Serum levels of hyaluronic acid and tissue metalloproteinase inhibitor-1 combined with age predict the presence of nonalcoholic steatohepatitis in a pilot cohort of subjects with nonalcoholic

Serum levels of hyaluronic acid and tissuemetalloproteinase inhibitor-1 combinedwith age predict the presence of nonalcoholicsteatohepatitis in a pilot cohort of subjectswith nonalcoholic fatty liver disease

LUCA MIELE, ALESSANDRA FORGIONE, GIUSEPPE LA TORRE, VITTORIA VERO,CONSUELO CEFALO, SIMONA RACCO, VALERIO G. VELLONE, FABIO M. VECCHIO,GIOVANNI GASBARRINI, GIAN LODOVICO RAPACCINI, MANUELA G. NEUMAN,and ANTONIO GRIECO

ROME, ITALY, AND TORONTO, CANADA

From the Institutes of Internal Me

Sacred Heart, Rome, Italy; Pub

Catholic University of the Sacred

Catholic University of the Sacred

Safety and Biotechnology Labo

Toronto, Ontario, Canada.

Supported by Grant Cofin Miu

Ricercatori 2002’’ (to L.M.) and b

Linea D.1. 2004 (to A.G.).

194

Hyaluronic acid (HA) and tissue inhibitor of metalloproteinase 1 (TIMP-1) are reliablemarkers of liver fibrosis and are closely linked to the proinflammatory status. In thispilot cohort study, we attempted to identify a clinical score that would predict theseverity of nonalcoholic fatty liver disease (NAFLD) based on clinical variables andserum markers of fibrosis and inflammation. The cohort included 46 patients with his-tologically confirmed NAFLD (76.1% male; mean age, 43 6 13 years; mean bodymass index [BMI], 27.8 6 3.5). Serum transforming growth factor beta (TGF-b), HA,TIMP, and matrix metalloproteinase (MMP) levels were measured with commercialenzyme-linked immunoassay (ELISA) kits. Demographic features and clinical andlaboratory findings were subjected to univariate and multivariate binary logisticregression analysis to construct the mathematical model. Receiver operating char-acteristic curve (ROC) analysis was used to identify a threshold value for diagnosisof NASH and to assess its sensitivity and specificity. Serum levels of HA and TIMP-1were statistically different in patients with nonalcoholic steatohepatitis (NASH)(P , 0.05). Logistic regression analysis of several clinical variables indicated patientage as the only independent predictor of NASH (odds ratio [OR], 1.129, 95% confi-dence interval [CI], 1.019–1.251, P 5 0.020). The mathematical model constructedon the basis of these results included age, TIMP-1, and HA levels. A value of 148.27or more identified patients with NASH with 85.7% sensitivity, 87.1% specificity, andnegative and positive predictive values of 96.4% and 60%, respectively. This modelseems to represent a reliable noninvasive tool for excluding the presence of NASH.If validated in larger prospective cohort studies, it might be useful for determiningwhen a liver biopsy is actually warranted in patients with NAFLD. (TranslationalResearch 2009;154:194–201)

dicine, Catholic University of the

lic Health (Biostatistics Unit),

Heart, Rome, Italy; Pathology,

Heart, Rome, Italy; In Vitro Drug

ratory, University of Toronto,

r-Universita Cattolica ‘‘Giovani

y Grant Miur-Universita Cattolica

Submitted for publication January 15, 2009, revision submitted June

18, 2009; accepted for publication June 20, 2009.

Reprint requests: Antonio Grieco, MD, Department of Internal

Medicine, Agostino Gemelli University Hospital, School of Medicine,

Catholic University of Rome, 8 Largo Agostino Gemelli - 00168

Rome, Italy; e-mail: [email protected].

1931-5244/$ – see front matter

� 2009 Published by Mosby, Inc.

doi:10.1016/j.trsl.2009.06.007

AT A GLANCE COMM

Background

Hyaluronic acid (HA) and

inhibitors (TIMPs) are w

liver fibrosis. Irrespective

tory, subjects with nonalc

(NAFLD) may develop no

tis (NASH), which is a ris

liver cancer. Basic studie

lationship between NASH

genesis.

Translational Significanc

HA and TIMP1 are signifi

jects with NASH, irrespec

combined with age, provi

tify NASH. The high neg

this algorithm could be tr

tice to avoid liver biopsy i

patients.

Translational ResearchVolume 154, Number 4 Miele et al 195

Abbreviations: Ab ¼ antibody; Ag ¼ antigen; ALT ¼ alanine aminotransferase; AST ¼ aspartateaminotransferase; AUROC ¼ area under the ROC curve; BMI ¼ body mass index; CI ¼ confi-dence interval; CNR ¼ Center for Neuroscience Research; ECM ¼ extracellular matrix; ELISA¼ enzyme-linked immunosorbent assay; HA ¼ hyaluronic acid; HDL ¼ high-density lipoprotein;HOMA ¼ homeostatic model assessment; MMP ¼ matrix metalloproteinases; MS ¼ metabolicsyndrome; NAFLD ¼ nonalcoholic fatty liver disease; NAS ¼ NAFLD activity score; NASH ¼ non-alcoholic steatohepatitis; NIH¼National Institutes of Health; ROC¼ receiver operating charac-teristic curve; SD ¼ standard deviation; TGFb1 ¼ transforming growth factor beta 1; TIMPs ¼tissue metalloproteinase inhibitors; PIIINP¼ plasma procollagen type III amino-terminal peptide

ENTARY

tissue metalloproteinase

idely used markers of

of its benign natural his-

oholic fatty liver disease

nalcoholic steatohepati-

k factor for cirrhosis and

s have shown a close re-

and activation of fibro-

e

cantly increased in sub-

tive of fibrosis stage, and

de an index able to iden-

ative predictive value of

anslated in clinical prac-

n a substantial number of

Nonalcoholic fatty liver disease (NAFLD) is a common

cause of liver disease in Western countries that is known

to be associated with obesity, diabetes, and the metabolic

syndrome (MS).1,2 It begins with simple fatty changes in

the liver, which progress to cause steatohepatitis and

ultimately cirrhosis. Despite the high prevalence of

NAFLD, there are still no serum markers for distinguish-

ing nonalcoholic steatohepatitis (NASH) from fatty

liver. Liver biopsy is currently the gold standard for

the diagnosis of NASH, but false negatives can occur

as a result of sampling errors. The morphologic features

of NASH include hepatocyte ballooning, necroinflam-

matory changes, and various degrees of fibrosis,3 which

may lead to cirrhosis and liver cancer.4

Fibrosis and inflammation are closely linked processes.

Inflammation is one of the factors that stimulates remodel-

ing of the extracellular matrix (ECM) within the liver,

which is a continuous, dynamic process that involves col-

lagen breakdown and synthesis, which is mediated in large

part by matrix metalloproteinases (MMPs) and their spe-

cific tissue metalloproteinase inhibitors (TIMPs). Liver

inflammation alters the balance between these 2 processes

and promotes collagen deposition and fibrogenesis. Colla-

gen deposition is also promoted by cytokines, such as

tumor necrosis factor-alpha and transforming growth

factor- b1 (TGFb1), which stimulate the conversion of

hepatic stellate cells into myofibroblast-like cells that

can secrete a wide range of ECM proteins.5

Assays are now available for measuring serum levels of

several proteins involved in fibrogenesis, such as plasma

procollagen type III amino-terminal peptide (PIIINP),

type IV collagen 7s domain, YKL-40, hyaluronic acid

(HA), and TIMP-1, and these tools have been used by

several research groups to develop clinical scoring

systems capable of predicting the severity of fibrosis.6

Tools of this sort are important to identify cases in which

liver biopsy is actually warranted. In the pilot study de-

scribed below, we tested the hypothesis that serum levels

of cytokines involved in the liver fibrosis cascade can re-

liably identify the presence of NASH in patients with

NAFLD.

PATIENTS AND METHODS

Study design and eligibility. The cohort investigated in

this prospective study consisted of 46 patients with

histologically confirmed NAFLD who were consecutively

observed in the Outpatient Liver Unit of the Catholic Uni-

versity of the Sacred Heart Medical Center in Rome, Italy.

The study was approved by the institutional review board,

and participants provided written informed consent to all

study procedures. We enrolled men and women between

the ages of 18 and 60 years, whose medical records

showed (1) persistent increases in liver function test pa-

rameters (at least 1.5 times the upper normal limit), (2)

findings indicative of fatty liver on the screening-visit

sonogram, and (3) no history of significant alcohol intake

(.20 g/day for either sex) or viral hepatitis. Exclusion cri-

teria were (1) evidence of drug and/or alcohol addiction,

(2) history of cancer or chemotherapy within the last 5

years, (3) evidence of drug-induced liver disease within

the last 5 years, (4) genetic or autoimmune liver disease,

(5) serologic evidence of hepatitis B and/or C infection

or autoimmune hepatitis (anti-hepatitis C virus immuno-

globulin G [IgG]; hepatitis B surface [HBs]-antigen;

HBs antibody [Ab] IgG; hepatitis B core-Ab IgG, anti-

Translational Research196 Miele et al October 2009

nuclear antibodies, anti-mitochondrial antibodies and anti-

live-kidney microsome antibodies), (6) regular use of

drugs known to promote steatosis (corticosteroids, tamox-

ifen, and amiodarone),7 (7) bariatric or gastrointestinal

surgery, and (8) clinical evidence of cirrhosis. Enrollment

began in January 2005 and ended in July 2007.

Enrolled subjects were observed on 2 occasions: the

screening visit, during which eligibility for enrolment

was assessed, and the enrolment visit, during which

anthropometric data were recorded, venous blood sam-

ples were collected, and liver biopsy was performed

according the current recommendations for the diagnosis

and staging of NAFLD.8

Data collection. Participants were interviewed on the

day of enrollment by a single physician, who measured

the subject’s height and weight, recorded the clinical

history (with current and past medication use) and

reviewed the medical records. The body mass index

(BMI) was calculated (weight in kilograms/height [m2]),

and obesity was defined as a BMI $ 30 kg/m2. The pres-

ence of diabetes was defined as a fasting gluco-

se $ 126 mg/dL or specific treatment with antidiabetic

drugs, as recommended by the World Health Organiza-

tion.9 The presence of MS was defined as positivity for

at least 3 of the following ATPIII criteria10: (1) central obe-

sity (waist circumference .102 cm for men and .88 cm

for women [measured midway between the lower border

of the rib cage and the iliac crest]), (2) hyperglycemia (fast-

ing blood glucose $110 mg/dL or previously diagnosed

type 2 diabetes), (3) hypertriglyceridemia (triglycerides

$150 mg/dL or current treatment for this lipid abnormal-

ity), (4) hypertension (blood pressure $130/$85 mm Hg

or current treatment for previously diagnosed hyperten-

sion), (5) low high-density lipoprotein (HDL) cholesterol

(,40 mg/dL in men or , 50 mg/dL in women, or current

specific treatment for this lipid abnormality).

The diagnosis of liver steatosis was based on compar-

ative assessment of liver and kidney echogenicity, as

recommended by the American Gastroenterology Asso-

ciation.11 All sonographic examinations were performed

by a single gastroenterologist experienced in hepatic

sonography. Alcohol use was defined on the basis of par-

ticipants’ reports and confirmed by interviews of 1 or 2

close relatives. Ethanol abuse was defined as daily intake

of .20 g for both sexes.

Liver biopsies and laboratory tests were performed on

the same day. The latter consisted of routine liver func-

tion tests (alanine and aspartate aminotransferase, total

bilirubin, albumin, alkaline phosphatase, and gamma

glutamyl transpeptidase), a complete blood count, total-

and HDL-cholesterol levels, total triglyceride levels, and

fasting glucose and fasting insulin levels. All tests were

performed with standard methods in the centralized

laboratory of our medical center. The degree of insulin

resistance was determined by homeostatic model assess-

ment (HOMA) using the formula: (insulin*glucose)/

22.5.12

Diagnosis of NAFLD and liver biopsy. The diagnosis of

NAFLD was made when all of the following criteria

were satisfied: (1) persistently abnormal liver function

tests for more than 6 months, (2) liver biopsy showing

steatosis involving at least 10% of hepatocytes, (3) no

evidence of ethanol abuse (as defined above), and (4)

exclusion of other liver diseases and other known causes

of steatosis (drug-induced liver disease, autoimmune or

viral hepatitis, and cholestatic or metabolic/genetic liver

disease) based on the results of specific clinical, bio-

chemical, radiographic, and/or histologic studies.

Ultrasound-guided liver biopsy specimens (at least

1.6 cm long and 5 mm thick) were stained with hematoxy-

lin/eosin, Masson’s trichrome, and PAS and examined

under blinded conditions by 2 experienced pathologists

(V.G.V. and F.M.V.). Steatosis, necroinflammation, and

fibrosis were assessed according to modified Brunt’s crite-

ria. The presence of NASH was defined according to the

NAFLD activity score (NAS) . 5 as proposed by Center

for Neuroscience Research (CNR) and National Institutes

of Health (NIH).13,14

Serum assay of MMP-1, MMP-2, TIMP-1, TIMP-2, TGF-b.and HA. On the day of the liver biopsy, peripheral

venous blood samples (10 cc) were drawn from all

patients. The blood was centrifuged for 15 min at 1500

3 g at 4 �C and within 2 h of collection the serum was

stored at –80 �C until assayed. Serum levels of TGF-b,

MMP-1, MMP2, and TIMP2 were measured with com-

mercial enzyme-linked immunoassay kits (ELISA) kits

(Biotrak; Amersham Pharmacia Biotech, Buckingham-

shire, UK) according to the manufacturer’s instructions.

Serum samples were diluted 1:4 for TIMP-2 assays (sen-

sitivity: 3 ng/mL), 1:50 for free MMP-2 assays (sensitiv-

ity: 0.37 ng/mL), and 1:20 for free MMP-1 assays

(sensitivity: 1.25 ng/mL). TIMP-1 was measured with

another ELISA kit (RayBio Human; RayBiotech, Inc.,

Norcross, Ga; sensitivity: 40 pg/mL). HA levels were

measured with an ELISA kit from Corgenix (Cam-

bridge, England). The normal range was 0–75 ng/mL.

Statistical analysis. Considering the pilot nature of this

study and the lack of an a priori hypothesis on possible

results, the sample size was chosen in accordance with

the criteria described by Rothman,15 and no statistical ad-

justment was considered to lead to fewer errors of

interpretation.16 The t-test and analysis of variance

were used for parametric analysis (after the normal

distribution of the data had been confirmed). Nonpara-

metric analysis was based on the Mann-Whitney and

chi-square tests (for quantitative and categorical

Table I. Clinical features of the NAFLD cohort

n 5 46

Males, number (%) 35 (76.1)Age (years) 43 6 13BMI (kg/m2) 27/8 6 3.5Waist (cm) 89 6 24Systolic BP (mm Hg) 131 6 8Diastolic BP (mm Hg) 80 6 7Fasting glucose (mg/dL) 93 6 14Fasting insulin (mUI/mL) 17 6 12HOMA 4 6 3.23Total cholesterol (mg/dL) 203 6 33HDL cholesterol (mg/dL) 48 6 13Triglycerides (mg/dL) 152 6 77ALT (IU/L) 75 6 54AST (IU/L) 49 6 29AST/ALT ratio 0.75 6 0.51GammaGT (IU/L) 90 6 103Diabetes (yes) 9 (19.6)MS ATP III (.3) 7 (15.2)Severe steatosis (.66%) 8 (17.4)NAS $ 5 8 (17.4)Advanced fibrosis (F $ 3) 3 (6.5)

ATP, adenosine triphosphate.

Translational ResearchVolume 154, Number 4 Miele et al 197

variables, respectively). Interobserver agreement for

pathologic diagnosis of NASH was expressed with the

Cohen kappa (k) coefficient. Means 6 standard deviation

are reported for continuous variables; numbers and per-

centages are shown for categorical variables. Differences

were considered to be statistically significant when Pwas , 0.05. Correlation was analyzed with the Spear-

man rank test. The influence of clinical variables on the

presence of severe NAFLD was assessed with a backward

stepwise logistic regression (Wald method) model that

included age, BMI, fasting blood glucose, platelet count,

and the aspartate aminotransferase (AST)/alanine amino-

transferase (ALT) ratio. The fitness of the model was

evaluated with the Hosmer-Lemeshow test for signifi-

cance, and the results were expressed as odds ratios

(OR) with 95% confidence intervals (95% CI). Our aim

was to construct an index for the detection of NASH

based on the clinical variables identified as significant

(P , 0.05) in logistic regression analysis plus serum

levels of TIMP-1 and HA. The overall diagnostic accu-

racy of this approach was assessed by ROC analysis.

The area under the receiver operating characteristic curve

ROC (AUROC) was calculated for each variable found to

be significant in the correlation analysis. The threshold

index for diagnosis of NASH was identified by ROC

analysis, and its sensitivity, specificity, positive, and

negative predictive values, and likelihood ratios were

calculated. All statistical analyses were performed

using the SPSS 13.0 software package (SPSS Inc.,

Cary, NC).

RESULTS

Clinical features. The clinical and biochemical

characteristics of the 46 NAFLD patients are shown in

Table I. Liver biopsy revealed steatosis that was mild

(involving ,33% of hepatocytes) in 13 patients

(28.3%), moderate (33–66%) in 25 patients (54.3%),

and severe (.66%) in 8 patients (17.4%) patients. Fibro-

sis was absent in 2 (4.3%), perivenular in 28 (60.9%),

periportal in 13 (28.3%), and severe with septa in 3

(6.4%) patients. None of the participants had cirrhosis.

Eight (17.4%) of the 46 participants had a NAS . 5.

Agreement between the 2 pathologists (F.M.V. and

V.V.) in the diagnosis of NASH was high (k 5 0.82).

Differences between patients with fatty liver and NASH(NAS . 5). To investigate potential correlations between

clinical features and the severity of liver disease docu-

mented by the histologic examination, we divided the

cohort into 2 groups based on the presence or absence

of NASH (reflected by NASs .5 or ,5, respectively)

(Table II).

We found that the NASH subgroup was significantly

older (P , 0.05) with a higher mean BMI (P , 0.05)

and a significantly increased prevalence of diabetes

(P 5 0.017), metabolic syndrome (P 5 0.025), and

mild or moderate steatosis (P 5 0.007) with severe fibro-

sis (P 5 0.020). This group also had significantly higher

mean serum levels of HA and TIMP-1 (P , 0.05), but the

2 groups were not significantly different in terms of se-

rum levels of the other cytokines (P 5 NS) (Table II).

Factors influencing the presence of NASH andconstruction of the NASH index. The presence of NASH

(NAS . 5) was significantly correlated with age, BMI,

and the presence of diabetes, whereas advanced fibrosis

was correlated with age, HOMA, and diabetes (Table

III). To assess the independence of each NASH predictor

identified as significant in the univariate analysis or

Spearman correlation analysis, we performed a logistic

binary regression with the clinical variables correlated

with severe inflammation and/or severe fibrosis. The

regression was adjusted for age, sex, BMI, fasting glu-

cose level, HOMA, and diabetes because these variables

have been shown to influence the severity and progres-

sion of NAFLD in several studies. As shown in Table

IV, age was the only independent predictor of severity

of disease (OR, 1.129 [1.019–1.251], P 5 0.020).

We constructed an index for assessing the severity of

NAFLD that was based on the following formula:

1.129(age) 1 serum TIMP-1 1 serum HA.

The AUROC was 0.931 (95% CI, 0.848–1.014;

P , 0.001). An index of 148.27 or more identified mem-

bers of our cohort with NASH with good sensitivity

(85.7% [95% CI, 0.598–1.116]) and specificity (87.1%

Table II. Clinical features of subjects with NASH (NAS . 5) and fatty liver

NAS Score

,5 (n 5 38) $ 5 (n 5 8) P

Clinical ParametersMales—no. (%) 31 (81,6) 4 (50) n.s.Age (years) 41 6 12 54 6 9 ,0.05BMI (kg/m2) 27.2 6 3.2 30.8 6 4 ,0.05Blood glucose (mg/dL) 91 6 13 101 6 15 n.s.Insulin (mUI/ml) 15 6 11 22 6 16 n.s.HOMA 3.51 6 2.80 5.66 6 4,17 n.s.Total cholesterol (mg/dL) 201 6 35 212 6 26 n.s.HDL-cholesterol (mg/dL) 48 6 13 48 6 12 n.s.Triglycerides (mg/dL) 158 6 84 126 6 21 n.s.ALT (IU/L) 78 6 58 60 6 27 n.s.AST (IU/L) 49 6 30 49 6 24 n.s.AST/ALT ratio 0.72 6 0.53 0.88 6 0,40 n.s.Albumin (g/dL) 4,5 6 0.3 4.7 6 0.7 n.s.Platelets (3 109/L) 241 6 53 239 6 87 n.s.Metabolic syndrome – no. (%) 5 (13.9) 2 (25) 0.025Diabetes—no. (%) 5 (13.2) 4 (50) 0.017Steatosis .66%—no. (%) 34 (89.5) 4 (50%) 0.007NAS $ 5–no. (%) — — —Septal/bridging fibrosis – no. (%) 1 (2.6) 2 (25) 0.020Serum MarkersMMP-1 (ng/mL) 4.15 6 4.01 3.79 6 3.27 n.s.MMP-2 (ng/mL) 1327.62 6 286.15 1419.34 6 405.14 n.s.TIMP-1 (pg/mlL) 22.48 6 1,41 32.60 6 1.06 ,0.05TIMP-2 (ng/mL) 131.14 6 68.36 118.14 6 96.70 n.s.TGF-b (ng/mL) 43.16 6 20.32 37.04 6 26.9 n.s.HA (ng/mL) 5.69 6 29.25 125.85 6 144.90 ,0.05

Table III. Correlation analysis

NASH (NAS $ 5)

Correlation Coefficient P

Age 0.367 0.012BMI 0.331 0.025HOMA 0.290 n.s.AST/ALT 0.274 n.s.Diabetes 0.352 0.016

Table IV. Logistic regression: Variables influencing

the severity of NAFLD (NAS . 5)

OR 95%CI P

Age 1.129 1.019–1.251 0.020BMI 1.298 0.974–1.729 0.075

Adjusted for age, sex, BMI, blood glucose, diabetes, and HOMA.

Translational Research198 Miele et al October 2009

[95% CI, 0.753–0.989]). The negative predictive value

of this threshold was 96.4%; the positive predictive

value was 60% (Fig 1).

DISCUSSION

Evidence has been steadily accumulating that

NAFLD, and particularly NASH, represents fertile soil

for the development of cirrhosis and liver cancer in

patients with diabetes or obesity. These findings have

stimulated attempts to develop simple tools that will

help clinicians identify patients with steatohepatitis and

plan adequate treatment and follow-up interventions.

Thus far, no surrogate marker has been identified that

can detect NASH. Neither ultrasound nor liver function

tests can discriminate between NASH and simple fatty

liver.17 As for liver biopsy, this approach may not be ac-

cepted by patients because fatty liver is widely consid-

ered a benign condition and there is still no real

consensus on the treatment of choice for NAFLD or

NASH. Nonetheless, liver biopsy is still considered the

gold standard for diagnosing NASH, and it is required

in randomized control trials to evaluate the outcome of

specific therapies. However, histology is not suitable

for identifying the subtle changes associated with

a dynamic process like ECM remodeling, and biopsy

is too invasive for monitoring treatment responses and

disease progression in clinical settings.

A ‘‘two-hit’’ model18 has been proposed to explain

how the normal liver is modified by insulin resistance,

obesity, and diabetes. In this model, biologic processes

1,00,80,60,40,20,01 - Specificity

1,0

0,8

0,6

0,4

0,2

0,0

Sen

sitivity

ROC Curve

Fig 1. ROC analysis showing the diagnostic performance of the algo-

rithm proposed for detecting NASH (NAS . 5). See text for details.

Translational ResearchVolume 154, Number 4 Miele et al 199

in the liver are subjected to increased oxidative stress

related to hyperinsulinemia and/or alterations involving

free fatty acid traffic,19 and the effects of this stress drive

the progression of fatty liver. Excessively high levels of

free fatty acids within hepatocytes can activate the

transcription factor, nucleic factor kappa-b, which

increases the expression of proinflammatory mediators

and adhesion molecules that lead to the development

of NASH.20 In this scenario, the ECM undergoes

continuous, dynamic remodeling characterized by the

breakdown and synthesis of collagens, leading to fibro-

genesis and deposition of collagen itself.21 The progres-

sion of fibrosis is closely related to cytokine-regulated

interaction between the MMPs and their TIMPs.22

The combination of high serum TIMP-1 levels with

near-normal levels of MMP-1 observed in our NASH

patients probably represents a profile of early fibrogene-

sis. TIMP-1 messenger RNA levels and serum levels of

the protein itself increase with fibrosis,23–25 and these

changes are correlated with the inflammatory processes

that occur during viral liver disease.26,27 Therefore,

elevated serum TIMP-1 levels may reflect the active

hepatic fibrogenesis that is associated with the inflamma-

tory stage of various liver diseases.28 Given the close

association between TIMP-1 and fibroproliferation, the

increased TIMP-1 levels found in our NASH patients

are an additional confirmation of the fibrogenetic poten-

tial of NASH. Later, collagen breakdown diminishes

considerably as a result of upregulated TIMP activity,

and the fibrotic changes eventually become irrevers-

ible.29,30 The findings reported above suggest that the

hepatic fibrogenesis in our NASH population was still

in an early stage, when reversal is still possible. This

interpretation is also consistent with the young age of

these patients, and the absence of factors known to accel-

erate NASH progression. The serum levels of TIMP-2

and MMP-2 were somewhat less informative. MMP-2,

which is also known as gelatinase A, is secreted as an in-

active precursor, pro-MMP-2, which preferentially com-

plexes with TIMP-2.31 Active MMP-2 digests collagen

types IV, V, VII, X, and XI, and it can also denature in-

terstitial collagens. When fibrogenesis occurs, MMP-2

activity is inhibited by the TIMPs, and it eventually de-

creases.

In our study, we found an increased level of HA in

NASH patients. To avoid the risk of intrarate variability

for the histological diagnosis, we used a NAS . 5 for

defining the presence of NASH. HA is a component of

mucopolysaccharide, and it is mainly synthesized by

fibroblasts. Serum HA levels have been reported to

increase progressively as liver fibrosis worsens.34,32

Recent studies have revealed significant differences

between the serum HA levels of NASH patients with

severe fibrosis and those with mild to moderate fibro-

sis.33–36 In our cohort, higher levels of HA were

observed in patients with a NAS .5 (P , 0.05). More

support to our results comes from studies on viral

chronic hepatitis, in which Ramadori et al37 concluded

that HA is correlated with fibrosis and hepatic inflamma-

tory activity.

Age was the only clinical variable that was indepen-

dently associated with NASH. This result is consistent

with the literature: As shown by a recent review, age is

a clinical variable that has been linked with progression

of NAFLD in several cohort studies.38 The main finding

of our study is that the presence of NASH, which is

defined according the recent CNR-NIH classification,

can be predicted with the combination a combination

of 2 classic surrogate markers of fibrosis, TIMP 1 and

HA, plus age. It is important to recall that this finding

is based on a pilot study, and our NASH index needs

to be prospectively validated in a much larger cohort

study. In our opinion, however, the formula we used is

robust because it was constructed with a multivariate

approach using variables that have already been shown

to influence the severity and progression of NAFLD in

several studies. The Enhanced Liver Fibrosis test is

based on the same combination used to calculate our

NASH index (HA, TIMP-1, age) plus PIIINP levels,39

and it has displayed good performance in diagnosing

severe fibrosis.40 The strong correlation between the

development of fibrosis and necroinflammatory activity

in fatty liver is well documented and implies a role for

inflammatory mediators in fibrogenesis.41 Indeed, this

correlation may explain the main result of our study.

Translational Research200 Miele et al October 2009

Using ROC analysis, we identified a threshold value

of 148.27, which identified NASH patients in our cohort

with a negative predictive value of 96.4% and a positive

predictive value of 60%. Most NAFLD patients do not

have NASH, but the identification of those who do is

critically important because these individuals are likely

to develop more advanced liver disease. Serum transam-

inase levels are not reliable surrogate markers of disease

progression. In fact, advanced fibrosis may be present in

patients with normal levels.17

However, the most important contribution of serum

biomarkers in clinical settings is related to their ability

to exclude cirrhosis.42 More accurate prediction of fibro-

sis can be achieved by combining panels of noninvasive

biomarkers whose accuracy has been analyzed, as we

did. All the possible combinations and their accuracy

have been extensively reviewed recently for different

liver diseases.43,44

Caution is necessary when a diagnosis is based on

a single serum biomarker. These proteins may in fact

be nonspecific. For example, TIMP has been shown to

have a dual action: It not only inhibits MMP activity

but also prevents stellate cell apoptosis.45

Although our NASH index needs to be prospectively

validated in larger cohort studies, its negative predictive

value seems to be high, and its use may help clinicians

avoid liver biopsy in patients who are not likely to

have NASH.

REFERENCES

1. Ratziu V, Giral P, Charlotte F, et al. Liver fibrosis in overweight

patients. Gastroenterology 2000;118:1117–23.

2. de Marco R, Locatelli F, Zoppini G, Verlato G, Bonora E,

Muggeo M. Cause-specific mortality in type 2 diabetes. The

Verona Diabetes Study. Diabetes Care 1999;22:756–61.

3. Angulo P. Nonalcoholic fatty liver disease. N Engl J Med 2002;

346:1221–31.

4. Di Bisceglie AM, Lyra AC, Schwartz M, et al. Hepatitis C-related

hepatocellular carcinoma in the United States: influence of ethnic

status. Am J Gastroenterol 2003;98:2060–3.

5. Geerts A. History, heterogeneity developmental biology and

functions of quiescent hepatic stellate cells. Semin Liver Dis

2001;21:311–37.

6. Guha IN, Parkes J, Roderick PR, Harris S, Rosenberg WM.

Non-invasive markers associated with liver fibrosis in non-alco-

holic fatty liver disease. Gut 2006;55:1650–60.

7. Pessayre D, Larrey D, Biour M. Drug induced liver injury. In:

Bircher J, Benhamou JP, McIntyre N, Rizzetto M, Rodes J, eds.

Oxford textbook of clinical hepatology. Oxford, UK: Oxford

University Press, 1999.

8. Rockey DC, Caldwell SH, Goodman ZD, Nelson RC, Smith AD.

American Association for the Study of Liver Diseases. Liver

biopsy. Hepatology 2009;49:1017–44.

9. Colman P, Thomas D, Zimmet P, Welborne T, Garcia-Webb P,

Moore M. New classification and criteria for diagnosis of diabetes

mellitus. MJA 1999;170:375–8.

10. Expert Panel on Detection Evaluation and Treatment of High Blood

Cholesterol in Adults. Executive summary of the third report of the

National Cholesterol Education Program (NCEP) expert panel on

dectection, evaluation, and treatment of high blood cholesterol in

adults (Adult Treatment Panel III). JAMA 2001;285:2486–97.

11. Sanyal AJ. AGA technical review on nonalcoholic fatty liver

disease. Gastroenterology 2002;123:1705–25.

12. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF,

Turner RC. Homeostasis model assessment: insulin resistance and

beta-cell function from fasting plasma glucose and insulin concen-

trations in man. Diabetologia 1985;28:412–9.

13. Kleiner DE, Brunt EM, Van Natta M, et al. Nonalcoholic Steato-

hepatitis Clinical Research Network. Design and validation of

a histological scoring system for nonalcoholic fatty liver disease.

Hepatology 2005;41:1313–21.

14. Brunt EM, Janney CG, Di Bisceglie AM, Neuschwander-

Tetri BA, Bacon BA. Nonalcoholic steatohepatitis: a proposal

for grading and staging the histological lesions. Am J Gastroen-

terol 1999;94:2467–74.

15. Rothman KJ. No adjustments are needed for multiple comparisons.

Epidemiology 1990;1:43–6.

16. Perneger TV. What’s wrong with Bonferroni adjustments. BMJ

1998;316:1236–8.

17. Mofrad P, Contos MJ, Haque M, et al. Clinical and histologic

spectrum of nonalcoholic fatty liver disease associated with normal

ALT values. Hepatology 2003;37:1286–92.

18. Day CP, James OF. Steatohepatitis: a tale of two ‘‘hits’’? Gastro-

enterology 1998;114:842–5.

19. Leclerque IA. Antioxidant defence mechanisms: new players in

the pathogenesis of non-alcoholic steatohepatitis? Clinical Science

2004;106:235–7.

20. Diehl AM, Li ZP, Lin HZ, Yang SQ. Cytokines and the pathogen-

esis of non-alcoholic steatohepatitis. Gut 2005;54:303–6.

21. Shuppan D, Ruehl D, Somasundaram R, Hahn EG. Matrix as

a modulator of hepatic fibrogenesis. Semin Liver Disease 2001;

21:351–73.

22. Arthur MJ. Collagenases and liver fibrosis. J Hepatol 1995;22:

43–8.

23. Murawaki Y, Yamamoto H, Kawasaki H, Shima H. Serum tissue

inhibitor of metalloproteinases in patients with chronic liver dis-

ease and with hepatocellular carcinoma. Clin Chim Acta 1993;

218:47–58.

24. Murawaki Y, Ikuta Y, Kawasaki H. Clinical usefulness of serum

tissue inhibitor of metalloproteinases (TIMP)-2 assay in patients

with chronic liver disease in comparison with serum TIMP-1.

Clin Chim Acta 1999;281:109–20.

25. Muzzillo DA, Imoto M, Fukuda Y, Koyama Y, Saga S, Nagai Y,

Hayakawa T. Clinical evaluation of serum tissue inhibitor of met-

alloproteinases-1 levels in patients with liver diseases. J Gastroen-

terol Hepatol 1993;8:437–41.

26. Flisiak R, Maxwell P, Prokopowicz D, Timms PM, Panasiuk A.

Plasma tissue inhibitor of metalloproteinases-1 and transforming

growth factor beta 1—possible non-invasive biomarkers of hepatic

fibrosis in patients with chronic B and C hepatitis. Hepatogastroen-

terology 2002;49:1369–72.

27. Patel K, Gordon SC, Jacobson I, Hezode C, Oh E, Smith KM,

Pawlotsky JM, McHutchison JG. Evaluation of a panel of non-in-

vasive serum markers to differentiate mild from moderate-to-ad-

vanced liver fibrosis in chronic hepatitis C patients. J Hepatol

2004;41:935–42.

28. Ueno T, Tamaki S, Sugawara H, et al. Significance of serum tissue

inhibitor of metalloproteinases-1 in various liver diseases.

J Hepatol 1996;24:177–84.

29. Maruyama K, Feinman L, Fainsilber Z, Nakano M, Okazaki I,

Lieber CS. Mammalian collagenase increases in early alcoholic

Translational ResearchVolume 154, Number 4 Miele et al 201

liver disease and decreases with cirrhosis. Life Sci 1982;30:

1379–84.

30. Montfort I, Perez-Tamayo R, Alvizouri AM, Tello E. Collagenase

of hepatocytes and sinusoidal liver cells in the reversibility of

experimental cirrhosis of the liver. Collagenase of hepatocytes

and sinusoidal liver cells in the reversibility of experimental

cirrhosis of the liver. Virchows Arch B Cell Pathol Incl Mol Pathol

1990;59:281–9.

31. Corcoran ML, Hewitt RE, Kleiner DE Jr., Stetler-Stevenson WG.

MMP-2: expression, activation and inhibition. Enzyme Protein

1996;49:7–19.

32. Palekar NA, Naus R, Larson SP, Ward J, Harrison SA. Clinical

model for distinguishing nonalcoholic steatohepatitis from simple

steatosis in patients with nonalcoholic fatty liver disease. Liver Int

2006;26:151–6.

33. Angulo P, Keach JC, Batts KP, Lindor KD. Independent predictors

of liver fibrosis in patients with nonalcoholic steatohepatitis.

Hepatology 1999;30:1356–62.

34. Sakugawa H, Nakayoshi T, Kobashigawa K, et al. A. Clinical

usefulness of biochemical markers of liver fibrosis in patients

with nonalcoholic fatty liver disease. World J Gastroenterol

2005;11:255–9.

35. Yoneda M, Mawatari H, Fujita K, et al. Type IV collagen 7s

domain is an independent clinical marker of the severity of fibrosis

in patients with nonalcoholic steatohepatitis before the cirrhotic

stage. J Gastroenterol 2007;42:375–81.

36. Suzuki A, Angulo P, Lymp J, et al. Chronological development of

elevated aminotransferases in a non-alcoholic population.

Hepatology 2005;41:64–71.

37. Ramadori G, Zohrens G, Manns M, et al. Serum hyaluronate and

type III procollagen aminoterminal propeptide concentration in

chronic liver disease. Relationship to cirrhosis and disease activity.

Eur J Clin Invest 1991;21:323–30.

38. Miele L, Forgione A, Gasbarrini G, Grieco A. Noninvasive

assessment of fibrosis in non-alcoholic fatty liver disease

(NAFLD) and non-alcoholic steatohepatitis (NASH). Transl Res

2007;149:114–25.

39. Rosenberg WM, Voelker M, Thiel R, et al. Serum markers detect

the presence of liver fibrosis: a cohort study. Gastroenterology

2004;127:1704–13.

40. Guha IN, Parkes J, Roderick P, et al. Noninvasive markers of

fibrosis in nonalcoholic fatty liver disease: validating the European

Liver Fibrosis Panel and exploring simple markers. Hepatology

2008;47:455–60.

41. Marra F, Gastaldelli A, Svegliati Baroni G, Tell G, Tiribelli C.

Molecular basis and mechanisms of progression of non-alcoholic

steatohepatitis. Trends Mol Med 2008;14:72–81.

42. Pilette C, Rousselet MC, Bedossa P, et al. Histopathological

evaluation of liver fibrosis: quantitative image analysis vs

semi-quantitative scores. J Hepatol 1998;28:439–46.

43. Mardini H, Record C. Detection assessment and monitoring of

hepatic fibrosis: biochemistry or biopsy? Ann Clin Biochem

2005;42:441–7.

44. Lydatakis H, Hager IP, Kostadelou E, Mpousmpoulas S, Pappas S,

Diamantis I. Noninvasive markers to predict the liver fibrosis in

non-alcoholic fatty liver disease. Liver Int 2006;26:864–71.

45. Murphy FR, Issa R, Zhou X, et al. Inhibition of apoptosis of

activated hepatic stellate cells by tissue inhibitor of metalloprotei-

nase- 1 is mediated via effects on matrix metalloproteinase

inhibition: implications for reversibility of liver fibrosis. J Biol

Chem 2002;277:11069–76.