosis when compared with consump-

ion of slowly absorbed carbohydrates with a lower gly-emic index.107

This field of diet composition manipulation representsnew and intriguing area of study in the treatment ofASH that requires further study. Large RCTs are needed

efore final recommendations can be made as to the idealiet for NAFLD patients. The general concept that foodsith low glycemic index as well as diets containing moreonounsaturated fatty acids and n-3 PUFAs, low SFAs,

nd limited to no fructose seem to be acceptable generaluidelines to offer patients who are agreeable to lifestylehanges in their treatment.

ExerciseExercise is another area of interest in the field of

ifestyle modification as therapy for NAFLD althoughhere are only limited data available. Most of the evidencehat does exist uses exercise in combination with diet toromote weight loss. The general consensus has beenhat exercise works synergistically with diet modification

able 4. Pharmacologic Therapies in the Treatment of NASH

Treatment Mechanism Biochemical e

rlistat Weight loss 2 LFTs and insulin

imonabant Weight loss, possibleperipheral effects

2 Insulin resistanctriglyceride levels1 HDL, adiponectin

ncretin analogs(exendin-4)

Weight loss 2 LFTs, insulin reshemoglobin A1C

ZDs PPAR-� agonists 2 LFTs, insulin resand TNF-�1 Adiponectin

etformin 1 AMP kinase 2 LFTs and insulinNo effect on adipon

itamin E 2 Oxidative stress 2 LFTs

etaine 2 Oxidative stress 2 LFTs

DCA Hepatoprotective No changeentoxifylline Hepatoprotective 2 LFTs, TNF-�MG CoA-reductaseinhibitors

Improve lipid panel ? LFTs

zetimibe Blocks cholesterolabsorption inintestine

? LFTs

ngiotensin-receptorblockers

? Inhibits stellatecells

2 LFTs

, decreases; AMP, adenosine 5=-monophosphate; LFT, liver func-hydroxy-3-methylglutaryl-coenzyme A; TNF, tumor necrosis factor.

o produce beneficial metabolic effects.108 Suzuki et al109 N

ound that in a population of 348 male patients whoeceived annual health physicals, regular exercise andeight loss over a 1-year period was associated with a

ignificant improvement in serum aminotransferase lev-ls. Another study that followed up 25 obese NAFLDatients prospectively for 3 months confirmed that reg-lar exercise defined as walking or jogging daily alongith a reduced-calorie diet produced significant improve-ent in serum aminotransferase levels, fasting glucose,

nd hepatic steatosis compared with a control popula-ion.110

Hickman et al111 followed up 31 NASH patients for 3onths with combined dietary modifications and mod-

rate exercise of 150 minutes of aerobic exercise weeklyith a mean weight loss of 4% of total body weight.atients who lost weight experienced improvement inerum aminotransferase levels that were maintained at 15

onths of follow-up evaluation provided the weight lossas sustained. Repeat liver biopsy in 14 patients 3– 6onths after lifestyle modification showed significant

mprovement in steatosis and fibrosis. This correspondso a recent study that inversely correlates the presence of

Histologic effects Comments

tance 2 Steatosis, inflammation,NAS score

Improvement ininflammation and NASseen if weight loss �9%

2 Steatosis Animal data, psychiatricside effects

ce, 2 Steatosis Animal and pilot studies inNAFLD; extensivelystudied in type 2diabetes mellitus

ce, 2 Steatosis, inflammationand fibrosis

Side effects: weight gain,peripheral edema,cardiac, fractures, needfor maintenance therapy

tance �/� improvement insteatosis, inflammation,and fibrosis

Data conflicting; no RCTs

Uncertain Large trials with histologicfollow-up evaluationrequired

2 Steatosis, inflammation,fibrosis

Pilot study only

No change Not beneficial in large RCT2 Steatosis, inflammation Pilot study onlyUncertain Conflicting studies

2 Steatosis and fibrosis Animal data

2 Fibrosis Animal and pilot studies

test; PPAR, peroxisome proliferator–activated receptor; HMG CoA,

ffects

resis

e,, LFTs

istan

istan

resisectin

tion

AFLD and a patient’s physical fitness level.112

tmhod

pwTismtpovrilbrmlmtomiTstbN

dwcitpet

dwobmtiisp

rspirTrao

tlmmehiiledrar4aoes

pcpesTdslgbb

piaspp1kst

1690 TORRES AND HARRISON GASTROENTEROLOGY Vol. 134, No. 6

Collectively these studies support the general concepthat slow gradual weight loss through diet and exercise

ay prove beneficial in the treatment of NASH. Clearly,owever, larger RCTs are needed to better define theptimal exercise regimen as well as the most beneficialietary composition and total calorie intake.

Medical Therapies That Augment Weight LossOrlistat. The reversible inhibitor of gastric and

ancreatic lipase, orlistat, appears to be the most studiedeight loss medication used in the treatment of NASH.his medication is taken with meals and blocks approx-

mately 30% of dietary triglycerides. Early pilot workuggested this medication may be beneficial in the treat-

ent of NASH.113,114 Two subsequent randomized con-rolled trials were performed to further investigate theotentially beneficial effects of orlistat in the treatmentf NASH.115,116 Six to 9 months of therapy with orlistats placebo produced 8% vs 6% total body weight loss,espectively. Zelber–Sagi et al115 showed steatosis onlymproved in the orlistat group despite significant weightoss in each group. Harrison et al116 obtained repeat liveriopsies at 9 months and showed that regardless ofegimen, a 9% body weight reduction produced improve-

ent in biochemical markers, serum aminotransferaseevels, steatosis, and necroinflammation, but no improve-

ent in fibrosis. The implications of these 2 RCTs werehat weight loss is the common pathway through whichrlistat may exert a beneficial effect. Other weight lossedications are under investigation in search of a med-

cation with minimal side effects and excellent efficacy.hese data, combined with diet and exercise studies,

uggest that relatively modest amounts of weight loss, onhe order of 5%–10%, will lead to improvement in bothiochemical and histopathologic abnormalities seen inASH.

Rimonabant. Increasing evidence suggests the en-ocannabinoid system is enhanced in the setting of over-eight or obesity. Activation of cannabinoid (CB-1) re-

eptors located both centrally and peripherally (tonclude organs involved in energy homeostasis such ashe liver, adipocytes, and possibly skeletal muscle andancreas), lead to enhanced weight gain and altered en-rgy metabolism. Inhibition of these receptors has beenhe subject of recent intense investigation.

Although not specifically studied in NASH patients toate, several large randomized placebo-controlled trialsith the cannabinoid-1 antagonist, rimonabant, in bothverweight/obese nondiabetic and diabetic patients haveeen performed.117 Rimonabant given at a dose of 20g/day for 1 year is associated with a weight loss of 4.7

o 5.4 kg and improved waist circumference. Significantmprovement in serum triglyceride levels, HDL levels,nsulin resistance, and adiponectin levels also has beenhown and seem to be greater than what would be ex-

ected with weight loss alone, suggesting an added pe- l

ipheral effect. In addition, significant improvement inerum ALT levels, �-glutamyltransferase levels, alkalinehosphatase levels, and hepatic steatosis has been shown

n an animal model of NAFLD.118 Multicentered RCTs ofimonabant in human beings with NASH are underway.he recent association of this medication with psychiat-

ic issues to include depression may limit its future use,nd it is currently not available in the United Statesutside of clinical research trials.

Glucagon-like protein-1–receptor agonist (incre-in analogs). Peptides that are derived from glucagon-ike protein-1–receptor agonists such as exenatide also

ay prove to be potential therapeutic agents in the treat-ent of NASH. These incretin analogs have been studied

xtensively in patients with type 2 diabetes mellitus andave been found to promote insulin secretion, suppress

nappropriate glucagon secretion, slow gastric emptying,nduce satiety, and are associated with modest weightoss. Nausea occurs in up to 57% of patients, typicallyarly in the course of therapy, and can be mitigated withose titration.119 A recent study in obese mice showededuced markers of oxidative stress and insulin resistances well as decreased hepatic steatosis.120 A recent caseeport of a diabetic male patient with NAFLD treated for4 weeks with exenatide showed improvement in serumminotransferase levels and hepatic steatosis.121 Numer-us other studies have supported the glycemic benefits ofxenatide (exendin-4) with further well-designed studiespecific to NASH populations required.

Surgical Therapy for Weight LossUntil medical therapy is developed that effectively

romotes and sustains weight loss, surgical therapy willontinue to remain a viable alternative in select patientopulations. Surgical therapy for obesity has becomexceedingly popular in the past decade and has evolvedignificantly from the early malabsorptive surgeries.hese early surgeries such as the jejunoileal bypass pro-uced dramatic weight loss but were shown to causeignificant worsening of liver disease postoperative-y.122,123 Newer surgical techniques such as adjustableastric banding and Roux-en-Y gastric bypass are overalletter tolerated and have been shown to improve meta-olic syndrome, insulin resistance, and hepatic histology.

Adjustable gastric banding in 22 obese NASH patientsroduced an average 34-kg weight loss as well as signif-

cant improvement in hepatic steatosis, inflammation,nd fibrosis at repeat liver biopsy more than 2 years afterurgery.82 Roux-en-Y gastric bypass has been used toroduce even greater weight loss and its utility in im-roving NASH has been well studied. A recent study of9 morbidly obese NASH patients who lost a mean of 52g of body weight after Roux-en-Y gastric bypass showedignificant improvement in steatosis, lobular inflamma-ion, and fibrosis on average 21.4 months postoperative-

y.83 Another study of 70 patients undergoing bariatric

saaspoivebahtol

moshietm

itppdtors

iwipmsmcalt1gat

ip

an4tafimaippptmbtAapwficfmibwg

ttstff4uitpo

ambrbcewna

May 2008 DIAGNOSIS AND THERAPY OF NASH 1691

urgery showed a mean weight loss of 59% at 15 monthsfter surgery and 82% improvement in steatohepatitisnd 39% improvement in fibrosis.124 Others also havehown large metabolic and histologic improvementsost– gastric bypass with no patients showing worseningf hepatic fibrosis and 50%–75% of patients showing

mprovement in hepatic fibrosis.125,126 These studies pro-ide compelling evidence that bariatric surgery, usingither adjustable gastric banding or Roux-en-Y gastricypass, results in substantial weight loss, and is associ-ted with a marked improvement in biochemical andistologic parameters of NASH patients. Subsequently,hese invasive therapies should be considered in morbidlybese patients with other medical conditions that cumu-

atively would warrant the risks of surgical intervention.

Medical Therapies That ImproveMetabolic ProfilesIn a similar manner to the tests under develop-

ent using specific biomarkers to diagnose NASH basedn a common pathway of insulin resistance and oxidativetress, pharmacologic therapies have been evaluated thatave targeted these same pathways. As the steps involved

n the development of steatohepatitis and fibrosis arelucidated further, pharmacologic therapies may be bet-er directed to cause histologic and biochemical improve-

ent.

Insulin-Sensitizing MedicationsThiazolidinediones. Thiazolidinediones (TZDs)

mprove insulin resistance in skeletal muscle, adiposeissue, and the liver through their action as peroxisomeroliferator–activated receptor � agonists that increaselasma adiponectin levels and fatty acid oxidation andecrease fatty acid synthesis.127 These agents have beenhe most studied insulin sensitizers used in the treatmentf NASH. Pioglitazone and rosiglitazone are the 2 cur-ently available TZDs and both have been evaluated inmall prospective trials.

A preliminary study of 22 NASH patients (�50% withmpaired glucose tolerance or diabetes) treated for 48eeks with rosiglitazone showed overall improvement in

nsulin sensitivity and serum aminotransferase levels. Re-eat liver biopsy at 48 weeks showed significant improve-ent in necroinflammation, ballooning, and zone 3 peri-

inusoidal fibrosis with 45% (10 patients) no longereeting criteria for NASH. However, this study lacked a

ontrol arm and within 6 months posttherapy serumminotransferase levels were almost back to pretreatmentevels.128 A subsequent randomized, placebo-controlledrial in 63 NASH patients (32% with diabetes) treated for

year with rosiglitazone 8 mg/day showed significantlyreater improvement in steatosis compared with placebo,nd significantly less histologic progression in reference

o ballooning, inflammation, and fibrosis.129 However, t

mprovement in steatosis was not seen uniformly in allatients treated with rosiglitazone.Pioglitazone also was evaluated in several small pilot

nd proof-of-concept trials. An open-label trial in 18ondiabetic NASH patients treated with 30 mg/day for8 weeks showed overall histologic improvement in twohirds of patients with significant improvement in ste-tosis, cellular injury, parenchymal inflammation, andbrosis.130 Another pilot study compared pioglitazone 30g/day plus vitamin E 400 IU/day with vitamin E alone

nd showed improvement in steatosis in both groups butmprovement in hepatic inflammation and fibrosis in theioglitazone plus vitamin E arm only.131 A more recentlacebo-controlled study by Belfort et al132 in 55 NASHatients with impaired glucose tolerance or type 2 diabe-es mellitus provided evidence that therapy with 6

onths of a low-calorie diet plus pioglitazone improvediochemical and histologic parameters significantly morehan patients prescribed a low-calorie diet and placebo.lthough significant improvement was seen for steatosisnd inflammation compared with placebo, fibrosis im-rovement did not reach significance when comparedith the placebo group (P � .08). These preliminaryndings are supported by another randomized placebo-ontrolled trial that followed up 74 nondiabetic patientsor 12 months and showed significant improvement in

etabolic and histologic parameters, including fibrosis,n the pioglitazone group when compared with place-o.133 Interestingly, lobular inflammation and steatosisere not improved between groups and further investi-ation is necessary to understand these findings.

Although these studies provided some evidence thathe TZDs may be beneficial in the short term, the long-erm benefits remain to be seen. Lutchman et al134

howed that continued therapy may be required to main-ain histologic benefit. In this study of 21 patients treatedor 48 weeks with pioglitazone, liver biopsies were per-ormed at baseline, at end of therapy, and, in 9 patients,8 weeks after stopping therapy. In the patients whonderwent the 3 serial liver biopsies, initial improvement

n biochemical and histologic parameters were not main-ained at the 48-week posttherapy evaluation, with 7 of 9atients developing histopathologic evidence of NASHnce again, despite no change in average body weight.

The side-effect profile of the TZD class of medicationlso must be addressed. Modest weight gain of approxi-ately 2–3 kg has been shown routinely and appears to

e caused by increased peripheral fat deposition. Someetrospective studies also suggest that TZDs promoteone loss and may contribute to osteoporosis and in-reased fractures, particularly in postmenopausal wom-n.135 If limited durations of therapy were adequate, thisould be less of an issue, but if long-term therapy isecessary to maintain hepatic benefit, this must be betterddressed. Further investigation is required to defini-

ively address this issue.

truaiauadm2v

tadtYafaoT

pomotfsatmodSgssgsmps

wmahbmtcr

snmg

mpfmshbtGrnhaqttTim

daaaictcvmbaelt

otisaampulms

s

1692 TORRES AND HARRISON GASTROENTEROLOGY Vol. 134, No. 6

Moreover, a recent meta-analysis of diabetic male pa-ients taking rosiglitazone has suggested an increasedisk of cardiovascular events.136 Although this study tab-lated an increased risk of myocardial infarction of 43%nd cardiovascular death of 64%, it has been widely crit-cized for its inclusion criteria, methods of statisticalnalysis, and overall low event rate.137 Other analysessing broader inclusion criteria and different statisticalnalysis failed to show a significant increase in myocar-ial infarction in this same patient population. Further-ore, a similar meta-analysis with pioglitazone showed a

0% reduction in cardiovascular events.138 Additional in-estigation is required to address the issue definitively.

The risk of congestive heart failure exacerbation withhe use of TZDs also has been reported and has beenttributed to TZD-related fluid retention and diastolicysfunction in susceptible individuals. Subsequentlyhese agents are contraindicated in patients with Nework Heart Classification stage 3– 4 heart failure. Lago etl139 recently showed an increased risk of congestive heartailure, but did not show increased cardiovascular deathmong patients taking TZDs. Large prospective trials arengoing and their results will determine the future ofZDs in the treatment of NASH.

Metformin. Metformin is a biguanide that im-roves insulin sensitivity by decreasing hepatic glucone-genesis and limiting triacylglycerol production. Animalodel data has been promising, with early studies of

bese mice given metformin showing improved serumransaminase levels, hepatomegaly, and steatosis.140 Datarom human studies are limited. Marchensini et al141 firsthowed improvement in insulin sensitivity and serumminotransferase levels in an open-label trial of 20 pa-ients treated with metformin 1500 mg/day for 4

onths. This was followed by a 6-month randomizedpen-label trial of metformin 850 mg twice daily plusiet vs dietary restriction of 1600 to 1800 kcal/day.142

ignificant improvement was seen in the metforminroup compared with the diet-alone group. However, noignificant improvement in necroinflammation or fibro-is was found. A similar open-label trial of metforminiven for 1 year in 15 patients showed improvement inerum aminotransferase levels and insulin sensitivity at 3

onths, but return of serum aminotransferase levels toretreatment levels were seen by the completion of thetudy.143

A subsequent open-label RCT in 110 NAFLD patientsas performed in Italy that compared metformin treat-ent with 2 control groups of vitamin E 800 IU/day orprescriptive, weight-reducing diet.144 Metformin showed

igher rates of aminotransferase normalization. Repeat liveriopsy in 31% of the metformin group showed improve-ent in steatosis, necroinflammation, and fibrosis, al-

hough there was no histologic follow-up evaluation in theontrol groups. Interestingly, metformin compared with

osiglitazone in a study of 20 patients with type 2 diabetes t

howed improvement in insulin resistance and transami-ase levels in both groups but decreased hepatic fat aseasured by proton spectroscopy in the rosiglitazone

roup.145

Overall, the insulin-sensitizing medications show theost promise to date in improving histopathology in

atients with NASH. Current data suggest that met-ormin may be less efficacious than the TZD class of

edications; however, randomized, placebo-controlledtudies with metformin using histopathologic end pointsave yet to be performed. Interestingly, one differenceetween the 2 classes of insulin sensitizers is the ability ofhe TZD class to up-regulate adiponectin expression.iven the multiple effects of adiponectin including up-

egulation of adenosine 5=-monophosphate (AMP) ki-ase and peroxisome proliferator–activated receptor-� in-ibition, the inability of metformin to positively impactdiponectin expression may have important conse-uences on the overall effectiveness of this medication inhe treatment of NASH, at least when used as mono-herapy. Combination therapy with metformin and theZD class of medications deserves consideration in clin-

cal trials because the untoward weight gain with TZDonotherapy may be mitigated.

Antioxidant therapy. The depletion of antioxi-ants within hepatocytes resulting in impaired ROS in-ctivation is the basis for antioxidant supplementation as

potential treatment for NASH.146 The lipid-solublentioxidant, � tocopherol (vitamin E), has been shown tonhibit lipid peroxidation and suppress inflammatoryytokines such as tumor necrosis factor-�, and its use inhe treatment of NASH has been studied. One studyomparing 6 months of 1000 IU vitamin E and 1000 mgitamin C daily vs placebo showed significant improve-ent in hepatic fibrosis within the vitamin group but not

etween groups and no change in serum aminotransfer-se levels or hepatic inflammation.147 Alternatively, oth-rs have shown improvement in serum aminotransferaseevels and insulin resistance although follow-up histopa-hology was not assessed.148,149

Betaine, another antioxidant studied in the treatmentf NASH, increases S-adenosylmethionine levels and ishought to reduce fat deposition in the liver.143 Markedmprovement in serum aminotransferase levels, hepaticteatosis, inflammatory activity, and fibrosis was seen in

pilot trial of 10 NASH patients treated with betainenhydrous in a 20-mg divided dose daily for 12onths.150 Rigorous follow-up studies are required with

articular attention paid to the formulation of betainesed because betaine anhydrous is a medication regu-

ated by the Food and Drug Administration whereas theore readily available betaine hydrochloride is a dietary

upplement found in nutrition stores.151

Phlebotomy targeting hepatic iron overload with re-ultant increased oxidative stress is another potential

herapeutic modality that has been investigated in the

tullprnhLnm

ctRba

trt(pctcewwdmo

eTgftnru

tlmlidbcuthN

irpomtbe

absiLAliw

mNicsrfcparltraat

octNbaarrCilmet

May 2008 DIAGNOSIS AND THERAPY OF NASH 1693

reatment of NASH. In a study of 64 NAFLD patientsndergoing therapeutic phlebotomy to a goal of a ferritin

evel of less than 80 ng/mL for 8 months in addition toifestyle modification vs lifestyle modification alone, thehlebotomy group showed significantly improved insulinesistance.152 Serum aminotransferase levels were not sig-ificantly different between the 2 groups and hepaticistology was not assessed in this preliminary study.arger studies with histologic follow-up evaluation areecessary to determine the utility of this treatmentodality.Overall, antioxidants represent a novel class of medi-

ations that have shown some promising initial results inhe treatment of NASH but further well-designed largeCTs are required and it is likely that these therapies wille used in combination regimens with other agents thatffect insulin sensitivity.

OthersCytoprotective agents. Cytoprotective agents are

hought to work by preventing apoptosis and down-egulating the inflammatory cascade that has been showno occur in NAFLD patients. Ursodeoxycholic acidUDCA) is the most studied and initial pilot trials provedromising as a NASH-specific therapy. However, a well-onducted, multicenter, double-blind, placebo-controlledrial of UDCA for 2 years failed to show significanthange in biochemical or histologic parameters.153 Inter-stingly, Dufour et al154 showed biochemical benefit asell as reduced steatosis with the combination of UDCAith vitamin E compared with UDCA plus placebo orouble-placebo groups after 2 years of therapy.154 It re-ains to be seen whether UDCA in combination with

ther medications may prove clinically efficacious.Pentoxifylline, another cytoprotective agent, exerts its

ffect through inhibition of tumor necrosis factor-�.welve months of pentoxifylline 400 mg 3 times dailyiven in 2 recent pilot trials improved serum aminotrans-erase levels as well as steatosis and lobular inflamma-ion.155,156 Although no dangerous side effects have beenoted, the high rates of nausea with this medicationesulting in a 50% drop-out rate may limit the clinicaltility of this medication.

3-hydroxy-3-methylglutaryl-coenzyme A reduc-ase therapy. Medications that exert beneficial effects onipid profiles offer another potential therapeutic class of

edications. The most studied of this family of lipid-owering medicines are the statins, which work throughnhibition of 3-hydroxy-3-methylglutaryl-coenzyme A re-uctase, which catalyses an essential step in cholesteroliosynthesis. Initially statins were noted to cause in-reases in hepatic transaminase levels in patients whensed to treat hyperlipidemia, and therefore were thoughto be harmful to liver function. However, recent studiesave shown statins to be safe when given to patients with

AFLD.157,158 An early pilot trial with atorvastatin raised T

nterest in using statins to treat NASH,159 but a moreecent study with simvastatin showed no affect on adi-onectin or improvement in insulin sensitivity in a groupf patients with metabolic syndrome.160 Questions re-ain in reference to the specific mechanisms of action of

hese drugs and it may be that they have positive effectseyond altering insulin sensitivity or adipocytokine lev-ls, but these studies have yet to be performed.

Ezetimibe. Ezetimibe (Zetia, Kenilworth, NJ) isnother lipid-lowering medication that may have hepaticenefits in the treatment of NASH. This medicationelectively inhibits intestinal cholesterol absorption andn controlled studies with statins appears to lower serumDL cholesterol by 24% as well as triglycerides by 16%.161

recent study by Deushi et al162 showed improved insu-in resistance and decreased lipid deposition and fibrosisn obese rats treated with ezetimibe. No clinical trialsith this therapy have been published.

Angiotensin-receptor blockers. A novel class ofedications that have been evaluated in the treatment ofASH are the angiotensin-receptor blockers. These med-

cations, which have been used to treat hypertension andongestive heart failure, have been shown to have liver-pecific beneficial effects. Obese mice given angiotensin-eceptor blockers showed improved serum aminotrans-erase levels and reduced expression of inflammatoryytokines and hepatic fibrosis through inhibition of he-atic stellate cell activity.163 Two human pilot studieslso have been completed and suggest that angiotensin-eceptor blockers may improve serum aminotransferaseevels and provide histologic benefit in NASH pa-ients.164,165 Similar to most of the research conducted inegards to NASH therapy, large placebo-controlled RCTsre required to ascertain if any substantial benefit can bechieved with long-term angiotensin-receptor blockerherapy in NASH patients.

ConclusionsThe myriad of public health issues related to the

besity epidemic continues to grow as more medicalonditions are linked to obesity, insulin resistance, andhe metabolic syndrome. The growing incidence ofAFLD, and subsequently NASH, almost certainly will

e reflected in subsequent increases in cirrhosis and hep-tocellular carcinoma in the future. Noninvasive andccurate screening tests to correctly identify patients atisk for disease progression as well as effective treatmentegimens must be developed for these high-risk patients.urrently, there is no ideal treatment regimen but those

nterventions that produce sustained substantial weightoss such as bariatric surgery seem to provide the single

ost benefit. Modest but sustained weight loss, regularxercise, and diet composition modification also appearo ameliorate biochemical and histologic abnormalities.

Thus far, medical therapies with the exception of the

ZDs have not been subjected to randomized, placebo-

cntlumramlthp

1694 TORRES AND HARRISON GASTROENTEROLOGY Vol. 134, No. 6

ontrolled trials and have shown only modest prelimi-ary benefits. The TZDs, although appearing to be effec-ive in reducing steatosis, inflammation, and fibrosis areimited by small proof-of-concept trials and potentiallyntoward effect profiles as well as an apparent need foraintenance therapy. Other therapies directed at insulin

esistance, oxidative stress, cytoprotection, and fibrosislso may offer benefit, but further study is required. Aultifaceted approach that relies on lifestyle changes,

imited weight loss, and pharmacotherapy requires fur-her investigation because no one treatment approachas proved universally applicable to the large generalopulation with NASH.

References

1. Beymer C, Kowdley KV, Larson AA, et al. Prevalence and predic-tors of asymptomatic liver disease in patients undergoing gas-tric bypass surgery. Arch Surg 2003;138:1240–1244.

2. Gholam PM, Flanebaum L, Machan JT, et al. Nonalcoholic fattyliver disease in severely obese subjects. Am J Gastroenterol2007;102:399–408.

3. Machado M, Marques-Vidal P, Cortez-Pinto H. Hepatic histologyin obese patients undergoing bariatric surgery. J Hepatol 2006;45:600–606.

4. Adams LA, Lymp JF, Sauver JS, et al. The natural history ofnonalcoholic fatty liver. Gastroenterology 2005;129:113–121.

5. Targher G, Bertolini L, Poli F, et al. Nonalcoholic fatty liverdisease and risk of future cardiovascular events among type 2diabetic patients. Diabetes 2005;54:3541–3546.

6. Hamaguchi M, Kojima T, Takeda N, et al. Nonalcoholic fatty liverdisease is a novel predictor of cardiovascular disease. World JGastroenterol 2007;13:1579–1584.

7. Arslan U, Turkoglu S, Balcioglu S, et al. Association betweennonalcoholic fatty liver disease and coronary artery disease.Coron Artery Dis 2007;18:433–436.

8. Liangpunsakul S, Chalasani N. Is hypothyroidism a risk factor fornon-alcoholic steatohepatitis? J Clin Gastroenterol 2003;37:340–343.

9. Harrison SA, Torgerson S, Hayashi P. The natural history ofnonalcoholic fatty liver disease: a clinical histopathologicalstudy. Am J Gastroenterol 2003;98:2042–2047.

10. Wanless IR, Lentz JS. Fatty liver hepatitis (steatohepatitis) andobesity: an autopsy study with analysis of risk factors. Hepatol-ogy 1990;12:1106–1110.

11. Harrison SA, Neuschwander-Tetri BA. Nonalcoholic fatty liverdisease and nonalcoholic steatohepatitis. Clin Liver Dis 2004;8:861–879.

12. Adams LA, Sanderson S, Lindor KD, et al. The histologicalcourse of nonalcoholic fatty liver disease: a longitudinal study of103 patients with sequential liver biopsies. J Hepatol 2005;42:132–138.

13. Pantsari M, Harrison SA. Nonalcoholic fatty liver disease pre-senting with an isolated elevated alkaline phosphatase. J ClinGastroenterol 2006;40:633–635.

14. Palmentieri B, de Sio I, La Mura V, et al. The role of bright liverecho pattern on ultrasound B-mod examination in the diagnosisof liver steatosis. Dig Liver Dis 2006;38:485–489.

15. Hamaguchi M, Kojima T, Itoh Y, et al. The severity of ultrasoundfindings in nonalcoholic fatty liver disease reflects the metabolicsyndrome and visceral fat accumulation. Am J Gastroenterol2007;102:1–8.

16. Mottin CC, Moretto M, Padoin AV, et al. The role of ultrasoundin the diagnosis of hepatic steatosis in morbidly obese patients.

Obes Surg 2004;14:635–637.

17. Celle G, Savarino V, Picciotto A, et al. Is hepatic ultrasonographya valid alternative tool to liver biopsy? Report on 507 casesstudied with both techniques. Dig Dis Soc 1988;33:467–471.

18. Saadeh S, Younossi ZM, Remer EM, et al. The utility of radio-logical imaging in nonalcoholic fatty liver disease. Gastroenter-ology 2002;123:745–750.

19. Park SH, Kim PN, Kimg KW, et al. Macrovesicular hepatic ste-atosis in living liver donors: use of CT for quantitative andqualitative assessment. Radiology 2006;239:105–112.

20. Duman DG, Celikel C, Tuney D, et al. Computed tomography innonalcoholic fatty liver disease: a useful tool for hepatosteato-sis assessment? Dig Dis Sci 2006;51:346–351.

21. Lee SW, Park SH, Kim KW, et al. Unenhanced CT for assess-ment of macrovesicular hepatic steatosis in living liver donors:comparison of visual grading with liver attenuation index. Radi-ology 2007;244:479–485.

22. Fishbein MK, Gardner KG, Potter CJ, et al. Introduction of fastMR imaging in the assessment of hepatic steatosis. MagnReson Imaging 1997;15:287–293.

23. Charatcharoenwitthaya P, Lindor KD. Role of radiologic modali-ties in the management of nonalcoholic steatohepatitis. ClinLiver Dis 2007;11:37–54.

24. Szczepaniak LS, Nurenberg P, Leonard D, et al. Magnetic reso-nance spectroscopy to measure hepatic triglyceride content:prevalence of hepatic steatosis in the general population. Am JPhysiol 2005;288:E462–E468.

25. Cheung O, Kapoor A, Puri P, et al. The impact of fat distributionon the severity of nonalcoholic fatty liver disease and metabolicsyndrome. Hepatology 2007;46:1091–1100.

26. Haukeland JW, Konopski Z, Linnestad P, et al. Abnormal glu-cose tolerance is a predictor of steatohepatitis and fibrosis inpatients with non-alcoholic fatty liver disease. Scand J Gastro-enterol 2005;40:1469–1477.

27. Balkau B. New diagnostic criteria for diabetes and mortality inolder adults. DECODE study group, European Diabetes Epide-miology Group. Lancet 1999;353:68–69.

28. Marchesini G, Bugianesi E, Forlani G, et al. Non-alcoholic ste-atohepatitis in patients cared in metabolic units. Diabetes ResClin Pract 2004;63:143–151.

29. Marchesini G, Bugianesi E, Forlani G, et al. Nonalcoholic fattyliver, steatohepatitis, and the metabolic syndrome. Hepatology2003;37:917–923.

30. Grundy SM, Cleeman JL, Daniels SR, et al. Diagnosis andmanagement of the metabolic syndrome. Circulation 2005;112:2735–2752.

31. Brunt EM, Janney CG, DiBisceglie AM, et al. Nonalcoholic ste-atohepatitis: a proposal for grading and staging the histologicallesions. Am J Gastroenterol 1999;94:2467–2474.

32. Kleiner DE, Brunt EM, Van Natta M, et al. Design and validationof a histological scoring system for nonalcoholic fatty liver dis-ease. Hepatology 2005;41:1313–1321.

33. Ratziu V, Bugianesi E, Dixon J, et al Histological progression ofnon-alcoholic fatty disease: a critical reassessment based onliver sampling variability. Aliment Pharmacol Ther 2007;26:821–830.

34. Iijima H, Moriyasu F, Tsuchiya K, et al. Decrease in accumula-tion of ultrasound contrast microbubbles in non-alcoholic ste-atohepatitis. Hepatol Res 2007;37:722–730.

35. Yoneda M, Fujita K, Inamori M, et al. Transient elastography inpatients with non-alcoholic fatty liver disease (NAFLD). Gut2007;56:1330–1331.

36. Talwalkar JA, Kurtz DM, Schoenleber SJ, et al. Ultrasound-basedtransient elastography for the detection of hepatic fibrosis:systematic review and meta-analysis. Clin Gastroenterol Hepa-

tol 2007;5:1214–1220.

May 2008 DIAGNOSIS AND THERAPY OF NASH 1695

37. Fraquelli M, Rigamonti C, Casazza G et al. Reproducibility oftransient elastography in the evaluation of liver fibrosis in pa-tients with chronic liver disease. Gut 2007;56:968–973.

38. Yin M, Talwalkar JA, Glaser KJ, et al. Assessment of hepaticfibrosis with magnetic reasonance elastography. Clin Gastroen-terol Hepatol 2007;5:1209–1213.

39. Sanyal AJ, Campbell-Sargent C, Mirshahi F, et al. Nonalcoholicsteatohepatitis: association of insulin resistance and mitochon-drial abnormalities. Gastroenterology 2001;120:1183–1192.

40. Chalasani N, Deeg MA, Crabb DW. Systemic levels of lipidperoxidation and its metabolic and dietary correlates in patientswith nonalcoholic steatohepatitis. Am J Gastroenterol 2004;99:1497–1502.

41. Bonnefont-Rousselot D, Ratziu V, Giral P. Blood oxidative stressmarkers are unreliable makers of hepatic steatosis. AlimentPharmacol Ther 2006;23:91–98.

42. Tsochatzis E, Papatheodororidis GV, Archimandritis AJ. Theevolving role of leptin and adiponectin in chronic liver diseases.Am J Gastroenterol 2006;101:1–12.

43. Anania FA, Parekh S. Abnormal lipid and glucose metabolism inobesity: implications for nonalcoholic fatty liver disease. Gas-troenterology 2007;132:2191–2207.

44. Uygun A, Kadayifci A, Yesilova Z, et al. Serum leptin levels inpatients with nonalcoholic steatohepatitis. Am J Gastroenterol2000;95:3584–3589.

45. Chitturi S, Farrell G, Forst L, et al. Serum leptin in NASH corre-lates with hepatitis steatosis but not fibrosis: a manifestation oflipotoxicity? Hepatology 2002;36:403–409.

46. Nakao K, Nakata K, Ohtsubo N, et al. Association betweennonalcoholic fatty liver, markers of obesity, and serum leptinlevels in young adults. Am J Gastroenterol 2002;97:1796–1801.

47. Chalasani N, Crabb DW, Cummings OW, et al. Does leptin playa role in the pathogenesis of human nonalcoholic steatohepa-titis? Am J Gastroenterol 2003;98:2771–2776.

48. Hui JM, Hodge A, Farrell GC, et al. Beyond insulin resistance inNASH: TNF-alpha or adiponectin? Hepatology 2004;40:46–54.

49. Musso G, Gambino R, Biroli G, et al. Hypoadiponectinemiapredicted the severity of hepatic fibrosis and pancreatic beta-cell dysfunction in nondiabetic nonobese patients with nonalco-holic steatohepatitis. Am J Gastroenterol 2005;100:2438–2446.

50. Bugianesi E, Pagotto U, Manini R, et al. Plasma adiponectin innonalcoholic fatty liver is related to hepatic insulin sensitivityand hepatic fat content of liver, not to liver disease severity.J Clin Endocrinol Metab 2005;90:3498–3504.

51. Kaser S, Moschen A, Canyon A, et al. Adiponectin and itsreceptors in nonalcoholic steatohepatitis. Gut 2005;54:117–121.

52. Shimada M, Kawahara H, Ozaki K, et al. Usefulness of a com-bined evaluation of the serum adiponectin level, HOMA-IR, andserum type IV collagen 7S level to predict the early stage ofnonalcoholic steatohepatitis. Am J Gastroenterol 2007;102:1931–1938.

53. Vuppalanchi R, Marri S, Kolwankar D, et al. Is adiponectininvolved in the pathogenesis of nonalcoholic steatohepatitis? Apreliminary human study. J Clin Gastroenterol 2005;39:237–242.

54. Targher G, Bertolini L, Rodella S, et al. Associations betweenplasma adiponectin concentrations and liver histology in pa-tients with nonalcoholic fatty liver disease. Clin Endocrinol2006;64:679–683.

55. Hui JM, Farrell GC, Kench JG, et al. High sensitivity C-reactiveprotein values do not reliably predict the severity of histologicalchanges in NAFLD. Hepatology 2004;39:1458–1459.

56. Yoneda M, Mawatari H, Fujita K, et al. High sensitivity C-reactive

protein is an independent clinical feature of nonalcoholic ste-

atohepatitis (NASH) and also of the severity of fibrosis in NASH.J Gastroenterol 2007;42:573–582.

57. Targher G. Relationship between high-sensitivity C-reactive pro-tein levels and liver histology in subjects with non-alcoholic fattyliver disease. J Hepatol 2006;45:879–882.

58. Feldstein AE, Canbay A, Angulo P, et al. Hepatic apoptosis andFas expression are prominent features of human nonalcoholicsteatohepatitis. Gastroenterology 2003;125:437–443.

59. Wieckowska A, Zein NN, Yerian LM, et al. In vivo assessment ofliver cell apoptosis as a novel biomarker of disease severity innonalcoholic fatty liver disease. Hepatology 2006;44:27–33.

60. Yilmaz Y, Dolar E, Ulakaya E, et al. Soluble forms of extracellularcytokeratin 18 may differentiate simple steatosis from nonalco-holic steatohepatitis. World J Gastroenterol 2007;13:837–844.

61. Mitry RR, De Bruyne R, Quaglia A, et al. Noninvasive diagnosisof nonalcoholic fatty liver disease using serum biomarkers.Hepatology 2007;46:2047–2048.

62. Palekar NA, Naus R, Larson SP, et al. Clinical model for distin-guishing nonalcoholic steatohepatitis from simple steatosis inpatients with nonalcoholic fatty liver disease. Liver Int 2005;26:151–156.

63. Poynard T, Ratziu V, Charlotte F, et al. Diagnostic value ofbiochemical markers (NASHTest) for the prediction of nonalco-hol steatohepatitis in patients with non-alcoholic fatty liver dis-ease. BMC Gastroenterol 2006;6:34.

64. Wieckowska A, McCullough AJ, Feldstein AE. Noninvasive diag-nosis and monitoring of nonalcoholic steatohepatitis: presentand future. Hepatology 2007;46:582–589.

65. Ratziu V, Giral P, Charlotte F, et al. Liver fibrosis in overweightpatients. Gastroenterology 2000;118:1117–1123.

66. Ratziu V, Massard J, Charlotte F, et al. Diagnostic value ofbiochemical markers (FibroTest-FibroSURE) for the prediction ofliver fibrosis in patients with non-alcoholic fatty liver disease.BMC Gastroenterol 2006;6:6.

67. Rosenberg WM, Voelker M, Thiel R, et al. Serum markers detectthe presence of liver fibrosis: a cohort study. Gastroenterology2004;127:1704–1713.

68. Angulo P, Hui JM, Marchesini G, et al. The NAFLD fibrosis score:a noninvasive system that identifies liver fibrosis in patientswith NAFLD. Hepatology 2007;45:846–854.

69. Kang H, Omo J, Joel G, et al. Validation of the NAFLD fibrosisscore in predicting advanced fibrosis in patients with histologi-cally proven non-alcoholic fatty liver disease. Hepatology 2007;46:750A.

70. Guha IN, Parkes J, Roderick PR, et al. Non-invasive markersassociated with liver fibrosis in non-alcoholic fatty liver disease.Gut 2006;55:1650–1660.

71. Capristo E, Miele L, Forgione A, et al. Nutritional aspects inpatients with non-alcoholic steatohepatitis (NASH). Eur Rev MedPharmacol Sci 2005;9:265–268.

72. Sato F, Tamura Y, Watada H, et al. Effects of diet-inducedmoderate weight reduction on intrahepatic and intramyocellulartriglycerides and glucose metabolism in obese subjects. J ClinEndocrinol Metab 2006;92:3326–3329.

73. Palmer M, Schaffner F. Effect of weight reduction on hepaticabnormalities in overweight patients. Gastroenterology 1990;99:1408–1413.

74. Anderson T, Gluud C, Franzmann MB, et al. Hepatic effects ofdietary weight loss in morbidly obese subjects. J Hepatol 1991;12:224–229.

75. Okita M, Hayashi M, Sasagawa T, et al. Effect of moderatelyenergy-restricted diet on obese patients with fatty liver. Nutrition2001;17:542–547.

76. Thomas EL, Brynes AE, Hamilton G, et al. Effect of nutritional

counseling on hepatic, muscle, and adipose fat tissue content

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1696 TORRES AND HARRISON GASTROENTEROLOGY Vol. 134, No. 6

and distribution in non-alcoholic fatty liver disease. World JGastroenterol 2006;12:5813–5819.

77. De Luis DA, Aller R, Izaola O, et al. Effect of a hypocaloric diet intransaminases in nonalcoholic fatty liver disease and obesepatients, relation with insulin resistance. Diabetes Res ClinPract 2008;79:74–78.

78. Luychkx FH, Desaive C, Thiry A, et al. Liver abnormalities inseverely obese subjects: effect of drastic weight loss aftergastroplasty. Int J Obes Relat Metab Disord 1998;22:222–226.

79. Kral JG, Thung SN, Biron S, et al. Effects of surgical treatmentof the metabolic syndrome on liver fibrosis and cirrhosis. Sur-gery 2004;135:48–58.

80. Fisler JS, Drenick EJ. Starvation and semistarvation diets in themanagement of obesity. Ann Rev Nutr 1987;7:465–484.

81. de Almeida SR, Rocha PR, Sanches MD, et al. Roux-en-Y gastricbypass improves the nonalcoholic steatohepatitis (NASH) ofmorbid obesity. Obes Surg 2006;16:270–278.

82. Dixon JB, Bhathal PS, O’Brien PE. Weight loss and non-alcoholicfatty liver disease: falls in gamma-glutamyl transferase concen-trations are associated with histologic improvement. Obes Surg2006;16:1278–1286.

83. Barker KB, Palekar NA, Bowers SP, et al. Non-alcoholic steato-hepatitis: effect of Roux-en-Y gastric bypass surgery. Am J Gas-troenterol 2006;101:368–373.

84. Clark JM, Alkhuraishi AR, Solga SF, et al. Roux-en-Y gastricbypass improves liver histology in patients with non-alcoholicfatty liver disease. Obes Res 2005;13:1180–1186.

85. Huang MA, Greenson JK, Chao C, et al. One-year intense nutri-tional counseling results in histological improvement in patientswith nonalcoholic steatohepatitis: a pilot study. Am J Gastroen-terol 2005;100:1072–1081.

86. Wang RT, Koretz RL, Yee HF. Is weight reduction an effectivetherapy for nonalcoholic fatty liver disease? A systematic re-view. Am J Med 2003;115:554–559.

87. Musso G, Gambino R, De Michieli F, et al. Dietary habits andtheir relations to insulin resistance and postprandial lipemia innonalcoholic steatohepatitis. Hepatology 2003;37:909–916.

88. Toshimitsu K, Matsuura B, Ohkubo I, et al. Dietary habits andnutrient intake in non-alcoholic steatohepatitis. Nutrition 2007;23:46–52.

89. Yancy WS, Olsen MK, Guyton JR, et al. A low-carbohydrate,ketogenic diet versus a low fat diet to treat obesity and hyper-lipidemia. Ann Intern Med 2004;140:769–777.

90. Tendler D, Lin S, Yancy WS, et al. The effect of low carbohy-drate, ketogenic diet on nonalcoholic fatty liver disease: a pilotstudy. Dig Dis Sci 2007;52:589–593.

91. Westerback J, Lammi K, Hakkienen A, et al. Dietary fat contentmodifies liver fat in overweight nondiabetic subjects. J ClinEndocrinol Metab 2005;90:2804–2809.

92. Petersen KF, Dufour S, Befroy D, et al. Reversal of nonalcoholichepatic steatosis, hepatic insulin resistance, and hyperglyce-mia by moderate weight reduction in patients with type 2 dia-betes. Diabetes 2005;54:603–608.

93. Zivkovic AM, German JB, Sanyal AJ. Comparative review of dietsfor the metabolic syndrome: implications for nonalcoholic fattyliver disease. Am J Clin Nutr 2007;86:282–300.

94. El-Badry AM, Moritz W, Contaldo C, et al. Prevention of reperfu-sion injury and microcirculatory failure in macrosteatotic mouseliver by omega-3 fatty acids. Hepatology 2007;45:841–845.

95. Wang D, Wei Y, Pagliassotti MJ. Saturated fatty acids promoteendoplasmic reticulum stress and liver injury in rats with hepaticsteatosis. Endocrine 2006;147:943–951.

96. Paniagua JA, de la Sancristana G, Romero I, et al. Monounsat-urated fat-rich diet prevents central body fat distribution anddecreases postprandial adiponectin expression induced by acarbohydrate-rich diet in insulin resistant subjects. Diabetes

Care 2007;30:1717–1723.

97. Laaksonen De, Nyyssonen K, Niskanen L, et al. Prediction ofcardiovascular mortality in middle-aged men by dietary and se-rum linoleic and polyunsaturated fatty acids. Arch Intern Med2005;165:193–199.

98. Sekiya M, Yahagi N, Matsuzaka T, et al. Polyunsaturated fattyacids ameliorate hepatic steatosis in obese mice by SREBP-1suppression. Hepatology 2003;38:1529–1539.

99. Cortez-Pinto H, Jesus L, Barros H, et al. How different is thedietary pattern in non-alcoholic steatohepatitis patients? ClinNutr 2006;25:816–823.

00. Capanni M, Calella F, Biagini, et al. Prolonged n-3 polyunsatu-rated fatty acid supplementation ameliorates hepatic steatosisin patients with non-alcoholic fatty liver disease: a pilot study.Aliment Pharmacol Ther 2006;23:1143–1151.

01. Spadaro L, Magliocco O, Spampinato D, et al. Omega-3 polyun-saturated fatty acids: a pilot trial in non-alcoholic fatty liverdisease. J Hepatol 2006;44:S264.

02. Tapsell LC, Gillen LJ, Patch CS, et al. Including walnuts in alow-fat/modified-fat diet improves HDL cholesterol-to-total cho-lesterol ratios in patients with type 2 diabetes. Diabetes Care2004;27:2777–2783.

03. Lopez-Garcia E, Schulze MB, Meigs JB, et al. Consumption oftrans fatty acids is related to plasma biomarkers of inflamma-tion and endothelial dysfunction. J Nutr 2005;135:562–566.

04. Mozaffarian D. Trans fatty acids—effects on systemic inflam-mation and endothelial dysfunction. Atheroscler Suppl 2006;7:29–32.

05. Zelber-Sagi S, Nitzan-Kaluski D, Goldsmith R, et al. Long termnutritional intake and the risk for non-alcoholic fatty liver dis-ease (NAFLD): a population based study. J Hepatol 2007;47:711–717.

06. Le KA, Tappy L, D’Alessio DA. Mitochondrial dysfunction andinsulin resistance: a matter of lifestyle? Curr Opin Nutr MetabCare 2007;10:494–497.

07. Scribner KB, Pawlak DB, Ludwig DS. Hepatic steatosis andincreased adiposity in mice consuming rapidly versus slowlyabsorbed carbohydrate. Obesity 2007;15:2190–2199.

08. Harrison SA, Day CP. Benefits of lifestyle modification in NAFLD.Gut 2007;20:29–43.

09. Suzuki A, Lindor K, St Saver J, et al. Effect of changes on bodyweight and lifestyle in nonalcoholic fatty liver disease. J Hepatol2005;43:1060–1066.

10. Ueno T, Sugawara H, Sujaku K, et al. Therapeutic effects ofrestricted diet and exercise in obese patients with fatty liver.J Hepatol 1997;27:103–107.

11. Hickman IJ, Jonsson JR, Prins JB, et al. Modest weight loss andphysical activity in overweight patients with chronic liver diseaseresults in sustained improvements in alanine aminotransfer-ase, fasting insulin, and quality of life. Gut 2004;53:413–419.

12. Church TS, Kuk JL, Ross R, et al. Association of cardiorespira-tory fitness, body mass index, and waist circumference to non-alcoholic fatty liver disease. Gastroenterology 2006;130:2023–2030.

13. Harrison SA, Fincke C, Helinski D, et al. A pilot study of orlistattreatment in obese, non-alcoholic steatohepatitis patients. Ali-ment Pharmacol Ther 2004;20:623–628.

14. Assy N, Hussein O, Abassi Z. Weight loss induced by orlistatreverses fatty infiltration and improves hepatic fibrosis in obesepatients with non-alcoholic steatohepatitis. Gut 2007;56:443–444.

15. Zelber-Sagi S, Kessler A, Brazowsky, et al. A double-blind ran-domized placebo controlled trial of orlistat for treatment ofnonalcoholic fatty liver disease. Clin Gastroenterol Hepatol2006;4:639–644.

16. Harrison SA, Brunt EM, Fecht WJ, et al. Orlistat in the treatment

of overweight patients with nonalcoholic steatohepatitis

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

May 2008 DIAGNOSIS AND THERAPY OF NASH 1697

(NASH): a multi-centered, randomized prospective trial.Gastroenterology 2007;132:T1059.

17. Scheen AJ. Cannabinoid-1 receptor antagonists in type-2 diabe-tes. Best Pract Res Clin Endocrinol Metab 2007;21:535–553.

18. Gary-Bobo M, Elachouri G, Gallas JF, et al. Rimonabant reducesobesity-associated hepatic steatosis and features of metabolicsyndrome in obese zucker fa/fa rats. Hepatology 2007;46:122–129.

19. Amori RE, Lau J, Pittas AG. Efficacy and safety of incretintherapy in type 2 diabetes mellitus. JAMA 2007;298:194–206.

20. Ding X, Saxena NK, Lin S, et al. Exendin-4, a glucagon-likeprotein-1 (GLP-1) receptor agonist, reverses hepatic steatosisin ob/ob mice. Hepatology 2006;43:173–181.

21. Tushuizen ME, Bunck MC, Pouwels PJ, et al. Incretin mimeticsas a novel therapeutic option for hepatic steatosis. Liver Int2006;26:1015–1017.

22. Marubbio AT, Buchwald H, Schwartz MZ, et al. Hepatic lesionsof central pericellular fibrosis in morbid obesity, and after jeju-noileal bypass. Am J Clin Pathol 1976;66:684–691.

23. Campbell JM, Hunt TK, Karam JH, et al. Jejunoileal bypass as atreatment of morbid obesity. Arch Intern Med 1976;137:602–610.

24. Mattar SG, Velcu LM, Rabinovitz M, et al. Surgically-inducedweight loss significantly improved nonalcoholic fatty liver dis-ease and the metabolic syndrome. Ann Surg 2005;242:610–620.

25. Furuya CK, de Olivereira CP, de Mello ES, et al. Effects ofbariatric surgery on nonalcoholic fatty liver disease: preliminaryfindings after 2 years. J Gastroenterol Hepatol 2007;22:510–514.

26. Liu X, Lazenby AJ, Clements RH, et al. Resolution of nonalco-holic steatohepatitis after gastric bypass surgery. Obes Surg2007;17:486–492.

27. Lutchman G, Promrat K, Kleiner DE, et al. Changes in serumadipokine levels during pioglitazone treatment for nonalcoholicsteatohepatitis, relationship to histologic improvement. ClinGastroenterol Hepatol 2006;4:1048–1052.

28. Neuschwander-Tetri BA, Brunt EM, Wehmeier KR, et al. Im-proved nonalcoholic steatohepatitis after 48 weeks of treat-ment with the PPAR-y ligand rosiglitazone. Hepatology 2003;38:1008–1017.

29. Ratziu V, Charlotte F, Jacqueminet S, et al. A one-year random-ized, placebo-controlled, double-blind trial of rosiglitazone innon alcoholic steatohepatitis. Results of the FLIRT pilot trial.J Hepatol 2006;44;A201.

30. Promrat K, Lutchman G, Uwaifo GI, et al. A pilot study ofpioglitazone for nonalcoholic steatohepatitis. Hepatology 2004;39:188–196.

31. Sanyal AJ, Mofrad PS, Contos MJ, et al. A pilot study of vitaminE versus vitamin E and pioglitazone for the treatment of nonal-coholic steatohepatitis. Clin Gastroenterol Hepatol 2004;2:1107–1115.

32. Belfort R, Harrison SA, Brown K, et al. A placebo controlled trialof pioglitazone in subjects with nonalcoholic steatohepatitis.N Engl J Med 2006;355:15–25.

33. Aithal GP, Thomas JA, Kaye P, et al. A randomized, double blind,placebo controlled trial of one year of pioglitazone in non-dia-betic subjects with nonalcoholic steatohepatitis. Hepatology2007;46:295A.

34. Lutchman G, Modi A, Kleiner DE, et al. The effects of discon-tinuing pioglitazone in patients with nonalcoholic steatohepati-tis. Hepatology 2007;46:424–429.

35. Murphy CE, Rodgers PT. Effects of thiazolidinediones on boneloss and fracture. Ann Pharmacother 2007;41:2014–2018.

36. Nissen SE, Wolski K. Effect of rosiglitazone on the risk ofmyocardial infarction and death from cardiovascular causes.

N Engl J Med 2007;356:2457–2471.

37. Diamond G, Bax L, Kaul S. Uncertain effects of rosiglitazone onthe risk of myocardial infarction and cardiovascular death. AnnIntern Med 2007;147:578–581.

38. Lincoff AM, Wolski K, Nicholls SJ, et al. Pioglitazone and risk ofcardiovascular events in patients with type 2 diabetes mellitus.A meta-analysis of randomized trials. JAMA 2007;298:1180–1188.

39. Lago RM, Singh PP, Nesto RW. Congestive heart failure andcardiovascular death in patients with prediabetes and type 2diabetes given thiazolidinediones: a meta-analysis of random-ized clinical trials. Lancet 2007;370:1129–1136.

40. Lin HZ, Yang SQ, Chuckaree C, et al. Metformin reverses fattyliver disease in obese leptin-deficient mice. Nat Med 2000;6:998–1003.

41. Marchensini G, Bianchi G, Tomassetti S, et al. Metformin innon-alcoholic steatohepatitis. Lancet 2001;348:893–894.

42. Uygun A, Kadayifci A, Isik AT, et al. Metformin in the treatmentof patients with non-alcoholic steatohepatitis. Aliment Pharma-col Ther 2004;19:537–544.

43. Nair S, Diehl AM, Wiseman M, et al. Metformin in the treatmentof non-alcoholic steatohepatitis: a pilot open label trial. AlimentPharmacol Ther 2004;20:23–28.

44. Bugianesi E, Gentilcore E, Manini R, et al. A randomized con-trolled trial of metformin versus vitamin E or prescriptive diet innonalcoholic fatty liver disease. Am J Gastroenterol 2005;100:1082–1090.

45. Tiikkainen M, Hakkinen AM, Korsheninnikova E, et al. Effects ofrosiglitazone and metformin on liver fat content, hepatic insulinresistance, insulin clearance, and gene expression in adiposetissue in patients with type 2 diabetes. Diabetes 2005;53:2169–2176.

46. Mehta K, Van Thiel DH, Shah N, et al. Nonalcoholic fatty liverdisease: pathogenesis and the role of antioxidants. Nutr Rev2002;60:289–293.

47. Harrison SA, Torgerson S, Hayashi P, et al. Vitamin E and Ctreatment improves fibrosis in patients with nonalcoholic ste-atohepatitis. Am J Gastroenterol 2003;98:2485–2490.

48. Yakaryilmaz F, Guiliter S, Savas B, et al. Effects of vitamin Etreatment on peroxisome proliferators activated receptor-a ex-pression and insulin resistance in patients with non-alcoholicsteatohepatitis: results of a pilot study. Intern Med J 2007;37:229–235.

49. Federico A, Trappoliere M, Tuccillo C, et al. A new silybin-vitaminE-phospholipid complex improves insulin resistance and liverdamage in patients with non-alcoholic fatty liver disease: pre-liminary observations. Gut 2006;55:901–902.

50. Abdelmalek MF, Angulo P, Jorgenssen R, et al. Betaine, apromising new agent for patients with nonalcoholic steatohepa-titis: results of a pilot study. Am J Gastroenterol 2001;96:2711–2717.

51. Houghton WC. Betaine and nonalcoholic steatohepatitis. Am JGastroenterol 2003;98:2803.

52. Valenti L, Fracanzani AL, Dongiovanni P, et al. Iron depletion byphlebotomy improves insulin resistance in patients with nonal-coholic fatty liver disease and hyperferritinemia: evidence froma case-control study. Am J Gastroenterol 2007;102:1251–1258.

53. Lindor KD, Kowdley KV, Heathcote EJ, et al. Ursodeoxycholicacid for treatment of nonalcoholic steatohepatitis: results of arandomized trial. Hepatology 2004;39:770–778.

54. Dufour JF, Oneta CM, Gonvers JJ, et al. Randomised placebo-controlled trial of ursodeoxycholic acid with vitamin E in nonal-coholic steatohepatitis. Clin Gastroenterol Hepatol 2006;4:1537–1543.

55. Adams LA, Zeine CO, Angulo P, et al. A pilot trial of pentoxifyllinein nonalcoholic steatohepatitis. Am J Gastroenterol 2004;99:

2365–2368.

1

1

1

1

1

1

1

1

1

1

mt(

t

1698 TORRES AND HARRISON GASTROENTEROLOGY Vol. 134, No. 6

56. Satapathy SK, Sakhuja P, Malhotra V, et al. Beneficial effects ofpentoxifylline on hepatic steatosis, fibrosis and necroinflamma-tion in patients with non-alcoholic steatohepatitis. J Gastroen-terol Hepatol 2006;22:634–638.

57. Browning JD. Statins and hepatic steatosis: perspectives fromthe Dallas heart study. Hepatology 2006;44:466–471.

58. Ekstedt M, Franzen LE, Mathiesen UL, et al. Statins in non-alco-holic fatty liver disease and chronically elevated liver enzymes: ahistopathological follow-up study. J Hepatol 2007;47:135–141.

59. Gomez-Dominguez E, Gisbert JP, Moreno-Monteagudo JA, et al. A pilotstudy of atorvastatin treatment in dyslipidemidic, non-alcoholic fattyliver patients. Aliment Pharmacol Ther 2006;23:1643–1647.

60. Devaraj S, Siegel D, Jialal I. Simvastatin (40mg/day), adiponec-tin levels, and insulin sensitivity in subjects with the metabolicsyndrome. Am J Cardiol 2007;100:1397–1399.

61. Mikhailidis DP, Sibbring GC, Ballantyne CM, et al. Meta-analysisof the cholesterol-lowering effect of ezetimibe added to ongoingstatin therapy. Curr Med Res Opin 2007;23:2009–2026.

62. Deushi M, Nomura M, Kawakami A, et al. Ezetimibe improvesliver steatosis and insulin resistance in obese rat model of

metabolic syndrome. FEBS Lett 2007;581:5664–5670. v

63. Hirose A, Ono M, Saibara T, et al. Angiotensin II type 1 receptorblocker inhibits fibrosis in rat nonalcoholic steatohepatitis.Hepatology 2007;45:1375–1381.

64. Yokohama S, Yoneda M, Haneda M. Therapeutic efficacy of anangiotensin II receptor antagonist in patients with nonalcoholicsteatohepatitis. Hepatology 2004;40:1222–1225.

65. Yokohama S, Tokusashi Y, Nakamura K, et al. Inhibitory effectof angiotensin II receptor antagonist on hepatic stellate activa-tion in non-alcoholic steatohepatitis. World J Gastroenterol2006;12:322–326.

Received December 24, 2007. Accepted February 25, 2008.Address requests for reprints to: Stephen A. Harrison, MD, Depart-ent of Gastroenterology, 3851 Roger Brooke Drive, Fort Sam Hous-

on, Texas 78234. e-mail: stephen.harrison@amedd.army.mil; fax:210) 916-5611.

The opinions or assertions contained herein are the private views ofhe authors and are not to be construed as official or reflecting the

iew of the Department of the Army or the Department of Defense.