ORIGINAL ARTICLE – TRANSLATIONAL RESEARCH AND BIOMARKERS

Recurrence and Poor Prognosis Following Resection of SmallHepatitis B-Related Hepatocellular Carcinoma Lesions AreAssociated with Aberrant Tumor Expression Profiles of Glypican3 and Osteopontin

Ming-Chin Yu, MD1,2, Yun-Shien Lee, PhD3,4, Sey-En Lin, MD, PhD5, Hsiang-Yao Wu, MS2,

Tse-Ching Chen, MD, PhD6, Wei-Chen Lee, MD1, Miin-Fu Chen, MD1, and Chi-Neu Tsai, PhD2

1Department of Surgery, Chang Gung Memorial Hospital, Taoyuan, Taiwan; 2Graduate Institute of Clinical Medical

Sciences, Chang Gung University, Taoyuan, Taiwan; 3Department of Biotechnology, Ming Chuan University, Taoyuan,

Taiwan; 4Genomic Medicine Research Core Laboratory, Chang Gung Memorial Hospital, Taoyuan, Taiwan; 5Department

of Pathology, Taipei Medical University Hospital, Taipei, Taiwan; 6Department of Pathology, Chang Gung Memorial

Hospital, Chang Gung University, Taoyuan, Taiwan

ABSTRACT

Background. Early detection and following appropriate

treatments of hepatocellular carcinoma (HCC) is still the

gold standard for favored outcome of HCC patients; nev-

ertheless, a small portion of hepatitis B virus (HBV)-

related small HCC (\5 cm) patients got poor prognosis.

Furthermore, the study for small HBV–HCC was limited.

Therefore, the aim of this study was to explore the potential

genetic signature for HBV-related small HCC as novel

prognostic factors.

Methods. We examined expression profiles of HBV-rela-

ted small HCC using an Affymetrix U133A GeneChip,

evaluated differential gene expression by quantitative real-

time polymerase chain reaction (qRT-PCR), and finally

validated these expression patterns by immunohistochem-

istry (IHC).

Results:. A total of 57 genes were differentially expressed

between tumor and normal parts (n = 20 pairs) using

Affymetrix U133A chip, and 16 genes were further eval-

uated by qRT-PCR. The result was compatible with the

finding of oligonucleotide microarray (Pearson’s correla-

tion, r = 0.87). Furthermore, the expression pattern in

HCC tissue by IHC in another group of small HBV–HCC

(n = 100) showed overexpression of either osteopontin

(OPN) or glypican 3 (GPC3) is an independent prognostic

factor for disease-free survival (DFS) in HBV-positive

small HCC (P \ 0.01 and 0.03, respectively). Long-term

DFS and overall survival (OS) for small HBV–HCC

patients with high risk (both elevated GPC3?/OPN?) were

DFS 0%, OS 0%, respectively; on the other hand, DFS and

OS in patients with moderate (only 1 gene elevated) or low

(OPN-/GPC3-) risk were 35.0 and 46.5%, respectively.

Conclusions. Elevation of both OPN and GPC3 may act as

an adverse indicator for HBV-related small HCC patients

after curative resection.

Primary liver cancer, hepatocellular carcinoma (HCC), is

one of the most common cancers worldwide. Globally, it is

the third highest cause of cancer mortality, having a yearly

fatality ratio of approximately 1.1 In Taiwan, HCC was

ranked in the annual report of the Department of Health

(DOH) as the first cause of cancer mortality in men and the

second in women.2 The main etiology of HCC is chronic

infection with hepatitis B or C virus (HBV or HCV). Tai-

wan has one of the highest prevalence rates in the world;

about 15–20% of the population are chronic carriers of

HBV, and the relative risks of hepatocarcinogenesis have

Electronic supplementary material The online version of thisarticle (doi:10.1245/s10434-011-1946-2) contains supplementarymaterial, which is available to authorized users.

� Society of Surgical Oncology 2011

First Received: 1 February 2011

M.-F. Chen, MD

e-mail: chenmf@adm.cgmh.org.tw

C.-N. Tsai, PhD

e-mail: pink7@mail.cgu.edu.tw

Ann Surg Oncol

DOI 10.1245/s10434-011-1946-2

been reported to be 9.6 for HBV surface antigen (HBsAg)-

seropositive patients and 60.2 for those positive for HBsAg

and HB envelope antigen (HBeAg).3 Recently, a large-scale

cohort study—the risk evaluation of viral load elevation and

associated liver disease/cancer-hepatitis B virus (REVEAL-

HBV) study—demonstrated that having a serum level

greater than 104 copies/ml of HBV–DNA upon entry into

the study was an important risk factor for HBV hepatocar-

cinogenesis, independent of HBe Ag seropositivity.4 HCC

commonly occurs between the ages of 50 and 60 years in

Taiwanese, at a time of life when these patients are a main

economic support for their families.5 Therefore, HBV-

related HCC is a major public health threat that has a sig-

nificant socioeconomic impact in our country.

Prognosis for these patients remains poor despite many

advances in HCC treatment regimens; therefore, predicting

clinical outcomes for patients is a challenge for clinicians

and researchers. Retrospective clinical analyses have been

performed in eastern and western countries regarding

patients who presented with pathological factors such as

multiplicity, satellite nodules, vascular invasion, and lack

of tumor encapsulation.6–9 Although partial hepatectomies

were potentially curative treatments, the recurrence rate of

all resectable HCC was still more than 75% in long-term

follow-up.8,10 For recurrence of small HCC, most reports

of recurrence were 50–70%, and long-term outcome is

acceptable.11,12

Despite the odds, early detection, especially of small

(\5 cm) HCC lesions (small HCC) coupled with appro-

priate treatment is very important for good prognosis.

However, this imposes diagnostic challenges for clinicians

because of the low sensitivity of currently available

screening biomarkers, such as a-fetoprotein (AFP), des-c-

carboxyprothrombin (DCP), and lens culinaris agglutinin

A-reactive fraction of AFP (AFP-L3).13 There are no ‘‘gold

standard’’ serum markers for small HCC at the present

time, and therefore the search for potential biomarkers for

early diagnosis or surveillance is an ongoing task.14

To investigate the roles of genetics or signaling pathway

in HCC carcinogenesis, methods such as comparative

genomic hybridization, transcriptome microarray, and

proteomics have been used. In recent years, oligonucleo-

tide microarray and systemic biology have been powerful

tools for finding potential useful biomarkers for clinical

application in signaling pathways.15–17 Most previous

studies analyzed gene signatures involved in vascular

invasion, recurrence, methylation, single or multiple nod-

ules, and differences between HBV- and HCV-related

HCC; however, only a few reports have addressed differ-

ential gene expression in small or early HCC.18–30 In this

study, we analyzed molecular signatures of HBV-related

small HCC by using Affymetrix human U133A Gene-

Chips. The molecular genetic signature was further

validated by quantitative real-time polymerase chain

reaction (qRT-PCR) and IHC analysis. In addition, we also

evaluated 2 independent significant prognostic factors in

survival analysis from the microarray genetic signatures in

early HBV–HCC patients.

PATIENTS AND METHODS

Patients

Microarray Studies From 2008 to 2009, 20 patients with

HBV-related HCC who underwent partial hepatectomy

were enrolled as the first cohort. Informed consents were

obtained under the approval of the Institutional Review

Boards (IRB 97-1371C) of Chang Gung Memorial Hospital

(CGMH). Inclusion criteria were defined as presence on an

image study (including sonography, computed tomog-

raphy [CT], and angiography) of a single hepatic lesion of

less than 5 cm. Patients with positivity in a resection

margin and HCV carriers were excluded, as were patients

who had unresectable disease, synchronous cancers, distant

metastasis, or prior history of another malignancy. Patients

with small HCC and/or decompensated liver cirrhosis were

suggested for liver transplantation and excluded in this study.

Noncancerous control samples were selected 3 cm away

from HCC lesions. Demographic data for these patients are

shown in Supplementary Table 1.

Immunohistochemical Studies Tumor specimens were

obtained by partial hepatectomy from another 100

patients with HBV-related small HCC recruited

consecutively at our institution from 2000 to 2008 as the

second cohort (Supplementary Table 2). The clinical

specimen used for PCR validation is the same as samples

applied for microarray studies (as training set in this study).

The formalin-embedded tissues used for IHC staining (as

validation set in this study) were different from samples

applying for microarray and PCR experiment. Clinical and

pathological data were collected; all patients had regular

follow-up examinations. Tumor recurrence was diagnosed

by elevated AFP levels or when dynamic CT revealed

typical features. Treatment for recurrence included

transarterial chemoembolization (TACE), radiofrequency

ablation (RFA), or chemotherapy.

RNA Extraction, Microarray Data Processing, and

Affymetrix GeneChip Analyses

The experimental procedures of RNA extraction and

microarray data processing was as previously described.31

To filter the lower variance genes, 6,252 probe sets were

M.-C. Yu et al.

filtered from the original 22,215 using a standard deviation

[0.5. Unsupervised (principal component analysis [PCA])

and supervised (t test) methods have been used for the

analysis of gene expression data.32 The t tests statistical and

PCAs were performed with MATLAB version 7.4 and Sta-

tistic toolbox version 3.1 (The MathWorks, Boston, MA).33

The differentially expressed genes were selected with paired

t tests. A total of 78 probe sets were selected using the criteria

of undergoing a fourfold change in gene expression with

P \ 0.0001. Some probe sets were annotated with the same

genes; from the 20 sample pairs a total of 59 (42 downreg-

ulated and 17 upregulated) genes were defined as

differentially expressed genes. The profiles of the selected 59

genes were visualized by the Cluster 3.0 and Tree View

programs using average-linkage clustering algorithms.16

qRT-PCR

The first-strand cDNA was synthesized from 5 lg of total

RNA, and 0.5 ll of cDNA was mixed with the Taqman probe

(Assay-on-Demand, Applied Biosystems, Foster City, CA),

RNase-free water, and TaqMan Universal PCR Master Mix.

Real-time PCR was performed according to the manufac-

turer’s protocol. Glyceraldehyde-3-phosphate dehydro-

genase (GAPDH) was used as an internal control. Paired

samples were processed and mean the fold-change of

duplicate samples was calculated. For each normalized

sample, relative gene expression was calculated using the

differences in threshold cycles (Delta Ct, DCt), a calibrated

DCt value (DCt = Ct,internal control - Ct,sample). For quantita-

tion of gene expression levels between tumor and normal

tissue from the same patient, the comparative Ct method (also

known as the 2�DDCt method), where DDCt = DCt,tumor -

DCt,normal, was used. The catalog numbers of the Taqman

probes used for these gene assays are listed in Supple-

mental Table 3.

Immunohistochemistry

Formalin-fixed and paraffin-embedded resection speci-

mens were sectioned to 4 lm in thickness and then

deparaffinized, rehydrated, and processed for antigen

retrieval. The slides were further incubated with appropriate

dilutions of antibodies at room temperature for 1 h; anti-

bodies and dilutions used are listed in Supplemental Table 3.

All tumor and normal tissues were examined by an inde-

pendent pathologist and the expression level was scored on a

scale from -3 (underexpression) to ?3 (overexpression).

Statistical Analysis

The Wilcoxon signed ranks test was used for comparison

of the relative signal intensities of immunohistochemical

staining scores between paired cancerous and normal tis-

sues. Cases and controls were compared using t tests for

continuous variables and the v2 test for categorical vari-

ables. Unconditional logistic regression analysis was used

to calculate odds ratios (ORs) and 95% confidence intervals

(95% CIs). All analyses were performed using Statistical

Analysis Software for Windows, version 8.12 (SAS insti-

tute, Gary, NC).

RESULTS

Differential Gene Expression Profiles in HBV-Related

Small HCC

Differential gene expression patterns between HCC

tumor and normal tissue pairs fell into 2 distinct clusters,

PC1 and PC2, using PCA (Fig. 1a). The goal of PCA is to

reduce the dimensionality of the data matrix by finding new

small set variables, termed principal components, and to

allow the experimental conditions to be compared directly

and visualized in 2 or 3 dimensions.33 In this study, the first

2 principal components can explain 80% of the variance in

the original gene expression data. The highest levels of

differential gene expression had a fourfold change (log

2 [ 2 and P \ 0.0001) between tumor and normal tissue.

In total 59 (78 probe sets) out of 6,252 genes were selected

from array data by volcano plot (Fig. 1b) and are displayed

graphically in Fig. 1c. The most differentially expressed

genes in HCC tumor specimens are shown in Supplemental

Tables 4a and 4b: 17 genes were upregulated and 42 were

downregulated. The most downregulated genes were

involved in oxidation–reduction (the cytochrome P450

gene family: CYP1A2, CYP2A6, CYP2E1, …, etc.),

alcohol metabolism (alcohol dehydrogenase gene family:

ADH1A, ADH1B, ADH1C), lipid metabolism (cholesteryl

ester transfer protein, plasma [CETP], hydroxyacid oxidase

2 [HOA2], phosphoenolpyruvate carboxykinase 1 [PCK1],

and some of the cytochrome P450 gene family), bile acid

catabolic processing (CYP39A1), detoxification (metallo-

thionein [MT]: MT1E, MT1F, MT1G, MT1H, MT1M,

MT1P2, MT1X, MT2A), and the immune response family

(complement component: C9, C8A, C-type lectin domain

family 1, member B [C1EC1B], C-type lectin domain

family 4, member M [CLEC4M]); all of which might

reflect the downregulation of liver function and the

immune response during hepatocarcinogenesis (Supple-

mentary Table 4a).

In contrast, the most upregulated genes were involved

in cell proliferation (abnormal spindle like microcephaly-

associated protein [ASPM], geminin, DNA replication

inhibitor [GMNN], karyopherin alpha 2 [KPNA2],

nucleolar and spindle associated protein 1 [NUSAP1],

Aberrant Gene Expression in Small HBV–HCC

ribonucleotide reductase M2 [RRM2], sex-determining

region Y [SRY]-box 4 [SOX4]), tumor growth signal-

ing pathways (gamma-aminobutyric acid B receptor 1

[GABBR1], glypican3 [GPC3], collagen type I alpha 2

[COL1A2]), response to hypoxia (CD24), chemotaxis/cell

adhesion (osteopontin [OPN], chemokine [C–C motif]

ligand 20 [CCL20], roundabout, axon guidance receptor

homolog 1 [Drosophila], [ROBO1]), and cellular alde-

hyde metabolic processes (aldo–keto reductase family 1,

member B10 [AKR1B10]); all of which might reflect

the uncontrolled nature of cell proliferation involved

in tumor growth (Supplementary Table 4b). To summa-

rize: The most differentially expressed genes in HCC

tumors are involved in abnormal cell proliferation

and chemotaxis/cell adhesion, as well as in repression

of normal liver function and the normal immune

response.

Microarray Validation

Of the 20 paired samples described previously, only 17

offered enough specimens to validate the results of

microarray by qRT-PCR. The most upregulated and

downregulated probe sets were selected; these included 7

upregulated and 9 downregulated genes in the microarray

assay. The results of qRT-PCR for 16 selected genes were

compatible with the corresponding microarray data

(Fig. 2), and the correlation was substantial (Pearson’s

correlation, r = 0.87).

Expression of 8 Selected Genes in 100 HBV-Positive

Small HCC Specimens Validated by IHC Staining

To distinguish specific cell types in the tumor mass

having aberrant gene expression, consecutive tissue sec-

tions from 100 additional patients with HBV-positive small

HCC (as validation set in this study) were subjected to

immunohistochemical staining. The median age of these

patients was 51 (ranging from 24 to 81 years); 90 were

male and 10 female. All were positive for HBs Ag and

negative for anti-HCV. The median tumor diameter was

2.75 cm (range 0.6–5.0 cm) and AFP level was 533 ng/ml

(range 1–14,665 ng/ml). Pathologic variables included:

cirrhosis (67%), microscopic vascular invasion (23%),

satellite lesion (12%), and encapsulation (76%). One

patient died of surgical complications. The median follow-

up was 66.5 months. These data are shown in Supple-

mentary Table 2.

Among the most upregulated and downregulated of 59

genes, the most differentially expressed 8 genes were

selected for validation, including 3 upregulated gene

products (OPN, GPC3, and AKR1B10) and 5 downregu-

lated gene products (CYP2E1, ADH1B, MT, NNMT, and

HAMP). All HCC tumor and normal tissues were exam-

ined by two independent pathologists, and the expression

level was scored on a scale from -3 (underexpression) to

?3 (overexpression). The expression patterns of these

protein are shown in Fig. 3. Association of selected IHC

expression profiles with clinical variables in this cohort

was determined.

0.4

0.3

0.2

0.1

0.0

−0.1

0.1680.1640.1600.1560.1520.148

PC1 (relative units)

PC2 (relative units)

18

16

14

12

10

8

6

4

2

531−1− 43 20−2−4−5

Log2 (T/N)

−Log10 (P value)

FIG. 1 Differential transcripts

distinguish tumor and normal

tissue counterparts in 20 pairs of

HCC samples. a Principle

component analysis using

expression levels of 22,215

probe sets (yellow, normal

counterpart; blue, HCC).

b Selection of the most

differentially expressed

transcripts, based on t tests

(P \ 0.0001) and a fourfold

change in expression (yellowcolor = selected probesets).

c Unsupervised hierarchical

clustering of 59 genes. Profiles

visualized by the Cluster 3.0

and Tree View programs using

average-linkage clustering

algorithms

M.-C. Yu et al.

Elevated GPC3 and OPN Associated with Poor

Prognosis of Small HBV–HCC

To examine if the aberrant expression of 8 proteins

associated with any clinicopathological factors of small

HBV–HCC, we retrospectively analyzed 16 clinical vari-

ables of patients with these genes expression by IHC

scoring systems. Underexpression of CYP2E1 was signif-

icantly associated with a higher grade (P = 0.038),

whereas underexpression of ADH1B was significantly

8

6

4

2

0

−2

−4

−6

−8

Gene expressions(log2 ratio)

AD

H1A

20

15

10

5

0

−5

qRT-PCR( CT)ΔΔ Spearman rho = 0.87

10

Microarray (log2 ratio)OPM

AD

H1B

AD

H1C

CX

CL

12

CX

CL

2

CY

P231

HA

MP

MT

1G

NN

MT

AK

R1B

10

AN

XA

2

CC

L20

CD

24

GPC

3

SPIN

K1 86420−2

FIG. 2 Validation of 16 differentially expressed genes. a Microarray

transcript levels: yellow bars; qRT-PCR assays: blue bars. See text

for calculation methods. Glyceraldehyde-3-phosphate dehydrogenase

(GAPDH) used as internal control in qRT-PCR assays. b Expression

of SPINK1 was as example of correlation between microarray data

and qRT-PCR results (Pearson’s correlation, r = 0.87). X axes, genes

selected from microarray results; Y axes, expression levels for each

gene in qRT-PCR experiment

HAMP

ADH1B

CYP2E1

NNMT

MTs

OPN

AKR1B10

GPC3

(A) (E) (I) (M)

(B)

(C)

(F) (J) (N)

(G) (K) (O)

(D) (H) (L) (P)

Under-expression Under-expressionOver-expression Over-expression

FIG. 3 Expression pattern of 8 selected genes in HCC tumors by

evaluation of immunohistochemical staining. Expression (brownstaining) of CYP2E1, NNMT, and MTs indicated that the proteins

were localized in the cytoplasm of HCC tumor cells; whereas the

expression of GPC3, OPN, HAMP, and ADH1B showed both

cytoplasmic and membranous staining pattern in HCC tumor cells.

The expression level of each proteins was scoring by two independent

pathologists and the scoring for underexpressed (a–d and i–l) or

overexpression (e–h and m–p) of each proteins were showed in this

figure. HAMP hepcidine, ADH1B alcohol dehydrogenase gene family

1B, NNMT nicotinamide N-methyltransferase, CYP2E1 cytochrome

P450 gene family 2E1, MTs metallothionein family, AKR1B10 aldo–

keto reductase family 1, member B10, OPN osteopontin, GPC3glypican3

Aberrant Gene Expression in Small HBV–HCC

associated with a higher indocyanine green retention test at

15 min (ICG), lower AFP, and a lower grade (P = 0.042,

0.043, and 0.046, respectively) (Supplemental Table 5).

Overexpression of OPN was associated with higher

Aspartate aminotransferase (AST) levels, and tumors

without necrosis (P = 0.030 and 0.017, respectively)

(Supplemental Table 5).

Since 14 of the 100 patients had incomplete IHC data,

86 patients were enrolled in Cox proportional hazard

models. Firstly, The 10 clinicopathological variables were

analyzed in Cox proportional hazard models, and only

vascular invasion was the independent prognostic factor

(Table 1). Tumor size and rupture were not incorporated in

analysis because all enrolled cases were less than 5 cm, and

there was no tumor rupture in the series. Secondly,

expression of 8 proteins was analyzed with stratification of

vascular invasion, the significant clinical factors. Only

elevation of OPN or GPC3 were independent prognostic

factors for disease-free survival (DFS) in HBV-positive

small HCC (P \ 0.01 and P = 0.03, respectively)

(Table 2). As shown by Kaplan–Meier plots, long-term

DFS and overall survival (OS) of 10 years for small HBV–

HCC patients dichotomized into high risk (elevated both

OPN and GPC3; GPC3?/OPN?), and moderate (elevated

OPN or GPC3; OPN?/GPC3-; OPN-/GPC3?) or low risk

(decreased OPN and GPC3; OPN-/GPC3-) were DFS 0%,

OS 0% and DFS 35.0%, OS 46.5%, respectively. This

difference in DFS and OS was statistically significant by a

log-rank test (P \ 0.001 and P = 0.003, respectively)

(Fig. 4). Elevated both OPN and GPC3 may act as an

adverse indicator for HBV-related small HCC patients after

curative resection.

DISCUSSION

Treatment following early detection of cancer is

considered essential for good prognosis; however, a

portion of HBV–HCC patients do poorly in spite of

having small tumors at the time of diagnosis. Therefore,

finding an aberrant tumor-specific genetic signature for

potential clinical use remains a focus of research. In this

study, we correlated patients’ clinicopathological char-

acteristics with aberrant gene expression in HBV-positive

small HCC lesions and retrospectively analyzed clinical

outcomes. Our results suggest that clinicians might

consider more aggressive treatments for those patients

having high risks identified in this study and that they

should possibly have more rigorous follow-up after sur-

gical resection.

Heat shock protein 70 (HSP70), GPC3, telomerase

reverse-transcriptase (TERT), serine/threonine kinase 15

(STK15), phospholipase A2 (PLAG12B), LYVE-1, and

survivin have been shown to be prognostic factors by

genome-wide DNA microarray or qRT-PCR.27,30,34 To

improve long-term outcome, the search for validated, novel

biomarkers continues through the wide implementation of

surveillance programs. Using either a genetic or proteomic

approach, several potentially therapeutic molecular targets

have been identified and clinical trials are ongoing;

examples include therapies targeting growth receptors,

vascular endothelial growth factor receptor (VEGF), and

mTOR, for advanced HCC, as systemic or adjuvant treat-

ments.35 Among new molecular medicines, sorafenib,

targeted at VEGF and platelet-derived growth factor

receptor (PDGFR), has proved effective for advanced HCC

treatment in western and eastern countries.36

TABLE 1 Cox proportional hazard models for disease-free survival

(DFS): analysis of clinicopathological factors in 100 HCC patients

Variables HR 95.0% CI P

Vascular invasion (yes vs. no) 3.40 1.56–7.40 \0.01*

Satellite lesions (yes vs. no) 0.87 0.281–2.4 0.73

Edmondson–Steiner grade

(III, IV vs. I, II)

0.59 0.23–1.09 0.08

Cirrhosis (yes vs. no) 1.60 0.76–3.35 0.22

Encapsulation (yes vs. no) 0.70 0.32–1.55 0.38

AFP ([400 vs. B400 ng/ml) 0.71 0.33–1.54 0.39

AST ([72 vs. B72 U/l) 2.04 0.58–7.18 0.27

ALP ([94 vs. B94 U/l) 1.67 0.65–4.29 0.28

ALB (B3.5 vs. [3.5 g/dl) 1.35 0.30–6.07 0.69

Margin involved (yes vs. no) 1.27 0.98–1.65 0.07

AFP a-fetoprotein, AST aspartate aminotransferase, ALP alkaline

phosphatase, ALB albumin

* Indicated statistically significance: P \ 0.05

TABLE 2 Cox proportional hazard models for disease-free survival

(DFS): gene expression analysis

Aberrant gene expression HR 95.0% CI P

HAMP 0.90 0.45–1.78 0.75

ADH1B 0.89 0.43–1.85 0.76

NNMT 1.35 0.52–3.51 0.54

CYP2E1 0.52 0.23–1.18 0.12

MTs 0.61 0.19–1.94 0.41

AKR1B10 0.62 0.32–1.21 0.16

OPN 3.74 1.65–8.47 \0.01*

GPC3 2.18 1.07–4.45 0.03*

HR was compared with overexpression versus underexpression

HAMP hepcidine, ADH1B alcohol dehydrogenase gene family 1B,

NNMT nicotinamide N-methyltransferase, CYP2E1 cytochrome P450

gene family 2E1, MTs metallothionein family, AKR1B10 aldo–keto

reductase family 1, member B10, OPN osteopontin, GPC3 glypican3

* Indicated statistically significance: P \ 0.05

M.-C. Yu et al.

In this study, downregulation of genes related to fatty

acid and alcohol metabolic processes and oxidative

reduction was seen. In alcohol-mediated hepatocarcino-

genesis, ADH and CYP2E1 have been reported to interact

with reactive oxygen species and play a role in DNA

adduction; changes in early HCC lesions were correlated

with clinical outcomes.37 Furthermore, loss of DNA repair

mechanisms occurred in Glycine-N methyltransferase

(GNMT) knockout and aflatoxin B(1)-treated mice,

meaning that DNA adduction and methyltransferase were

affected at an early stage.38 Genes involved in detoxifica-

tion, such as cytochrome P450 family and NNMT, were

downregulated in this study, which may alter the cellular

microenvironment. Interestingly, CYP2E1 was induced

following exposure to alcohol and acetone or in nonalco-

holic steatohepatitis, but decreased expression of CYP2E1

in tumors has been reported to be associated with absence

of a tumor capsule and poorly differentiated HCC.39

GPC3, HSP70, adenylate cyclase-associated protein 2

(CAP2), and glutamine synthetase have been proposed as

early molecular markers for HCC in Japan; GPC3 was

assumed to be an oncofetal protein, a marker of liver

regeneration that is inactive in normal adult liver.40,41

GPC3 also functions as a secretory protein and has been

used as a serum marker (cutoff: 300 ng/l, sensitivity

47.0%, and specificity 93.5%) in HBV-positive HCC

patients.42

OPN was the most upregulated gene in our microarray

and qRT-PCR assays, but only 13 of 98 patients had high

expression in the IHC analysis; these data are compatible

with a previous study in which 24.2% of HCC patients had

upregulated OPN.43 There have been some studies dem-

onstrating that OPN is a useful marker for vascular

invasion, tumor grading, and predicting surgical out-

come.44,45 OPN has been identified as a transformation-

associated protein in some human cancers, such as stomach

and breast, and functions as a metastatic signal in HCC cell

lines.46,47 Furthermore, elevation of serum OPN has been

associated with advanced HCC, and a level of 1.85 log

10 ng/ml was significantly predictive of outcome following

transarterial chemoembolization.48 The current study

demonstrates that OPN can be an independent predictor of

outcome for patients with HBV-positive small HCC.

In conclusion, this study began with an oligonucleotide

microarray, which identified 59 genes undergoing expres-

sion changes early in the development of HBV-positive

small HCC. Microarray results were validated by qRT-

PCR and IHC analysis. In a cohort analysis of DFS,

aberrant tumor expression of OPN and GPC3 were prog-

nostic factors for recurrence and poor outcome. Therefore,

the elevation both of OPN and GPC3 was a genetic sig-

nature as one prognostic factor for DFS and OS in small

HBV–HCC following resection.

ACKNOWLEDGMENT We are grateful to all our colleagues in

the Department of Cancer Center, Pathology and the Genomic

Medicine Research Core Laboratory of Chang Gung Memorial

Hospital for their technical assistance. This study was supported

by Chang Gung Memorial Hospital (CMRPG361521-2, Chen MF;

CMRPG391001, Yu MC) and the National Science Council

(NMRPG380081, Yu MC), Taiwan.

REFERENCES

1. Bosch FX, Ribes J, Diaz M, Cleries R. Primary liver cancer:

worldwide incidence and trends. Gastroenterology. 2004;127:S5–

16.

100

80

60

40

20

90

70

50

30

10

Moderate to low risk (n = 80)High risk (n = 6)

100

Years

Proportional disease-free survival P < 0.001

8642 91 753

100

80

60

40

20

90

70

50

30

10

Moderate to low risk (n = 80)High risk (n = 6)

100

Years

Proportional overall survival P = 0.003

8642 91 753

(A) (B)

FIG. 4 Elevated expression of OPN and GPC3 in tumors was

associated with shorter overall survival (OS) and disease-free survival

(DFS) for HBV-related small HCC. Long-term DFS and OS for small

HBV–HCC patients with high risk (both elevated GPC3?/OPN?)

were DFS 0%, OS 0%, respectively; on the other hand, DFS and OS

in patients with moderate (elevated only one gene) or low (OPN-/

GPC3-) risk were 35.0 and 46.5%, respectively. a Kaplan–Meier plot

for 10-year DFS, P \ 0.001. b Kaplan–Meier plot for overall

survival, P = 0.003

Aberrant Gene Expression in Small HBV–HCC

2. Department of Health, Taiwan. Health statistics in Taiwan, 2006.

http://www.doh.gov.tw/ufile/doc/Chapter%202.pdf.

3. Chen DS. Hepatocellular carcinoma in Taiwan. Hepatol Res.2007;37:S101–5.

4. Chen CJ, Yang HI, Su J, Jen CL, You SL, Lu SN, et al. Risk of

hepatocellular carcinoma across a biological gradient of serum

hepatitis B virus DNA level. JAMA. 2006;295:65–73.

5. Lu SN, Su WW, Yang SS, Chang TT, Cheng KS, Wu JC, et al.

Secular trends and geographic variations of hepatitis B virus and

hepatitis C virus-associated hepatocellular carcinoma in Taiwan.

Int J Cancer. 2006;119:1946–52.

6. Qin LX, Tang ZY. The prognostic significance of clinical and

pathological features in hepatocellular carcinoma. World J Gas-troenterol. 2002;8:193–9.

7. Lee WC, Jeng LB, Chen MF. Estimation of prognosis after

hepatectomy for hepatocellular carcinoma. Br J Surg. 2002;89:

311–6.

8. Tung-Ping PR, Fan ST, Wong J. Risk factors, prevention, and

management of postoperative recurrence after resection of

hepatocellular carcinoma. Ann Surg. 2000;232:10–24.

9. Bruix J, Llovet JM. Prognostic prediction and treatment strategy

in hepatocellular carcinoma. Hepatology. 2002;35:519–24.

10. Yanaga K. Current status of hepatic resection for hepatocellular

carcinoma. J Gastroenterol. 2004;39:919–26.

11. Poon RT, Fan ST, Lo CM, Liu CL, Wong J. Long-term survival

and pattern of recurrence after resection of small hepatocellular

carcinoma in patients with preserved liver function: implications

for a strategy of salvage transplantation. Ann Surg. 2002;235:

373–82.

12. Zhou Y, Zhao Y, Li B, Xu D, Yin Z, Xie F, et al. Meta-analysis

of radiofrequency ablation versus hepatic resection for small

hepatocellular carcinoma. BMC Gastroenterol. 2010;10:78.

13. Carr BI, Kanke F, Wise M, Satomura S. Clinical evaluation of lens

culinaris agglutinin-reactive alpha-fetoprotein and des-gamma-

carboxy prothrombin in histologically proven hepatocellular car-

cinoma in the United States. Dig Dis Sci. 2007;52:776–82.

14. Malaguarnera G, Giordano M, Paladina I, Berretta M, Cappellani

A, Malaguarnera M. Serum markers of hepatocellular carcinoma.

Dig Dis Sci. 2010;55:2744–55.

15. Ge X, Yamamoto S, Tsutsumi S, Midorikawa Y, Ihara S, Wang

SM, et al. Interpreting expression profiles of cancers by genome-

wide survey of breadth of expression in normal tissues. Genom-ics. 2005;86:127–41.

16. Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis

and display of genome-wide expression patterns. Proc Natl AcadSci USA. 1998;95:14863–68.

17. Lee YS, Chen CH, Tsai CN, Tsai CL, Chao A, Wang TH.

Microarray labeling extension values: laboratory signatures for

Affymetrix GeneChips. Nucleic Acids Res. 2009;37:1–12.

18. Ho MC, Lin JJ, Chen CN, Chen CC, Lee H, Yang CY, et al. A

gene expression profile for vascular invasion can predict the

recurrence after resection of hepatocellular carcinoma: a micro-

array approach. Ann Surg Oncol. 2006;13:1474–84.

19. Wang SM, Ooi LL, Hui KM. Identification and validation of a

novel gene signature associated with the recurrence of human

hepatocellular carcinoma. Clin Cancer Res. 2007;13:6275–83.

20. Iizuka N, Oka M, Yamada-Okabe H, Nishida M, Maeda Y, Mori

N, et al. Oligonucleotide microarray for prediction of early

intrahepatic recurrence of hepatocellular carcinoma after curative

resection. Lancet. 2003;361:923–9.

21. Yoshioka S, Takemasa I, Nagano H, Kittaka N, Noda T, Wada H,

et al. Molecular prediction of early recurrence after resection of

hepatocellular carcinoma. Eur J Cancer. 2009;45:881–9.

22. Moribe T, Iizuka N, Miura T, Kimura N, Tamatsukuri S, Ishit-

suka H, et al. Methylation of multiple genes as molecular markers

for diagnosis of a small, well-differentiated hepatocellular car-

cinoma. Int J Cancer. 2009;125:388–97.

23. Okamoto M, Utsunomiya T, Wakiyama S, Hashimoto M, Fuku-

zawa K, Ezaki T, et al. Specific gene-expression profiles of

noncancerous liver tissue predict the risk for multicentric

occurrence of hepatocellular carcinoma in hepatitis C virus-

positive patients. Ann Surg Oncol. 2006;13:947–54.

24. Lee CF, Ling ZQ, Zhao T, Lee KR. Distinct expression patternsin hepatitis B virus- and hepatitis C virus-infected hepatocellular

carcinoma. World J Gastroenterol. 2008;14:6072–7.

25. Okabe H, Satoh S, Kato T, Kitahara O, Yanagawa R, Yamaoka

Y, et al. Genome-wide analysis of gene expression in human

hepatocellular carcinomas using cDNA microarray: identification

of genes involved in viral carcinogenesis and tumor progression.

Cancer Res. 2001;61:2129–37.

26. Nam SW, Park JY, Ramasamy A, Shevade S, Islam A, Long PM,

et al. Molecular changes from dysplastic nodule to hepatocellular

carcinoma through gene expression profiling. Hepatology.2005;42:809–18.

27. Paradis V, Bieche I, Dargere D, Laurendeau I, Laurent C, Bioulac

Sage P, et al. Molecular profiling of hepatocellular carcinomas

(HCC) using a large-scale real-time RT-PCR approach: deter-

mination of a molecular diagnostic index. Am J Pathol. 2003;

163:733–41.

28. Lee S, Kim S. Gene regulations in HBV-related liver cirrhosis

closely correlate with disease severity. J Biochem Mol Biol.2007;40:814–24.

29. Wurmbach E, Chen YB, Khitrov G, Zhang W, Roayaie S, Sch-

wartz M, et al. Genome-wide molecular profiles of HCV-induced

dysplasia and hepatocellular carcinoma. Hepatology. 2007;45:

938–47.

30. Chuma M, Sakamoto M, Yamazaki K, Ohta T, Ohki M, Asaka M,

et al. Expression profiling in multistage hepatocarcinogenesis:

identification of HSP70 as a molecular marker of early hepato-

cellular carcinoma. Hepatology. 2003;37:198–207.

31. Huang CC, Lee TJ, Chang PH, Lee YS, Chuang CC, Jhang YJ,

et al. Expression of cathepsin S and its inhibitor stefin A in sin-

onasal inverted papilloma. Rhinology. 2010;48:352–7.

32. Quackenbush J. Computational analysis of microarray data. NatRev Genet. 2001;2:418–27.

33. Raychaudhuri S, Stuart JM, Altman RB. Principal components

analysis to summarize microarray experiments: application to

sporulation time series. Pac Symp Biocomput. 2000:455–66.

34. Smith MW, Yue ZN, Geiss GK, Sadovnikova NY, Carter VS,

Boix L, et al. Identification of novel tumor markers in hepatitis C

virus-associated hepatocellular carcinoma. Cancer Res. 2003;63:

859–64.

35. Ishii H, Yamamoto J, Ikari T. Adjuvant treatments for resectable

hepatocellular carcinoma. J Hepatobiliary Pancreat Surg.2008;15:459–62.

36. Cheng AL, Kang YK, Chen Z, Tsao CJ, Qin S, Kim JS, et al.

Efficacy and safety of sorafenib in patients in the Asia-Pacific

region with advanced hepatocellular carcinoma: a phase III ran-

domised, double-blind, placebo-controlled trial. Lancet Oncol.2009;10:25–34.

37. Seitz HK, Stickel F. Molecular mechanisms of alcohol-mediated

carcinogenesis. Nat Rev Cancer. 2007;7:599–612.

38. Liu SP, Li YS, Lee CM, Yen CH, Liao YJ, Huang SF, et al.

Higher susceptibility to aflatoxin B(1)-related hepatocellular

carcinoma in glycine N-methyltransferase knockout mice. Int JCancer. 2011;128:511–23.

39. Ho JC, Cheung ST, Leung KL, Ng IO, Fan ST. Decreased

expression of cytochrome P450 2E1 is associated with poor

prognosis of hepatocellular carcinoma. Int J Cancer. 2004;111:

494–500.

M.-C. Yu et al.

40. Nakatsura T, Yoshitake Y, Senju S, Monji M, Komori H, Mo-

tomura Y, et al. Glypican-3, overexpressed specifically in human

hepatocellular carcinoma, is a novel tumor marker. BiochemBiophys Res Commun. 2003;306:16–25.

41. Sakamoto M. Early HCC: diagnosis and molecular markers.

J Gastroenterol. 2009;44:108–11.

42. Liu H, Li P, Zhai Y, Qu CF, Zhang LJ, Tan YF, et al. Diagnostic

value of glypican-3 in serum and liver for primary hepatocellular

carcinoma. World J Gastroenterol. 2010;16:4410–5.

43. Yang GH, Fan J, Xu Y, Qiu SJ, Yang XR, et al. Osteopontin

combined with CD44, a novel prognostic biomarker for patients

with hepatocellular carcinoma undergoing curative resection.

Oncologist. 2008;13:1155–65.

44. Korita PV, Wakai T, Shirai Y, Matsuda Y, Sakata J, Cui X, et al.

Overexpression of osteopontin independently correlates with

vascular invasion and poor prognosis in patients with hepato-

cellular carcinoma. Hum Pathol. 2008;39:1777–83.

45. Chen RX, Xia YH, Cui JF, Xue TC, Ye SL. Osteopontin, a single

marker for predicting the prognosis of patients with tumor-node-

metastasis stage I hepatocellular carcinoma after surgical resec-

tion. J Gastroenterol Hepatol. 2010;25:1435–42.

46. Takafuji V, Forgues M, Unsworth E, Goldsmith P, Wang XW.

An osteopontin fragment is essential for tumor cell invasion in

hepatocellular carcinoma. Oncogene. 2007;26:6361–71.

47. Ue T, Yokozaki H, Kitadai Y, Yamamoto S, Yasui W, Ishikawa

T, et al. Co-expression of osteopontin and CD44v9 in gastric

cancer. Int J Cancer. 1998;79:127–32.

48. Kim SH, Chung YH, Yang SH, Kim JA, Jang MK, Kim SE, et al.

Prognostic value of serum osteopontin in hepatocellular carci-

noma patients treated with transarterial chemoembolization.

Korean J Hepatol. 2009;15:320–30.

Aberrant Gene Expression in Small HBV–HCC