Retinol-Binding Protein 4: A New Marker ofVirus-Induced Steatosis in Patients Infected with

Hepatitis C Virus Genotype 1Salvatore Petta,1 Calogero Camma,1 Vito Di Marco,1 Nicola Alessi,1 Francesco Barbaria,1 Daniela Cabibi,2

Rosalia Caldarella,3 Stefania Ciminnisi,1 Anna Licata,1 Maria Fatima Massenti,3 Alessandra Mazzola,1

Giuseppe Tarantino,1 Giulio Marchesini,4 and Antonio Craxı1

Retinol-binding protein 4 (RBP4) is an adipocytokine associated with insulin resistance (IR). WetestedserumlevelsofRBP4toassess its linkwithsteatosis inpatientswithgenotype1chronichepatitisC (CHC) or nonalcoholic fatty liver disease (NAFLD). Nondiabetic patients with CHC (n � 143) orNAFLD (n � 37) were evaluated by liver biopsy and anthropometric and metabolic measurements,including IR by the homeostasis model assessment. Biopsies were scored by Scheuer classification forCHC,andKleinerforNAFLD.Steatosiswastestedasacontinuousvariableandgradedasabsent-mild<30%, or moderate-severe >30%. Thirty nondiabetic, nonobese blood donors served as controls.RBP4 levels were measured by a human competitive enzyme-linked immunosorbent assay kit (Adi-poGen). Mean values of RBP4 were similar in NAFLD and CHC (35.3 � 9.3 �g/L versus 36.8 �17.6; P � 0.47, respectively), and both were significantly higher than in controls (28.9 � 12.1; P �0.02 and P � 0.01, respectively). RBP4 was higher in CHC patients with steatosis than in NAFLD(42.1 � 19.7 versus 35.2 � 9.3; P � 0.04). By linear regression, RBP4 was independently linked tosteatosis only (P � 0.008) in CHC, and to elevated body mass index (P � 0.01) and low grading (P �0.04) in NAFLD. By linear regression, steatosis was independently linked to homeostasis modelassessmentscore(P�0.03)andhighRBP4(P�0.003) inCHC.Bylogisticregression,RBP4wastheonly variable independently associated with moderate-severe steatosis in CHC (odds ratio, 1.045;95% confidence interval, 1.020 to 1.070; P � 0.0004), whereas waist circumference was associatedwithmoderate-severesteatosis inNAFLD(oddsratio,1.095;95%confidence interval,1.007to1.192;P � 0.03). Conclusion: In nondiabetic, nonobese patients with genotype 1 CHC, serum RBP4 levelsmight be the expression of a virus-linked pathway to steatosis, largely unrelated to IR. (HEPATOLOGY

2008;48:28-37.)

See Editorial on Page 4 Steatosis is the hallmark of nonalcoholic fatty liverdisease (NAFLD),1 and is also a frequent histolog-ical finding in patients with chronic hepatitis C

(CHC). Its prevalence in CHC ranges between 40% and80% overall, and remains around 40% after exclusion ofall known causes of fatty liver.2 The relevance of hepaticfat content comes from its effects on the course of viraland nonviral chronic liver damage. Steatosis is considereda major determinant of fibrosis in both NAFLD3 andCHC,4,5 and negatively affects the rate of sustained viro-logical response in patients with genotype 1 (G1) CHCtreated with pegylated-interferon and ribavirin.5,6

The biological mechanism underlying steatosis is notentirely understood and is probably multifactorial. InNAFLD patients, liver fat content is specifically related toinsulin resistance (IR) and results from enhanced freefatty acid influx, together with de novo lipogenesis.1,7 ForCHC, in which the rate of steatosis is approximately two

Abbreviations: ALT, alanine aminotransferase; BMI, body mass index; CI, con-fidence interval; ELISA, enzyme-linked immunosorbent assay; �-GT, �-glutamyl-transferase; HCV, hepatitis C virus; G1, genotype 1; G3, genotype 3; CHC, chronichepatitis C; IR, insulin resistance; HOMA, homeostasis model assessment; NAFLD,nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; OR, odds ratio;RBP4, retinol-binding protein 4; RXR, retinol X receptor.

From the 1Cattedra ed Unita Operativa di Gastroenterologia, 2Cattedra diAnatomia Patologica, and 3Dipartimento di Igiene e Microbiologia–Sezione Igiene,University of Palermo, Palermo, Italy; and 4Dipartimento di Medicina e Gastro-enterologia, Alma Mater Studiorum, University of Bologna, Bologna, Italy.

Received December 21, 2007; accepted February 29, 2008.Supported by the Italian Association for the Study of the Liver (to S.P.).Address reprint requests to: Salvatore Petta, Gastroenterologia and Epatologia,

Piazza delle Cliniche 2, 90127 Palermo, Italy. E-mail: petsa@inwind.it; fax: 39091 655 2156.

Copyright © 2008 by the American Association for the Study of Liver Diseases.Published online in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/hep.22316Potential conflict of interest: Nothing to report.

28

times higher than in chronic hepatitis B8 or autoimmunehepatitis,9 two distinct “viral” and “metabolic” pathwaysto steatosis probably exist (and possibly coexist in a largeproportion of cases).

Among hepatitis C virus (HCV) genotypes, G1 HCVis generally considered devoid of any intrinsic steatogenicactivity, as opposed to genotype 3 (G3) HCV.10 HCVmay interfere with lipid metabolism through at least threedistinct, nonmutually exclusive mechanisms, identifiedusing G1-derived constructs (impaired secretion, im-paired degradation, and increased synthesis).10 More re-cently, studies have focused on G3 CHC, in whichsteatosis is more prevalent and severe, correlates with thelevel of HCV replication,11 disappears in patients withsustained virological response to antivirals,12 and re-oc-curs after viral relapse.13 Conversely, in G1 CHC, hepaticfat content is associated with elevated body mass index(BMI), central adiposity, and other clinical risk factors forNAFLD,11 and it is not responsive to antiviral therapy.12

In line with these data, several studies of G1 CHC haveidentified IR as the main determinant in the pathogenesisof steatosis4 and fibrosis,14,15 and have shown a higherprevalence of IR in HCV patients when compared tocontrol populations.14 This evidence prompted a numberof experimental studies showing that HCV per se is capa-ble of inducing IR, possibly by interfering with insulinsignaling.16,17

Recently attention has been paid to the role of retinol-binding protein 4 (RBP4) in the pathogenesis of IR. Thisprotein, a member of the lipocalin family,18 is secretedmainly by hepatocytes (80%), but also by adipose tissue(20%). It is the only specific transport protein for reti-nol19 and, by interacting with nuclear retinol X receptor(RXR), it takes part in the control of metabolic and pro-liferative cell functions,20 including steatogenesis.21

Mouse and human studies have highlighted a pathogeniclink among IR, diabetes, and high serum and adiposelevels of RBP4,21-24 identifying RBP4 as a novel adipocy-tokine. There is clinical evidence that circulating RBP4levels are related to the severity of IR and to the variousfeatures of the metabolic syndrome.25-29 Specifically,raised serum levels of RBP4 have been linked to NAFLD,assessed by ultrasonography in a cohort of subjects with30

and without diabetes.31

A recent study found a direct relation between hepaticfat content, evaluated by magnetic resonance spectros-copy, and blood levels of RBP4 in healthy subjects,32

whereas other studies linked RBP4 levels to the inflam-matory response in obese, insulin-resistant patients.33,34

To our knowledge, no studies have specifically evalu-ated the serum levels of RBP4 in patients with histologicalproven NAFLD or with G1 CHC. The issue of HCV G1

inducing steatosis is of particular relevance, since a re-markable proportion of subjects chronically infected bythis “nonsteatogenic” genotype do have steatosis, also inthe absence of predisposing factors.35

The aim of the present study was to assess the possiblerole of serum levels of RBP4 in hepatic steatosis in bothG1 CHC and NAFLD, and whether RBP4 levels wereassociated with the severity of steatosis in both conditions.

Patients and Methods

Patients. The study was carried out on a convenientsample of 180 consecutive patients (143 with G1 CHCand 37 with NAFLD), recruited at the Gastrointestinaland Liver Unit of the University Hospital in Palermo,between January 2006 and March 2007 and fulfilling allinclusion and exclusion criteria detailed below. Patientswere included if they had a histological diagnosis of G1CHC or NAFLD on a liver biopsy performed less than 6months before enrollment. G1 CHC patients were char-acterized by the presence of anti-HCV and HCV RNA,with persistently abnormal alanine aminotransferase(ALT) and liver histology of chronic hepatitis (any degreeof fibrosis, including cirrhosis), and by alcohol consump-tion �20 g/day in the last year or more. The diagnosis ofNAFLD was based on chronically elevated ALT for atleast 6 months, alcohol consumption �20 g/day in thelast year, steatosis (�5% of hepatocytes) at histologywith/without necroinflammation and/or fibrosis, andnegative anti-HCV.

Exclusion criteria were as follows: (1) advanced cirrho-sis (Child-Turcotte-Pugh B and C); (2) hepatocellularcarcinoma; (3) other causes of liver disease or mixedetiologies (excessive alcohol consumption, hepatitis B,autoimmune liver disease, Wilson’s disease, hemochro-matosis, or �1-antitrypsin deficiency); (4) human immu-nodeficiency virus infection; (5) previous treatment withantiviral therapy, immunosuppressive drugs, and/or reg-ular use of steatosis-inducing drugs, evaluated by a ques-tionnaire (for example, corticosteroid, valproic acid,tamoxifen, amiodarone); (6) previous diagnosis of type 1or type 2 diabetes mellitus, and/or fasting blood glucose�126 mg/dL; or (7) active intravenous drug addiction oruse of cannabis.

Thirty nondiabetic, nonobese (BMI �30 kg/m2), ap-parently healthy blood donors were enrolled as controls.Alcohol consumption �20 g/day for at least the last yearand/or concomitant use of any drugs were additional ex-clusion criteria. All had normal fasting glucose (�100mg/dL), cholesterol (�200 mg/dL) and triglycerides(�150 mg/dL), normal ALT (�30 UI/mL), and no evi-dence of viral infection (HCV, human immunodeficiency

HEPATOLOGY, Vol. 48, No. 1, 2008 PETTA ET AL. 29

virus, and hepatitis B surface antigen negativity) or steato-sis at ultrasonography scan.

The study was performed in accordance with the prin-ciples of the Declaration of Helsinki, and its appendices,and with local and national laws. Approval was obtainedfrom the hospital’s Internal Review Board and the EthicalCommittee, and written informed consent was obtainedfrom all patients and controls.

Clinical and Laboratory Assessment. Clinical andanthropometric data were collected at the time of liverbiopsy (Table 1). BMI was calculated on the basis ofweight (in kilograms) and height (in meters), and subjectswere classified as normal weight (BMI 18.5 to 24.9 kg/m2), overweight (BMI 25 to 29.9 kg/m2), mildly obese(BMI 30 to 34.9 kg/m2), and moderately-severely obese(BMI �35 kg/m2). Hip circumference was measured atthe widest point between hip and buttocks. Waist circum-

ference was measured at the midpoint between the lowerborder of the rib cage and the iliac crest. Visceral obesitywas diagnosed at a waist circumference �88 cm (women)and �102 cm (men).

A 12-hour overnight fasting blood sample was drawn atthe time of biopsy to determine serum levels of ALT, �-glu-tamyltransferase (�-GT), total cholesterol, high-density li-poprotein-cholesterol, triglycerides, ferritin, plasma glucoseconcentration, and platelet count. Serum insulin was deter-mined by a two-site enzyme-linked immunosorbent assay(ELISA; Mercodia Insulin ELISA; Arnika). IR was deter-mined by the homeostasis model assessment (HOMA)method, using the following equation36: IR (HOMA-IR) �fasting insulin (�U/mL) � fasting glucose (mmol/L)/22.5.HOMA-IR has been validated in comparison with the eu-glycemic/hyperinsulinemic clamp technique in both dia-betic and nondiabetic subjects.37

Table 1. Demographic, Laboratory, Metabolic, and Histologic Features of 180 Consecutive Nondiabetic, Nondrinking Patientswith Viral (HCV G1) and Nonviral (NAFLD) Chronic Liver Disease

Variable HCV G1 (n � 143) NAFLD (n � 37) P

Age (years) 50.2 � 12.4 40.6 � 11.4 �0.0001Male gender 76 (53.1) 29 (78.4) 0.005Body mass index (kg/m2) 27.4 � 4.3 28.9 � 4.8 0.07

�25 35 (24.5) 9 (24.4)25-29.9 79 (55.2) 15 (40.5)�30 29 (20.3) 13 (35.1)

Waist circumference (cm) 93.1 � 11.4 96.0 � 13.8 0.20Visceral obesity (ATPIII criteria) 55 (38.5) 14 (37.9) 0.94Hip circumference (cm) 104.7 � 10.4 101.4 � 13.2 0.11Alanine aminotransferase (IU/L; NL � 30 IU/L) 91.2 � 79.4 78.4 � 53.9 0.25Platelet count (�103/mm3) 210.9 � 52.9 217.6 � 53.6 0.49�-Glutamyl transferase (IU/L) 53.6 � 44.8 115.8 � 142.1 0.01Ferritin (ng/mL) 206.8 � 189.4 275.0 � 302.7 0.19Cholesterol (mg/dL) 177.2 � 34.9 209.5 � 43.1 �0.0001HDL cholesterol (mg/dL) 51.6 � 15.8 51.2 � 18.6 0.90Triglycerides (mg/dL) 95.3 � 37.8 136.6 � 84.8 0.006Blood glucose (mg/dL) 89.0 � 15.7 95.0 � 26.8 0.08Insulin (�U/mL) 12.2 � 6.7 16.5 � 10.5 0.02HOMA score 2.76 � 1.73 4.19 � 4.10 0.04RBP4 (�g/L) 36.8 � 17.5 35.2 � 9.3 0.47HCV-RNA (IU/mL) 765,680 � 900,429 — —Histology at biopsy

SteatosisContinuous variable 12.6 � 17.8 39.2 � 25.8 �0.0001Categorical variable

�5% 68 (47.6) 0 (0)�5% to �30% 41 (28.6) 14 (37.8)�30% 34 (23.8) 23 (62.2) �0.0001

Stage of fibrosis0-1 68 (33.6) 31 (83.8)2 48 (33.5) 4 (10.8)3-4 27 (23.7) 2 (5.4) �0.0001

Grade of activity1 34 (23.8) — —2 83 (58.0) — —3 26 (18.2) — —

Data are given as mean � SD or as number of cases (%). HCV-RNA indicates hepatitis C virus ribonucleic acid; HDL, high-density lipoprotein; HOMA, homeostasismodel assessment; NL, normal limit; and RBP4, retinol-binding protein 4.

30 PETTA ET AL. HEPATOLOGY, July 2008

Serum RBP4 levels were measured in duplicate by ahuman competitive ELISA kit (AdipoGen, Inc.). The co-efficient of variation within each assay was less than 5%.The coefficient of variation between assays was less than10%.

All patients were tested at the time of biopsy for HCV-RNA by qualitative polymerase chain reaction (CobasAmplicor HCV Test, version 2.0; limit of detection 50IU/mL). HCV-RNA–positive samples were quantified byVersant HCV RNA 3.0 bDNA (Bayer, Tarrytown, NY),expressed in IU/mL. Genotyping was performed byINNO-LiPA HCV II (Bayer).

Assessment of Histology. Slides were coded and readby a single pathologist (D.C.) unaware of the patient’sidentity and history. A minimum length of 15 mm ofbiopsy specimen or the presence of at least 10 completeportal tracts was required.38 The percentage of hepato-cytes containing fat was determined for each 10� field.An average percentage of steatosis was then determinedfor the entire specimen.

Steatosis was assessed as the percentage of hepatocytescontaining fat droplets (minimum 5%), and evaluated ascontinuous variable. Steatosis was also classified as: ab-sent-mild �30%; moderate-severe �30%.

In the NAFLD group, the Kleiner classification39 wasused to compute the NAFLD activity score (from 0 to 8,on a scale including separate scores for steatosis, lobularinflammation, and hepatocellular ballooning) and tostage fibrosis from 0 to 4. Nonalcoholic steatohepatitis(NASH) was defined and graded according to the Bruntglobal grade of activity.40

Biopsies of CHC patients were classified for grade andstage according to the Scheuer system.41

Statistics. Continuous variables were summarized asmean � standard deviation and categorical variables asfrequency and percentage. The t test and analysis of vari-ance test were used as appropriate. Multiple linear regres-sion analysis was performed to identify independentpredictors of serum RBP4 levels as continuous dependentvariables in both CHC and NAFLD patients. In thesemodels, we selected as explanatory variables age, gender,BMI, waist and hip circumference, baseline ALT, plateletcount, �-GT, ferritin, total and high-density lipoprotein-cholesterol, triglycerides, blood glucose, insulin, HOMAscore, HCV viral load, fibrosis, steatosis, activity score (inCHC patients), and lobular inflammation, hepatocellularballooning, and NAFLD activity score in NAFLD pa-tients.

Multiple logistic regression models were used to assessthe relationship of steatosis to demographic, virological(in CHC patients), metabolic and histological character-istics of both CHC and NAFLD patients. In these mod-

els, the dependent variable was steatosis coded as 0 �absent-mild (�30%); or 1 � moderate-severe (�30%).As candidate risk factors for moderate-severe steatosis weselected the same independent variables included in theRBP4 model, with RBP4 added as an additional indepen-dent variable. Multiple linear regression analysis was alsoperformed to identify independent predictors of steatosisas continuous dependent variables. In CHC patients,multiple logistic models were also used to identify inde-pendent risk factors of steatosis, coded as: 0 � absent(�5%); 1 � mild (�5% to �30%); 2 � moderate-severe(� 30%).

Variables found to be associated with the dependentvariable at univariate analyses (probability threshold, P �0.10) were included in multivariate regression models. Toavoid the effect of the colinearity, HOMA score, bloodglucose levels, and insulin levels, as well as waist circum-ference, hip circumference, and BMI, were not includedin the same multivariate model. Regression analyses wereperformed using PROC LOGISTIC, PROC REG, andsubroutines in SAS (SAS Institute, Inc., Cary, NC).42

Results

Patient Features and Histology. CHC patients werecharacterized by older age (by 10 years) and different gen-der and weight class distribution, compared with NAFLD(Table 1). Visceral obesity was identified in about 40% ofcases in both groups. Metabolic abnormalities were morecommon in NAFLD, with higher cholesterol and triglyc-erides and a higher HOMA score.

At liver biopsy (Table 1), steatosis was present in half ofthe CHC patients, but a moderate-severe grade was iden-tified only in approximately a quarter of the cases, lessfrequently than in NAFLD patients (62%). Twenty-seven subjects with NAFLD (72.9%; 95% CI, 59 to 87)were histologically diagnosed as NASH, graded as mild in12 cases (44.4%), moderate in 12 (44.4%), and severe inthree (11.2%). According to Kleiner et al.,39 43% ofNAFLD were classified as NASH, 12% as non-NASH,and 43% were indeterminate (Table 2).

The control subjects had a mean age of 40.0 � 12.5years, and 63.4% were males. Their mean BMI was24.6 � 2.8 kg/m2. All had normal ALT (24.6 � 2.8 IU),fasting glucose (76.6 � 10.4 mg/dL), cholesterol(163.8 � 22.4 mg/dL), and triglycerides (78.2 � 30.8mg/dL). The mean HOMA score was 1.98 � 0.60�U/mL (range, 0.56 to 2.55). None had arterial hyper-tension. Visceral obesity was not assessed.

Serum RBP4 Levels. Mean serum values of RBP4were similar in NAFLD and CHC (36.8 � 17.5 versus35.2 � 9.3; P � 0.4; Fig. 1), but both were significantly

HEPATOLOGY, Vol. 48, No. 1, 2008 PETTA ET AL. 31

higher than controls (28.9 � 12.1; P � 0.01 and P �0.02, respectively; Fig. 1). CHC patients with steatosishad higher RBP4 levels than in NAFLD (42.1 � 19.7versus 35.2 � 9.3; P � 0.04; Fig. 2). In the CHC cohortonly, a progressive increase of RBP4 levels was observedfrom patients without steatosis to patients with mild(�30%) and with moderate-severe steatosis (�30%; P �0.0002; Fig. 3).

In the subset of 34 CHC patients with moderate-severesteatosis, 21 (61.7%) were males, 24 (70.6%) had a BMIof �30 kg/m2, and 50% had visceral obesity. CHC sub-jects with moderate-severe steatosis without visceral obe-sity had lower HOMA score (2.61 � 1.27 versus 3.90 �

2.36; P � 0.05) but higher RBP4 levels (57.5 � 24.0versus 35.0 � 10.6 P � 0.003), compared to viscerallyobese patients.

High �-GT levels (P � 0.06) and hepatic steatosis(P � 0.008), evaluated as continuous variables, were as-sociated with higher RBP4 levels in CHC (Table 3), butonly steatosis was an independent factor by multiple lin-ear regression (P � 0.008). In NAFLD, elevated BMI,low ALT and �-GT, and high cholesterol, and low degreeof fibrosis and necroinflammation were associated withhigh RBP4, but only elevated BMI (P � 0.01) and lownecroinflammatory activity (P � 0.04) were maintainedin multiple linear regression analysis.

Finally, in the NAFLD cohort, RBP4 levels were notsignificantly different between patients with NASH andall the others (37.1 � 7.4 versus 34.5 � 5.8; P � 0.31).

Risk Factors for Steatosis. The characteristics ofCHC patients according to steatosis grade are listed inTable 4. Elevated BMI and waist circumference, as well ashigh RBP4 and HOMA score were associated with mod-erate-severe steatosis, but only higher RBP4 levels [oddsratio (OR), 1.045; 95% confidence interval (CI), 1.020 to1.070; P � 0.0004] were maintained in multivariate anal-ysis.

By multiple logistic regression analysis, carried out ac-cording to steatosis grade and the presence and the degreeof steatosis (absent versus mild versus moderate-severe)were independently linked to HOMA score (OR, 1.379;95% CI, 1.094 to 1.738; P � 0.006) and high RBP4levels (OR, 1.046; 95% CI, 1.024 to 1.068; P � 0.0001).These data were confirmed by multiple linear regressionanalyses in CHC patients, in whom the degree of steatosiswas predicted quantitatively by HOMA score (P � 0.03)and high RBP4 (P � 0.003).

Table 2. Histological Features by Kleiner et al.’s40

Classification of 37 Consecutive Nondiabetic Patients withNonalcoholic Fatty Liver Disease

Histology NAFLD (n � 37)

NAFLD activity score0-2 5 (13.6)3-4 16 (43.2)5-8 16 (43.2)

Lobular inflammation0 10 (27.0)1 20 (54.0)2 7 (19.0)3 0 (0)

Steatosis grade0 (�5%) 0 (0)1 (5%-33%) 22 (59.5)2 (�33%-66%) 6 (16.2)3 (�66%) 9 (24.3)

Hepatocellular ballooning0 8 (21.6)1 9 (24.3)2 20 (54.1)

Data are given as number of cases (%).

Fig. 1. Serum RBP4 levels in viral (G1 CHC) and nonviral (NAFLD)chronic liver disease, and in healthy control subjects.

Fig. 2. Serum RBP4 levels in viral (G1 CHC) and nonviral (NAFLD)steatosis.

32 PETTA ET AL. HEPATOLOGY, July 2008

In NAFLD (Table 5), elevated BMI, waist and hipcircumference, and HOMA score were all related to mod-erate-severe steatosis. By multivariate analysis, only a largewaist circumference (OR, 1.095; 95% CI, 1.007 to1.192; P � 0.03) was independently associated withmoderate-severe steatosis. By multiple linear regressionanalysis, the degree of steatosis was quantitatively pre-dicted by solely the waist circumference (P � 0.009).

DiscussionThis is the first study that provides evidence of an

association between elevated RBP4 levels and steatosis inG1 CHC patients, largely independent of obesity and thedegree of IR, pointing to a possible direct involvement ofthe virus in RBP4 abnormalities. In contrast, in NAFLDthe elevated RBP4 levels are more closely associated withIR and obesity, in keeping with a general metabolic dis-order.

These conclusions are based on several findings: (1)both HCV-infected and NAFLD patients have higher

serum levels of RBP4 than nonobese, nondiabetic healthyblood donors without metabolic disturbances; (2) HCV-infected patients with steatosis are at low metabolic risk,but their RBP4 levels are higher compared to NAFLDsubjects; (3) RBP4 levels increase progressively in directrelation to the histological degree of steatosis in G1 CHC,but not in NAFLD; and (4) in patients with CHC andmoderate-severe steatosis, subjects without metabolic dis-turbances have higher RBP4 levels than their counterpartswith metabolic risk factors, and an independent link existsbetween the degree of steatosis and RBP4 levels.

In patients with G1 HCV infection the mechanism(s)underlying the occurrence of liver steatosis are not entirelyunderstood. A high hepatic fat content is associated withIR4 due to metabolic and viral factors, and with its clinicalexpressions, such as elevated BMI and central adiposity.11

However, G1 HCV seems also able to induce steatosis bydirect interference with lipid metabolism,10 and thepresent study is fairly in keeping with this hypothesis. InG1 CHC patients (all nondrinkers, naive to antiviral ther-

Fig. 3. Serum RBP4 levels according to the degree of steatosis in G1 CHC and NAFLD.

Table 3. Univariate and Multivariate Analysis of Host Factors Associated with Elevated RBP4 Levels in Patients withGenotype 1 Chronic Hepatitis C and with Nonalcoholic Fatty Liver Disease

Variable

Univariate Analysis Multivariate Analysis

� SE P Value � SE P Value

Genotype 1 chronic hepatitis C�-Glutamyl transferase (IU/L) 0.060 0.032 0.06 0.042 0.032 0.19Steatosis 0.214 0.80 0.008 0.215 0.079 0.008

Nonalcoholic fatty liver diseaseBody mass index (kg/m2) 0.584 0.337 0.09 0.750 0.297 0.01Alanine aminotransferase (IU/L) �0.066 0.029 0.03 �0.055 0.032 0.10�-Glutamyl transferase (IU/L) �0.026 0.010 0.02 �0.014 0.014 0.32Cholesterol (mg/dL) �0.077 0.036 0.03 0.000 0.094 0.98

Histology at biopsyStage of fibrosis �3.127 1.830 0.08 �1.468 2.079 0.48Lobular inflammation �5.733 2.368 0.02 �5.297 2.472 0.04

SE indicates standard error.

HEPATOLOGY, Vol. 48, No. 1, 2008 PETTA ET AL. 33

apy and with a low rate of obesity, and independently ofthe classic measures of visceral adiposity), 52% had histo-logical steatosis, which was in the moderate-severe rangein 24%.

Although this study was not designed to clarify thepathogenesis of elevated RBP4 levels in G1 CHC, a fewhypotheses may be put forward on the basis of the presentand literature data. G1 HCV might induce the hepaticoverexpression of RBP4, and this protein may in turnhave an important role in the control of hepatic fat infil-tration. Tsutsumi et al.43 showed that HCV core proteinenhances the transcriptional activity regulated by RXR�homodimers and heterodimers, also inducing steatosis.Retinol is the precursor for the synthesis of ligands ofthese nuclear hormone receptors; RBP4 is the only trans-port protein of retinol in the blood,19 and its expression inanimal models is associated with elevation of messengerRNA of several retinoic acid-responsive genes. HCV, byinteracting with the same metabolic pathway, might in-duce the hepatic expression of RBP4, favoring the RXR-retinol interaction and therefore determining hepatic fatinfiltration.

However, other mechanism(s) linking RBP4 to steato-sis may be operative. Ost et al.44 found that RBP4 inter-feres with insulin signaling in adipocytes. Accordingly,RBP4 might also interact with a specific, so far unidenti-

fied, hepatocyte membrane receptor, acting as a modula-tor of steatogenesis. Finally, the evidence that both HCV-infected and NAFLD patients have RBP4 levels higherthan healthy controls could stem from totally indepen-dent mechanism(s), for example, adipose overexpressionof RBP4 in NAFLD subjects with metabolic distur-bances, and virus-induced hepatic overproduction ofRBP4 in G1 CHC.

We did not find a direct relation between HCV-RNAlevels and serum RBP4, which, given the fluctuating levelsof viremia in HCV infection,45 does not preclude a directassociation of RBP4 with infection with G1 HCV. Sim-ilarly, in keeping with literature data,4 we did not observea direct relation between the degree of steatosis and serumHCV-RNA levels.10

The elevated RBP4 levels of the NAFLD cohort aremore easily explained. Our study confirms the link be-tween steatosis and IR and identifies a pivotal role ofvisceral fat. Recent studies in experimental animals and inhumans22-24 have suggested that RBP4, when overex-pressed by visceral fat, is implicated in the pathogenesis ofIR and is directly associated with many features of themetabolic syndrome,25-29 including NAFLD.30 Accord-ing to these data, we found a direct relationship betweenRBP4 levels and BMI in NAFLD, but could not confirmthe recently proven association of RBP4 levels with AST

Table 4. Univariate and Multivariate Analysis of Host and Viral Factors Associated with Moderate-Severe Steatosis (>30%Hepatocytes Involved) in 143 Consecutive Nondrinking, Nondiabetic Patients with Chronic Hepatitis C Genotype 1

Variable

Absent-Mild Steatosis(<30% Hepatocytes;

n � 109)

Moderate-SevereSteatosis

(>30% Hepatocytes;n � 34)

Univariate Analysis(P Value)

Multivariate Analysis

OR (95% CI) P Value

Age (years) 50.0 � 12.1 51.1 � 13.5 0.64 —Male gender 55 (50.4) 21 (61.7) 0.25 —Body mass index (kg/m2) 27.1 � 4.4 28.5 � 3.8 0.09 —Waist circumference (cm) 92.1 � 11.5 96.5 � 10.3 0.05 1.034 (0.992-1.079) 0.11Hip circumference (cm) 104.3 � 10.4 106.3 � 10.1 0.37 —Alanine aminotransferase (IU/L; NL � 30 IU/L) 87.6 � 71.4 103.3 � 101.8 0.32 —Platelet count (�103/mm3) 214.6 � 53.7 198.9 � 49.5 0.13 —�-Glutamyl transferase (IU/L) 52.2 � 47.0 58.2 � 36.8 0.49 —Ferritin (ng/mL) 193.2 � 169.9 254.5 � 243.6 0.12 —Cholesterol (mg/dL) 175.0 � 32.9 184.0 � 40.4 0.18 —HDL cholesterol (mg/dL) 51.8 � 16.6 50.8 � 13.1 0.74 —Triglycerides (mg/dL) 93.1 � 34.3 102.9 � 47.3 0.19 —HOMA score 2.61 � 1.63 3.25 � 1.98 0.07 1.148 (0.886-1.488) 0.29RBP4 (�g/L) 33.8 � 15.0 46.3 � 21.5 0.0009 1.045 (1.020-1.070) 0.0004HCV-RNA (IU/mL) 752,361 � 924,554 803,687 � 843,391 0.83 —Histology at biopsy

Stage of fibrosis1-2 88 (80.7) 28 (82.3)3-4 21 (19.3) 6 (17.7) 0.84 —

Grade of activityMild 27 (24.8) 7 (14.9)Moderate-severe 82 (75.2) 27 (85.1) 0.97 —

Data are given as mean � standard deviation or as number of cases (%). HCV-RNA indicates hepatitis C virus ribonucleic acid; HDL, high-density lipoprotein; HOMA,homeostasis model assessment; NL, normal limit; and RBP4, retinol-binding protein 4.

34 PETTA ET AL. HEPATOLOGY, July 2008

and �-GT levels.31 Differences in age, anthropometricindices, and ALT and �-GT levels among studies mayexplain this discrepancy.

In our setting, RBP4 was inversely associated with he-patic necroinflammatory activity. Conflicting data areavailable on this issue. Clinical studies have reported adirect association between adipose or serum expression ofRBP4 and adipose or systemic inflammation,33,34 whereasin an experimental mouse model RXR�-deficiency wasreported to promote inflammatory-related liver injury.46

The effects of RBP4 on liver cells might differ from thoseon adipose tissue33 and on systemic inflammation,34 thusthe results observed in our small NAFLD group need tobe confirmed in a larger study. The main limitation of thisstudy lies in its cross-sectional nature and its inability todissect the temporal relation between RBP4 and steatosis.A further methodological issue is the potentially limitedexternal validity of the results to different populations andsettings, particularly for NAFLD. Our study included acohort of Italian subjects enrolled at a tertiary care center,who may be different from the majority of prevalent casesof NAFLD in the general population.

Lack of data on the liver expression of RBP4, and onother potential confounders, such as proinflammatory cy-tokines and adipocytokines, could also affect the interpre-tation of the results. In addition, we cannot exclude thepossibility that hidden abuse of alcohol may be responsi-

ble for steatosis in a few subjects, although this bias, ifpresent, might be equally shared by the two groups, aswell as by controls, given the similar methodology of as-certainment.

In conclusion, this study has shown a remarkable asso-ciation between the degree of hepatic steatosis and RBP4levels, restricted to G1 CHC patients and unrelated toabnormal metabolic features. This would suggest a viralpathway to steatosis, involving direct overexpression ofRBP4 by the hepatocytes. Experimental studies areneeded to determine the virus-induced mechanism(s) re-sponsible for the association between steatosis and RBP4in G1 HCV.

Acknowledgment: We thank Warren Blumberg forhis help in editing this article. S. Petta designed the study,contributed to data acquisition, was responsible for writ-ing the manuscript, and participated in statistical analysis.C. Camma participated in the writing of the manuscriptand was responsible for statistical analysis. V. Di Marco,A. Craxı (Director of the Gastrointestinal and LiverUnit), and G. Marchesini were responsible for writing themanuscript and have seen and approved the final version.N. Alessi, F. Barbaria, D. Cabibi, R. Caldarella, S. Cim-innisi, A. Licata, F. Massenti, A. Mazzola, and G.Tarantino participated in patient management and datacollection. All authors have seen and approved the finalversion of the manuscript.

Table 5. Univariate and Multivariate Analysis of Host Factors Associated with Moderate-Severe Steatosis in 37 ConsecutivePatients with Nonalcoholic Fatty Liver Disease

Variable

Mild Steatosis(<30% Hepatocytes;

n � 14)

Moderate-SevereSteatosis (>30%Hepatocytes; n �

23)Univariate Analysis

(P Value)

Multivariate Analysis

OR (95% CI) P Value

Age (years) 43.4 � 13.8 39.0 � 9.6 0.25 —Male gender 10 (71.4) 19 (82.6) 0.42 —Body mass index (kg/m2) 26.1 � 4.4 30.5 � 4.3 0.01 —Waist circumference (cm) 87.0 � 15.0 101.2 � 10.0 0.01 1.095 (1.007-1.192) 0.03Hip circumference (cm) 92.8 � 17.6 105.5 � 8.1 0.02 —Alanine aminotransferase (IU/L; NL � 30 IU/L) 62.5 � 37.9 88.1 � 60.4 0.18 —Platelet count (�103/mm3) 220.3 � 45.7 215.9 � 58.8 0.80 —�-Glutamyl transferase (IU/L) 124.6 � 139.0 110.4 � 146.7 0.76 —Ferritin (ng/mL) 191.7 � 95.7 325.6 � 370.7 0.25 —Cholesterol (mg/dL) 208.0 � 30.0 210.5 � 50.1 0.86 —HDL cholesterol (mg/dL) 55.6 � 17.2 48.6 � 11.9 0.17 —Triglycerides (mg/dL) 118.0 � 46.4 147.9 � 100.7 0.30 —HOMA score 2.60 � 1.64 5.16 � 4.83 0.06 1.527 (0.826-2.825) 0.17RBP4 (�g/L) 33.5 � 7.7 36.3 � 10.1 0.38 —Histology at biopsy

Stage of fibrosis0-2 13 (92.8) 22 (95.6)3-4 1 (7.2) 1 (4.4) 0.81

Lobular inflammation0-1 12 (85.7) 18 (78.2)2-3 2 (14.3) 5 (21.8) 0.28 —

Data are given as mean � standard deviation or as number of cases (%). HDL indicates high-density lipoprotein; HOMA, homeostasis model assessment; NL, normallimit; and RBP4, retinol-binding protein 4.

HEPATOLOGY, Vol. 48, No. 1, 2008 PETTA ET AL. 35

References1. Bugianesi E, McCullough AJ, Marchesini G. Insulin resistance: a met-

abolic pathway to chronic liver disease. HEPATOLOGY 2005;42:987-1000.

2. Asselah T, Rubbia-Brandt L, Marcellin P, Negro F. Steatosis in chronichepatitis C: why does it really matter? Gut 2006;55:123-130.

3. Bugianesi E, Marchesini G, Gentilcore E, Cua IH, Vanni E, Rizzetto M, etal. Fibrosis in genotype 3 chronic hepatitis C and nonalcoholic fatty liverdisease: Role of insulin resistance and hepatic steatosis. HEPATOLOGY 2006;44:1648-1655.

4. Camma C, Bruno S, Di Marco V, Di Bona D, Rumi M, Vinci M, et al.Insulin resistance is associated with steatosis in nondiabetic patients withgenotype 1 chronic hepatitis C. HEPATOLOGY 2006;43:64-71.

5. Leandro G, Mangia A, Hui J, Fabris P, Rubbia-Brandt L, Colloredo G, etal. Relationship between steatosis, inflammation, and fibrosis in chronichepatitis C: a meta-analysis of individual patient data. Gastroenterology2006;130:1636-1642.

6. Poynard T, Ratziu V, McHutchison J, Manns M, Goodman Z, Zeuzem S,et al. Effect of treatment with peginterferon or interferon alfa-2b andribavirin on steatosis in patients infected with hepatitis C. HEPATOLOGY

2003;38:75-85.7. Choudhury J, Sanyal AJ. Insulin resistance and the pathogenesis of nonal-

coholic fatty liver disease. Clin Liver Dis 2004;8:575-594.8. Gordon A, McLean CA, Pedersen JS, Bailey MJ, Roberts SK. Hepatic

steatosis in chronic hepatitis B and C: predictors, distribution and effect onfibrosis. J Hepatol 2005;43:38-44.

9. Yatsuji S, Hashimoto E, Kaneda H, Taniai M, Tokushige K, Shiratori K.Diagnosing autoimmune hepatitis in nonalcoholic fatty liver disease: is theInternational Autoimmune Hepatitis Group scoring system useful? J Gas-troenterol 2005;40:1130-1138.

10. Clement S, Negro F. Hepatitis C virus: the viral way to fatty liver. J Hepa-tol 2007;46:985-987.

11. Adinolfi LE, Gambardella M, Andreana A, Tripodi MF, Utili R, RuggieroG. Steatosis accelerates the progression of liver damage of chronic hepatitisC patients and correlates with specific HCV genotype and visceral obesity.HEPATOLOGY 2001;33:1358-1364.

12. Castera L, Hezode C, Roudot-Thoraval F, Lonjon I, Zafrani ES, PawlotskyJM, et al. Effect of antiviral treatment on evolution of liver steatosis inpatients with chronic hepatitis C: indirect evidence of a role of hepatitis Cvirus genotype 3 in steatosis. Gut 2004;53:420-424.

13. Rubbia-Brandt L, Giostra E, Mentha G, Quadri R, Negro F. Expression ofliver steatosis in hepatitis C virus infection and pattern of response toalpha-interferon. J Hepatol 2001;35:307.

14. Hui JM, Sud A, Farrell GC, Bandara P, Byth K, Kench JG, et al. Insulinresistance is associated with chronic hepatitis C virus infection and fibrosisprogression. Gastroenterology 2003;125:1695-1704.

15. Svegliati-Baroni G, Bugianesi E, Bouserhal T, Marini F, Ridolfi F, TarsettiF, et al. Post-load insulin resistance is an independent predictor of hepaticfibrosis in virus C chronic hepatitis and in non-alcoholic fatty liver disease.Gut 2007;56:1296-1301.

16. Shintani Y, Fujie H, Miyoshi H, Tsutsumi T, Tsukamoto K, Kimura S, etal. Hepatitis C virus infection and diabetes: direct involvement of the virusin the development of insulin resistance. Gastroenterology 2004;126:840-848.

17. Pazienza V, Clement S, Pugnale P, Conzelman S, Foti M, Mangia A, et al.The hepatitis C virus core protein of genotypes 3a and 1b downregulatesinsulin receptor substrate 1 through genotype-specific mechanisms. HEPA-TOLOGY 2007;45:1164-1171.

18. van Dam RM, Hu FB. Lipocalins and insulin resistance: etiological role ofretinol-binding protein 4 and lipocalin-2? Clin Chem 2007;53:5-7.

19. Blaner WS. Retinol-binding protein: the serum transport protein for vita-min A. Endocr Rev 1989;10:308-316.

20. Desvergne B. RXR: from partnership to leadership in metabolic regula-tions. Vitam Horm 2007;75:1-32.

21. Larter CZ, Farrell GC. Insulin resistance, adiponectin, cytokines inNASH: which is the best target to treat? J Hepatol 2006;44:253-261.

22. Yang Q, Graham TE, Mody N, Preitner F, Peroni OD, Zabolotny JM, etal. Serum retinol binding protein 4 contributes to insulin resistance inobesity and type 2 diabetes. Nature 2005;436:356-362.

23. Graham TE, Yang Q, Bluher M, Hammarstedt A, Ciaraldi TP, Henry RR,et al. Retinol-binding protein 4 and insulin resistance in lean, obese, anddiabetic subjects. N Engl J Med 2006;354:2552-2263.

24. Kloting N, Graham TE, Berndt J, Kralisch S, Kovacs P, Wason CJ, et al.Serum retinol-binding protein is more highly expressed in visceral than insubcutaneous adipose tissue and is a marker of intra-abdominal fat mass.Cell Metab 2007;6:79-87.

25. von Eynatten M, Lepper PM, Liu D, Lang K, Baumann M, NawrothPP, et al. Retinol-binding protein 4 is associated with components ofthe metabolic syndrome, but not with insulin resistance, in men withtype 2 diabetes or coronary artery disease. Diabetologia 2007;50:1930-1937.

26. Takebayashi K, Suetsugu M, Wakabayashi S, Aso Y, Inukai T. Retinolbinding protein-4 levels and clinical features of type 2 diabetes patients.J Clin Endocrinol Metab 2007;92:2712-2719.

27. Broch M, Vendrell J, Ricart W, Richart C, Fernandez-Real JM. Circulat-ing retinol-binding protein-4, insulin sensitivity, insulin secretion, andinsulin disposition index in obese and nonobese subjects. Diabetes Care2007;30:1802-1806.

28. Lee DC, Lee JW, Im JA. Association of serum retinol binding protein 4and insulin resistance in apparently healthy adolescents. Metabolism 2007;56:327-331.

29. Wu H, Jia W, Bao Y, Lu J, Zhu J, Wang R, et al. Serum retinol bindingprotein 4 and nonalcoholic fatty liver disease in patients with type 2 dia-betes mellitus. Diabetes Res Clin Pract 2007;79:185-190.

30. Janke J, Engeli S, Boschmann M, Adams F, Bohnke J, Luft FC, et al.Retinol-binding protein 4 in human obesity. Diabetes 2006;55:2805-2810.

31. Seo JA, Kim NH, Park SY, Kim HY, Ryu OH, Lee KW, et al. Serumretinol-binding protein 4 levels are elevated in non-alcoholic fatty liverdisease. Clin Endocrinol (Oxf) 2007;68:555-560.

32. Stefan N, Hennige AM, Staiger H, Machann J, Schick F, Schleicher E, etal. High circulating retinol-binding protein 4 is associated with elevatedliver fat but not with total, subcutaneous, visceral, or intramyocellular fatin humans. Diabetes Care 2007;30:1173-1178.

33. Balagopal P, Graham TE, Kahn BB, Altomare A, Funanage V, George D.Reduction of elevated serum retinol binding protein in obese children bylifestyle intervention: association with subclinical inflammation. J ClinEndocrinol Metab 2007;92:1971-1974.

34. Yao-Borengasser A, Varma V, Bodles AM, Rasouli N, Phanavanh B, LeeMJ, et al. Retinol binding protein 4 expression in humans: relationship toinsulin resistance, inflammation, and response to pioglitazone. J Clin En-docrinol Metab 2007;92:2590-2597.

35. Petta S, Camma C, Di Marco V, Alessi N, Cabibi D, Caldarella R. Insulinresistance and diabetes increase fibrosis in the liver of patients with geno-type 1-HCV infection. Am J Gastroenterol 2008. Available at: http://www.amjgastro.com.

36. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, TurnerRC. Homeostasis model assessment: insulin resistance and beta-cell func-tion from fasting plasma glucose and insulin concentrations in man. Dia-betologia 1985;28:412-419.

37. Ikeda Y, Suehiro T, Nakamura T, Kumon Y, Hashimoto K. Clinical sig-nificance of the insulin resistance index as assessed by homeostasis modelassessment. Endocr J 2001;48:81-86.

38. Colloredo G, Guido M, Sonzogni A, Leandro G. Impact of liver biopsy sizeon histological evaluation of chronic viral hepatitis: the smaller the sample,the milder the disease. J Hepatol 2003;39:239-244.

39. Kleiner DE, Brunt EM, Van Natta M, Behling C, Contos MJ, Cum-mings OW, et al. Design and validation of a histological scoring systemfor nonalcoholic fatty liver disease. HEPATOLOGY 2005;411:313-321.

40. Brunt EM, Janney CG, Di Bisceglie AM, Neuschwander-TetriBA, Bacon BR. Nonalcoholic steatohepatitis: a proposal for gradingand staging the histological lesions. Am J Gastroenterol 1999;9:2467-2474.

36 PETTA ET AL. HEPATOLOGY, July 2008

41. Scheuer PJ. Classification of chronic viral hepatitis: a need for reassess-ment. J Hepatol 1991;13:372-374.

42. SAS Technical Report, SAS/STAT software: changes and enhancement,Release 6.07. Cary, NC: SAS Institute, Inc.; 1992.

43. Tsutsumi T, Suzuki T, Shimoike T, Suzuki R, Moriya K, Shintani Y, etal. Interaction of hepatitis C virus core protein with retinoid X receptoralpha modulates its transcriptional activity. HEPATOLOGY

2002;35:937-946.44. Ost A, Danielsson A, Liden M, Eriksson U, Nystrom FH, Strålfors P.

Retinol-binding protein-4 attenuates insulin-induced phosphorylation

of IRS1 and ERK1/2 in primary human adipocytes. FASEB J 2007;21:3696-3704.

45. Pontisso P, Bellati G, Brunetto M, Chemello L, Colloredo G, Di Ste-fano R, et al. Hepatitis C virus RNA profiles in chronically infectedindividuals: do they relate to disease activity? HEPATOLOGY 1999;29:585-589.

46. Gyamfi MA, He L, French SW, Damjanov I, Wan YJ. Hepatocyte retinoidX receptor alpha-dependent regulation of lipid homeostasis and inflamma-tory cytokine expression contributes to alcohol-induced liver injury.J Pharmacol Exp Ther 2007;324:443-453.

HEPATOLOGY, Vol. 48, No. 1, 2008 PETTA ET AL. 37