Imaging, Diagnosis, Prognosis
Association of MicroRNA Expression in Hepatocellular Carcinomas
with Hepatitis Infection, Cirrhosis, and Patient Survival
Jinmai Jiang,1 Yuriy Gusev,2 Ileana Aderca,3 Teresa A. Mettler,3 David M. Nagorney,3
Daniel J. Brackett,2 Lewis R. Roberts,3 and Thomas D. Schmittgen1
Abstract
Hepatocellular carcinoma (HCC) is the third most common
cause of death from cancer worldwide, with >600,000 cases in
2002 (1). The major etiologies of HCC include chronic liver
disease due to chronic hepatitis B or hepatitis C virus infection,
metabolic causes such as alcoholic or nonalcoholic steatohepatitis, hereditary hemochromatosis, and a-1-antitrypsin deficiency, or immune-related causes such as primary biliary
cirrhosis and autoimmune hepatitis. In parts of Asia and subSaharan Africa, dietary fungal aflatoxins have a synergistic effect
with chronic hepatitis in the pathogenesis of HCC (2 – 5).
Chronic liver injury with associated inflammation leads to
accelerated cycles of cell death, regeneration, and repair that
ultimately lead to premature senescence of the liver. As the
regenerative capacity of the liver becomes exhausted, aberrant
repair processes in the context of ongoing inflammation result
in the development of nodular regeneration, stromal expansion, and fibrosis, the end stage of which is called cirrhosis.
Cirrhosis is a major risk factor for the development of HCC;
individuals with cirrhosis have a 2% to 6% risk per year of
developing HCC (6). Previous studies have identified a number
of genetic and epigenetic alterations associated with cirrhosis,
including allelic imbalance at multiple genetic loci, p53
mutations, promoter hypermethylation of the p16INK4a tumor
suppressor gene, and telomere shortening with replicative
senescence and associated chromosomal instability. The development of HCC is associated with the development of
additional genetic and epigenetic alterations, coupled with
telomerase activation and consequent cellular immortalization
(7). Important molecules and pathways involved in hepatocarcinogenesis include cell cycle regulatory proteins such as
p53, c-Myc, and cyclin D1, the Wnt/h-catenin signaling
pathway, and multiple receptor tyrosine kinase growth factor
ligands and receptors, including epidermal growth factor,
Authors’ Affiliations: 1College of Pharmacy, Ohio State University, Columbus,
Ohio, 2Department of Surgery, University of Oklahoma Health Sciences Center and
Veterans Affairs Medical Center, Oklahoma City, Oklahoma, and 3Divisions of
Gastroenterology and Hepatology and Gastroenterological Surgery, Mayo Clinic
College of Medicine, Rochester, Minnesota
Received 3/5/07; revised 8/10/07; accepted 9/27/07.
Grant support: NIH grant CA107435 (T.D. Schmittgen), grant CA100882
(L.R. Roberts), the Richard M. Schulze Family Foundation, and the Miles and
Shirley Fiterman Center for Digestive Diseases at the Mayo Clinic, Rochester, MN.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance
with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Clinical Cancer Research
Online (http://clincancerres.aacrjournals.org/).
Requests for reprints: Thomas D. Schmittgen, College of Pharmacy, Ohio State
University, 500 West 12th Avenue, Columbus, OH 43210. Phone: 614-292-3456;
Fax: 614-292-7766; E-mail: schmittgen.2@ osu.edu.
F 2008 American Association for Cancer Research.
doi:10.1158/1078-0432.CCR-07-0523
www.aacrjournals.org
419
Clin Cancer Res 2008;14(2) January 15, 2008
Downloaded from http://aacrjournals.org/clincancerres/article-pdf/14/2/419/1977120/419.pdf by guest on 17 June 2022
Purpose: MicroRNA (miRNA) is a new class of small, noncoding RNA.The purpose of this study
was to determine if miRNAs are differentially expressed in hepatocellular carcinoma (HCC).
Experimental Design: More than 200 precursor and mature miRNAs were profiled by realtime PCR in 43 and 28 pairs of HCC and adjacent benign liver, respectively, and in normal liver
specimens.
Results: Several miRNAs including miR-199a, miR-21, and miR-301 were differentially expressed in the tumor compared with adjacent benign liver. A large number of mature and precursor miRNAs were up-regulated in the adjacent benign liver specimens that were both cirrhotic
and hepatitis-positive compared with the uninfected, noncirrhotic specimens (P < 0.01). Interestingly, all of the miRNAs in this comparison had increased expression and none were decreased.
The expression of 95 randomly selected mRNAs was not significantly altered in the cirrhotic and
hepatitis-positive specimens, suggesting a preferential increase in the transcription of miRNA.
Comparing the miRNA expression in the HCC tumors with patient’s survival time revealed two
groups of patients; those with predominantly lower miRNA expression and poor survival and
those with predominantly higher miRNA expression and good survival (P < 0.05). A set of
19 miRNAs significantly correlated with disease outcome. A number of biological processes including cell division, mitosis, and G1-S transition were predicted to be targets of the 19 miRNAs in
this group.
Conclusion: We show that a global increase in the transcription of miRNA genes occurs in cirrhotic and hepatitis-positive livers and that miRNA expression may prognosticate disease outcome in HCC.
Imaging, Diagnosis, Prognosis
fibroblast growth factor, hepatocyte growth factor, and vascular
endothelial growth factor, which activate the mitogen-activated
protein kinase and phosphoinositide-3-kinase/AKT kinase
pathways (8, 9).
MicroRNA (miRNA) are endogenously expressed, small
interfering RNAs, discovered during studies of Caenorhabditis
elegans development (10). miRNAs are transcribed as precursor
molecules that are subsequently processed into the active f21
nucleotide mature miRNA. The mature miRNA binds to the
3¶ untranslated region of the target mRNA through imperfect
base pairing, producing translational arrest and/or degradation
of the mRNA. Conceptually, miRNAs regulate gene expression
in a manner similar to transcription factors. Both miRNAs and
transcription factors are trans-acting factors that bind to
composite cis-regulatory elements that are ‘‘hard-wired’’ into
RNA and DNA, respectively (11). Although putative roles for
the vast majority of mammalian miRNAs remain unknown,
miRNAs have been implicated in a diverse number of
mammalian cellular processes including insulin secretion in
the pancreas (12), differentiation of adipocytes (13), and
regulation of embryonic stem cell development (14).
A growing number of both direct and indirect evidence
suggests a relationship between differential miRNA expression
and cancer. These include miR-15a and miR-16-1 in chronic
lymphocytic leukemia (15, 16), miR-143 and miR-145 in
colorectal cancer (17), let-7 in lung cancer (18, 19), and miR155 in diffuse large B cell lymphoma (20). Expression profiling
has identified other cancers with differential expression of
miRNAs including breast cancer (21), papillary thyroid cancer
(22), and glioblastoma (23, 24). A polycistron encoding
miRNAs miR-17, -18, -19a, -19b-1, and -92-1 is amplified in
human B-cell lymphomas and forced expression of the
polycistron along with c-myc was tumorigenic, suggesting that
this group of miRNAs may function as oncogenes (25).
The purpose of this study was to profile the expression of
miRNAs in clinical specimens of HCC, adjacent benign tissue,
and in liver specimens from nondiseased livers and to compare
the miRNA expression profiles among the patients with HCC
including those with cirrhosis and hepatitis infection.
Table 1. Clinical and pathologic features of
patients with HCC
28 (51.9%)
26 (48.1%)
5 (9.3%)
9 (16.7%)
2 (3.7%)
6 (11.1%)
1 (1.9%)
3 (5.6%)
10 (18.5%)
18 (33.3%)
5
27
18
4
(9.3%)
(50.0%)
(33.3%)
(7.4%)
24 (44.4%)
26 (48.1%)
4 (9.4%)
*Either HBV or HCV infection.
somewhat higher percentage of patients with no known risk factors in
our study is likely due to a combination of factors. First, a number of
studies from the United States have documented a relatively high
percentage (15-50%) of HCCs that have no known risk factors, this is
typical of the population of HCC patients that are seen in the Upper
Midwest (26, 27). Secondly, patients with cirrhosis are less likely to be
candidates for surgical resection. Studies of resected tissues therefore
tend to have a higher percentage of samples from patients without
cirrhosis and without known risk factors for HCC.
Real-time PCR for miRNA expression. RNA was treated with RNasefree DNase I and cDNA was synthesized from 600 ng of total RNA using
gene-specific primers to 182 miRNA precursors and U6 RNA as
previously described (28, 29). The expression of the miRNA precursors
was determined using real-time quantitative PCR assay, as described
with the exception that 35 cycles of PCR was used rather than 40 cycles
(28, 29). TaqMan miRNA Assays (Applied Biosystems) were used to
quantify mature miRNA. cDNA was synthesized by priming with 196
different gene-specific looped primers and the reverse primer for 18S
rRNA as previously described (30). All real-time PCR data was analyzed
using the comparative CT method, data were multiplied by 106 to
simplify presentation.
Real-time PCR for mRNA expression. Real-time PCR to measure the
mRNA expression of 95 genes was done using SYBR green detection and
standard techniques as mentioned above. cDNA was synthesized on
total RNA using random primers. PCR primers to the 95 genes were
collected from our laboratory archives or those of colleagues or were
randomly selected from the PrimerBank database (31). Gene expression
was presented relative to 18S rRNA.
Statistics. Differences between the various groups (benign/tumor,
cirrhosis/no cirrhosis, virus/nonvirus) were determined using the
Student’s t test. The Survival Risk Group Prediction algorithm (PAM
software package) was used to develop a miRNA expression – based
predictor of survival risk groups. The survival risk groups were
constructed using the supervised principal component method from
ref. (32). This method uses a Cox proportional hazards model to relate
survival time to k ‘‘supergene’’ expression levels, where k is selectable by
Materials and Methods
Tissue procurement and RNA isolation. Fifty-four pairs of HCC
tumors and adjacent benign liver were available for the study. For each
case, tumor samples with matched adjacent benign tissue were collected
during surgical resections at the Mayo Clinic between 1991 and 2001,
frozen in liquid nitrogen, and stored at -80jC. The study was approved
by the Mayo Clinic Institutional Review Board. Sections from each
specimen were examined by a pathologist and graded histologically.
Patients were classified for common etiology of HCC including
hepatitis B or C, cirrhosis, and alcoholic liver disease. Patients who
developed HCC in the context of a normal liver with no known risk
factors were negative for cirrhosis, hepatitis B, and hepatitis C. Total
RNA was isolated from the tissues using Trizol reagent (Invitrogen).
Some of the total RNA specimens were further purified using a Qiagen
midi column (Qiagen). Four normal liver samples were purchased from
Ambion as part of their First Choice total RNA normal human tissue.
These patients died from complications other than liver disease. Eight
additional normal liver tissues were received from Dr. Snorri S.
Thorgeirsson, National Cancer Institute. Clinical data on the patient
specimens are included in Table 1 and Supplemental Table S1. The
Clin Cancer Res 2008;14(2) January 15, 2008
74 (36-90)
44 (81.5%)
10 (18.5%)
420
www.aacrjournals.org
Downloaded from http://aacrjournals.org/clincancerres/article-pdf/14/2/419/1977120/419.pdf by guest on 17 June 2022
Age, y (median, range)
>60
<60
Gender
Male
Female
Etiology
HBV
HCV
HBV + HCV
ALD
ALD + viral infection*
Hemochromatosis
Other
No known risk factors
Edmonson grade
Grade 1
Grade 2
Grade 3
Grade 4
Cirrhosis
No
Yes
Unknown
MicroRNA Expression in LiverTissues
analysis was performed. This provides a regression coefficient (weight)
for each principal component. This method provides a prognostic index
for a patient with a log expression profile given by a set of miRNA
expression data. A high value of the prognostic index corresponds to a
high value of hazard of death, and consequently, a relatively poor
predicted survival. Unsupervised hierarchical cluster analysis was done
for samples and genes using mean centered miRNA expression data,
average linkage, and uncentered Pearson correlation as a distance.
the user (usually 1-3). The supergene expression levels are the first k
principal component, i.e., linear combinations of expression levels of
the subset of genes that are univariately correlated with survival. In our
analysis, we used the first three principal components. The significance
of each gene was measured based on a univariate Cox proportional
hazards regression of survival time versus the log expression level for
the gene. After selecting the genes, the principal component was
computed, and the k variable Cox proportional hazard regression
Downloaded from http://aacrjournals.org/clincancerres/article-pdf/14/2/419/1977120/419.pdf by guest on 17 June 2022
Fig. 1. miRNA expression in patients with cirrhosis and/or hepatitis infection.The expression of 182 precursors and 196 mature miRNAs was profiled in specimens of adjacent
benign tissue or normal liver. Approximately half of the patients had cirrhosis and one third were infected with hepatitis B or C. Comparisons were made between patients
infected with hepatitis B or C (V+) or not (V-), and those who had cirrhosis (C+) or not (C-). Points, P values from the Student’s t test comparing the miRNA expression in the
two groups or compared with liver from patients without HCC (N). Groups were considered statistically different at P < 0.01 (dashed line, P = 0.01). Precursor miRNAs are
presented in corresponding benign (A) or normal (B) liver tissue. C, P values from the expression of 95 randomly selected mRNAs. P values generated from the mature
miRNA expression profiling compared with benign (D) or normal (E) liver tissue.
www.aacrjournals.org
421
Clin Cancer Res 2008;14(2) January 15, 2008
Imaging, Diagnosis, Prognosis
Computational prediction of potential miRNA targets. A list of
predicted targets was generated for the group of 19 coexpressed
miRNAs that were identified by PAM survival analysis as strongly
associated with survival of patients with HCC. A combinatorial target
prediction algorithm was applied (miRgate 2.2 software suite,
Actigenics/Cepheid Europe). Initially, a list of all predicted target genes
which are targeted by any miRNA from that group, was generated.
Secondly, this list of potential targets was analyzed using gene ontology
(GO) enrichment analysis according to the total number of miRNAs
that were targeting the same GO categories in order to determine the
biological processes and functions that were most likely to be affected
by a group of miRNAs. A short list of the three top GO categories, which
are targeted by at least 80% of the miRNAs from the group, was
selected. A list of 84 target genes from those top three categories was
further analyzed using the Ingenuity Pathway Analysis system (IPA 5.0,
Ingenuity Systems). This method of miRNA combinatorial target
analysis has been described in details elsewhere (33).
miRNA expression in cirrhotic and hepatitis-positive tissues.
The expression of 182 miRNA precursors was determined in
Fig. 2. Precursor and mature miRNA expression
in benign liver tissue. A, the expression of miRNA
precursors was determined by real-time PCR in
specimens of uninfected, noncirrhotic liver
(gray columns, n = 17) and virally infected, cirrhotic
livers (open columns, n = 12). B, the expression of
mature miRNA was determined by real-time PCR
in specimens of uninfected, noncirrhotic liver
(gray columns, n = 17) and virally infected, cirrhotic
livers (open columns, n = 9). *, P < 0.05.
Clin Cancer Res 2008;14(2) January 15, 2008
422
www.aacrjournals.org
Downloaded from http://aacrjournals.org/clincancerres/article-pdf/14/2/419/1977120/419.pdf by guest on 17 June 2022
Results
43 specimens of HCC, 43 adjacent benign tissues, and in 12
normal liver tissues using real-time PCR. The raw data from the
expression profiling is included in Supplemental Table S2.
About half of the HCC patients had underlying cirrhosis and
one third was infected with hepatitis B and/or hepatitis C
(Table 1). To determine the contribution of hepatitis infection
or cirrhosis to miRNA precursor expression, the data from the
adjacent benign tissues was classified into the following four
categories: V-C-, V-C+, V+C-, and V+C+, where V represents viral
hepatitis infection (hepatitis B or C infection) and C represents
cirrhosis. The number of patients in each of the four categories
was: V-C-, 17 patients; V-C+, 10 patients; V+C-, 4 patients; and
V+C+, 12 patients.
Comparing the miRNA precursor expression in the V-C- and
+ +
V C groups yielded the largest number of differentially
expressed miRNAs, with 50 miRNA precursors differentially
expressed in the benign tissue of patients with hepatitis
infection and cirrhosis compared with those patients with
histologically normal livers (Supplemental Table S3; Fig. 1A).
The expression of all 50 of the miRNA precursors is increased
MicroRNA Expression in LiverTissues
Table 2. Mature miRNA expression in HCC and adjacent benign tissue
miRNA
Relative gene expression
P
Mean benign
Tumor/benign
Tumor vs. benign
0.10
15.57
0.03
0.21
0.35
0.40
0.02
6.09
0.01
1.23
0.00
0.51
1.99
0.44
9.59
0.15
0.05
7.38
0.01
0.43
0.13
0.04
0.06
14.08
0.03
3.22
0.01
1.23
4.78
0.13
20.37
0.07
2.03
2.11
3.68
-2.07
2.67
10.5
-2.51
-2.31
-2.59
-2.62
-2.32
-2.42
-2.40
3.42
-2.12
2.19
0.039718
0.001165
0.000381
0.037011
0.000566
0.029641
0.000002
0.041504
0.012791
0.003364
0.020543
0.034570
0.006759
0.011714
0.000311
0.021348
in the V+C+ group (z2-fold, P < 0.01) and none had reduced
expression. When the V+C+ benign tissues were compared with
normal liver in patients without HCC, 20 miRNA precursors
had increased expressions (z2-fold, P < 0.01). Twelve of the
20 miRNAs were increased in the benign tissue comparison
(Supplemental Table S3; Fig. 1). These data show that the
expression of a large number of miRNA precursors was
increased in patients with cirrhosis and concomitant hepatitis
infection.
To determine if the increase in miRNA precursor expression
was due to cirrhosis or viral hepatitis infection alone, we
compared the V-C-/V-C+ and V-C-/V+C- groups (benign tissues).
In these comparisons, only one of the miRNA precursors was
significantly changed (z2-fold, P < 0.01; Fig. 1A). Comparing
the V-C+ or V+C- to normal liver yielded four and two
significantly changed miRNAs, respectively (Fig. 1B). These
data suggest that neither hepatitis viral infection nor cirrhosis
alone is sufficient to induce major changes in miRNA
expression, however the combination of both viral hepatitis
infection and cirrhosis significantly enhances the expression of
a large number of miRNA precursors.
To determine if the increase in miRNA precursor expression
of the virally infected specimens had a preference for hepatitis B
or hepatitis C, we compared the differences in the miRNA
expression in the hepatitis B (n = 3) and hepatitis C (n = 6)
samples (cirrhosis-positive group only). Comparing the miRNA
expression between these groups yielded nine miRNA precursors (miR-145, -9-2, -138-1,2, -320, -33, -10a,b, -21, -146,
and -220) that were differentially expressed (>2-fold increased
expression; P < 0.05). For all nine miRNAs, expression in the
hepatitis B – infected patients was greater than those infected
with hepatitis C. These data suggest that hepatitis B is a greater
contributor to the increase in expression compared with
hepatitis C, although it must be noted that the sample size
was small.
miRNA versus mRNA expression. Our data describe two
rather unusual findings: (a) a large percentage of miRNA
precursors are differentially expressed in liver tissues that were
hepatitis-positive and cirrhotic, and (b) in each of these cases,
the miRNA expression was increased, not a single statistically
significant decrease in miRNA expression was observed. This
www.aacrjournals.org
observation could be explained by biological or technical
factors. Possibilities include fundamental differences in the
regulation of miRNAs compared with the regulation of mRNAs,
or that the sensitive, real-time PCR used to profile miRNA gene
expression biased the results in some way. To address the
possibility that miRNA genes are expressed differently than
mRNA in these tissues, the expression of 95 randomly selected
mRNA genes was quantified in cDNA from the identical RNA
used in the miRNA expression profiling study. The entire list of
genes studied is presented in Supplemental Table S4. Only 3 of
the 95 (3%) mRNAs were differentially expressed in the benign
tissues, compared with 50 of the 182 (27%) miRNAs (z2-fold,
P < 0.01; Fig. 1C). Of the differentially expressed mRNAs, two
were up-regulated and one was down-regulated (Supplemental
Table S4). These data suggest that transcription of miRNA genes
was more predominantly increased compared with mRNA
genes in liver tissues that are cirrhotic and infected with
hepatitis virus.
Technical validation. The total RNA available in the archives
of one of the authors (L.R. Roberts) was isolated with Trizol
reagent followed by a clean-up using a Qiagen Midi column
(Qiagen). Column purification has the potential of removing
smaller RNAs. We determined that the expression of the 106
nucleotide U6 RNA were comparable in filtered and unfiltered
RNA (data not shown). The miRNA precursor expression in
RNA from normal liver purchased from Ambion (and not
filtered) was comparable to that in the filtered, benign tissue
(Supplemental Table S2).
The real-time PCR method that was used quantifies the
miRNA precursors (28, 29) and not the mature miRNA. We
and others have shown that in most cases, miRNA precursor expression correlates with the mature miRNA expression
(28 – 30, 34 – 37). However, situations exist in which the
precursor expression does not correlate with the mature miRNA
expression (29). Although column filtration did not alter the
miRNA precursor levels, column filtration did remove a large
portion of the f21-nucleotide mature miRNAs (data not
shown). Because the mature miRNA is the active species, we
wanted to validate the expression of the miRNA precursors. RNA
was isolated by the Trizol procedure from liver tissues that were
available in our archives; this included 13 of the 43 specimens
423
Clin Cancer Res 2008;14(2) January 15, 2008
Downloaded from http://aacrjournals.org/clincancerres/article-pdf/14/2/419/1977120/419.pdf by guest on 17 June 2022
miR-18
miR-21
miR-33
miR-101
miR-130b
miR-135a
miR-139
miR-150
miR-199a
miR-199a*
miR-199b
miR-200b
miR-214
miR-221
miR-223
miR-301
Fold change
Mean tumor
Imaging, Diagnosis, Prognosis
PCR assay for mature miRNA (38). Like the miRNA precursors,
the mature miRNA expression was increased in the V+C+
specimens compared with the V-C- specimens, however, only
6 of the 10 comparisons were statistically significant (Fig. 2).
Even though there was good correlation between the
expression of precursor and mature miRNA in these samples
(Fig. 2), we wanted to profile several hundred mature miRNAs
listed in Table 1. Total RNA was isolated from another 13 liver
tissues that were not among the 43 profiled for miRNA
precursor expression (patient data are presented in Supplemental Table S1). Ten miRNAs were selected from the 50 miRNA
precursors that significantly differed among the V-C- and V+C+
liver tissues (Fig. 1A). The expression of these 10 miRNAs was
measured in the 26 benign liver specimens using a real-time
Clin Cancer Res 2008;14(2) January 15, 2008
424
www.aacrjournals.org
Downloaded from http://aacrjournals.org/clincancerres/article-pdf/14/2/419/1977120/419.pdf by guest on 17 June 2022
Fig. 3. Prognostic value of miRNA
expression in patients with HCC. A, Kaplan-Meyer
survival curves for two groups of patients.
Prognostic scores for each patient were estimated
using semisupervised principal component analysis
(PAM algorithm). Red, data for patients with low
prognostic scores (group 1, higher survival time);
green, patients with high prognostic scores
(group 2, poorer survival; P < 0.05). B, heatmap
of the results of hierarchical clustering of samples
and genes. Expression data of 19 miRNAs are
presented for 25 tumor samples from patients
with known survival times. Nineteen miRNAs
were identified by supervised principal
component analysis (PAM algorithm) as being
significantly associated with survival risk.Two major
clusters of samples showed distinct patterns of
miRNA expression of these 19 genes and
correspond (except for two samples) with two
classes of patients predicted by supervised
principal component analysis (PAM algorithm) as
having good (group 1) and poor survival (group 2).
Red, expression levels above the pool average;
green, expression levels below the pool average;
black, no difference compared with the pool
average; and gray, undetectable expression.
MicroRNA Expression in LiverTissues
using a commercially available real-time PCR assay (38). RNA
was isolated by the Trizol procedure from liver tissues that were
available in our archives. This includes another 11 liver tissues
that were not among the 43 profiled for miRNA precursor
expression. In addition to these 11 patients, we were able to
obtain unfiltered RNA from 17 of the original 43 patients.
Therefore, mature miRNA was profiled on a total of 28 patients
(patient data are presented in Supplemental Table S1). The
number of patients in each of the four categories was: V-C-,
15 patients; V-C+, 4 patients; V+C-, 2 patients; and V+C+,
7 patients. Fourteen mature miRNAs were increased in the V+C+
compared with the V-C- group (benign samples, P < 0.01;
Fig. 1D). Comparing the mature miRNA expression in the V+C+
livers to the normal livers showed that 61 mature miRNAs were
increased (Fig. 1E). The small number of samples in the V-C+
and V+C- groups precluded us from analyzing these data. miR181b and miR-214 were increased in all four comparisons
(Fig. 1A, B, D, and E; Supplemental Table S4).
miRNA expression in HCC tissues. The expression of 196
mature miRNAs was compared in 28 specimens of HCC and
adjacent benign tissues and in 6 normal liver tissues using realtime PCR. The raw data from the expression profiling is
included in Supplemental Table S3. The mature miRNA
expression in the HCC was compared with the adjacent benign
or the normal liver tissue. Differentially expressed miRNA were
defined as those having a 2-fold or greater change in gene
expression and P < 0.05 (Student’s t test). Sixteen miRNAs were
differentially expressed when the tumor data was compared
with adjacent benign tissue (Table 2). The mature miRNA
expression data is presented as a heatmap (Supplemental
Fig. S1). Hierarchical clustering yielded four clusters. Clusters 3
and 4 contained only tumor, cluster 1 contained only benign
tissue, and cluster 2 contained all benign plus two tumors.
Prognostic benefit of miRNA signature in HCC. To determine
if a relationship exists between miRNA expression and the
patient’s survival, additional analyses were done on the mature
miRNA expression data from the tumors of patients with
unfiltered RNA and for which the survival time was known
(25 patients). A semisupervised analysis of survival risks was
conducted using the survival risk prediction algorithm (PAM
software package) and Kaplan-Meier estimates of disease-free
survival. Unfiltered expression data for all genes and survival
times were analyzed using the PAM program and showed two
groups of patients with significantly different survival curves;
group 1 had better survival rates and group 2 had poor survival
rates (P < 0.05; Fig. 3A; Table 3). PAM analysis of correlation
between miRNA expression and survival determined that a set
of 19 genes significantly correlated with the outcome of disease
(P < 0.05, Supplemental Table S5). These included let-7c,
let-7g, miR-26b, -29c, -30e-3p, -31, -99a, -99b, -100, -125b,
-139, -148a, -150, -200c, -220, -221, -345, -372, and -377.
Hierarchical clustering of the expression data from these 19 genes
showed that they correspond (except for two samples) with the
patient survival groups; patients in group 1 (good survival rates)
had overall higher mature miRNA expression levels compared
with those in group 2 (poor survival rates; Fig. 3B).
Prediction of miRNA targets. miRNAs function by binding to
conserved sequences within the 3¶ untranslated regions of their
respective, target mRNAs. Using a combinatorial target prediction algorithm and the gene ontology enrichment analysis, we
determined biological processes categories (GO BP) that are
Patient ID
10T
14T
26T
32T
33T
44T
69T
85T
86T
87T
88T
91T
92T
94T
97T
99T
103T
109T
309T
312T
337T
361T
363T
367T
368T
Group
Status
Survival (mo)
1
2
2
2
1
2
2
1
1
2
2
1
2
1
2
1
2
1
1
1
2
2
1
1
1
0
1
1
0
1
0
1
1
0
1
1
0
1
1
1
0
1
0
0
0
1
1
1
0
0
240
29
17
201
3
127
70
81
146
9
18
79
18
38
20
73
30
127
118
82
3
6
6
114
122
NOTE: Censoring status: 0, alive; 1, deceased.
targeted by multiple miRNAs from the group of 19 miRNAs
which we found to be associated with patient survival. Three
of these GO BP categories (cell division, mitosis, and G1-S
transition of mitotic cell cycle) are targeted by at least 80% of
miRNAs from this group (Table 4). The top three categories
which are most likely to be collectively affected by this group of
miRNAs include a total of 84 predicted target genes (Supplemental Table S6), all of which are related to cell division and/or
cell cycle regulation. Thirty-nine genes from this group of 84
were reported to be associated with several types of human
cancers. Nine of these genes were reported to be associated with
liver cancer (ACVR1B, APC11, CCND1, CDC25A, CDKN3,
HGF, PLK1, RAN, and TPX2). Because miRNA negatively
regulates translation, the expression of these 19 miRNAs could
regulate the protein levels of these cell cycle – related targets, for
example, by producing reduced protein expression in the good
survivors or increased protein levels in the poor survivors. A list
of relevant biological functions and diseases that are known to
be associated with the 84 genes from Supplemental Table S6 are
presented in Supplemental Table S7.
Discussion
We report the results of PCR-based, miRNA expression
profiling study in liver tissues. Our data show that a large
number of miRNAs have increased expression in the hepatitispositive and cirrhotic specimens compared with the liver tissues
that developed HCC in the context of a histologically normal
liver or in normal liver not associated with HCCs. More than 20
miRNA precursors were differentially expressed in the cirrhotic
and hepatitis-positive tissues (Fig. 1; Supplemental Table S3).
425
Clin Cancer Res 2008;14(2) January 15, 2008
Downloaded from http://aacrjournals.org/clincancerres/article-pdf/14/2/419/1977120/419.pdf by guest on 17 June 2022
www.aacrjournals.org
Table 3. Patient identifiers along with grouping,
survival time, and censored status
Imaging, Diagnosis, Prognosis
Table 4. Enrichment of biologically significant targets of 19 miRNAs from PAM analysis: top three GO
biological process categories
Biological process
Cell division (GO:0051301)
Mitosis (GO:0007067)
G1-S transition of mitotic cell cycle (GO:0000082)
Significance
No. of target genes
0.0174
0.0126
0.0317
61
46
15
miRNA
100% (19/19)
100% (19/19)
84% (16/19)
Acknowledgments
We thank Caifu Chen for his assistance with the TaqMan mature miRNA assays,
Snorri Thorgeirsson for providing RNA from liver specimens, and Wendy Frankel
for her assistance with the pathology.
References
1. Ferlay J, Bray F, Pisani P, et al: GLOBOCAN 2002:
cancer incidence, mortality and prevalence worldwide. In IARC Cancer Base no. 5. version 2.0: IARC
Press, 2002.
2. Bosch FX, Ribes J, Borras J. Epidemiology of primary
liver cancer. Semin Liver Dis 1999;19:271 ^ 85.
3. Jackson PE, Groopman JD. Aflatoxin and liver cancer. Baillieres Best Pract Res Clin Gastroenterol 1999;
13:545 ^ 55.
4. Bruix J, Llovet JM. Hepatitis B virus and hepatocellular carcinoma. J Hepatol 2003;39 Suppl 1:
S59 ^ 63.
5. Bruix J, Boix L, Sala M, et al. Focus on hepatocellular
carcinoma. Cancer Cell 2004;5:215 ^ 9.
6. Okuda H. Hepatocellular carcinoma development in
cirrhosis. Best Pract Res Clin Gastroenterol 2007;21:
161 ^ 73.
7. Herath NI, Leggett BA, MacDonald GA. Review
Clin Cancer Res 2008;14(2) January 15, 2008
426
of genetic and epigenetic alterations in hepatocarcinogenesis. J Gastroenterol Hepatol 2006;21:
15 ^ 21.
8. Roberts LR, Gores GJ. Hepatocellular carcinoma:
molecular pathways and new therapeutic targets.
Semin Liver Dis 2005;25:212 ^ 25.
9. Lee JS, Thorgeirsson SS. Comparative and integrative functional genomics of HCC. Oncogene 2006;
25:3801 ^ 9.
www.aacrjournals.org
Downloaded from http://aacrjournals.org/clincancerres/article-pdf/14/2/419/1977120/419.pdf by guest on 17 June 2022
Comparing the miRNA expression within the patients with
HCC, two different groups of patients emerged: those with
reduced miRNA expression and poor survival and those
with increased miRNA expression and good survival (Fig. 3;
Supplemental Fig. S1). A set of 19 miRNA genes significantly
correlated with disease outcome (Fig. 3; Supplemental
Table S5). Many of the predicted targets of these miRNAs
regulate cell division, mitosis, and G1-S transition (Table 4). In
addition to these theoretical issues, it may be possible to
prognosticate patients with HCC based on their miRNA
expression, similar to what was done for patients with lung
cancer (19, 37). Additional testing of the prognostic value of
this miRNA expression signature needs to be done on an
independent set of samples.
In summary, we have identified a number of miRNAs that are
differentially expressed in HCC. Our data also suggest that
important changes in miRNA expression occur during the
development of chronic viral hepatitis and cirrhosis, and that
the combination of viral hepatitis and cirrhosis is significantly
more likely to result in changes in miRNA expression.
Subsequent changes in miRNA expression in the transition
from cirrhosis to HCC are much less marked. It is possible that
most of the miRNA changes that occur during carcinogenesis
occur early, so that by the stage of replicative senescence,
characteristic of cirrhosis, the key miRNA changes that may
predispose to carcinogenesis have already occurred. These
miRNA changes may reflect and/or mediate a preneoplastic
‘‘field effect’’ in individuals with cirrhosis and hepatitis who are
at an increased risk of the development of HCC. Studies of
mRNA gene expression analysis during the process of liver
carcinogenesis show significant changes in mRNA expression,
both in the process of development of cirrhosis and in the
transition from cirrhosis to HCC. The results of our miRNA
study are in contrast with this and may suggest fundamental
differences in the regulation of miRNAs and mRNAs in
carcinogenesis that are worthy of further study.
Surprisingly, the expression of all of these miRNAs was
increased in the hepatitis-positive, cirrhotic tissues—and not a
single miRNA was reduced to a statistically significant degree. A
recent study in human lymphoblastoid cells reports a global
increase in miRNA expression in response to stress induced by
folate deficiency (39). Many of the miRNAs increased in
response to cellular stress (e.g., miR-22, miR-182b, miR-198,
miR-221, and miR-222) reported in ref. (39), were increased in
cirrhotic and hepatitis-positive livers (Supplemental Table S3),
suggesting that increased transcription of certain miRNAs may
be a general response to stress induced by a combination of
hepatitis and cirrhosis.
Other interesting observations from our study is that the
increased miRNA expression in cirrhotic and hepatitis-positive
liver specimens is modest in magnitude (2- to 3-fold),
consistent across most samples, and highly statistically significant (Supplemental Table S3). The 2- to 3-fold global increase
in miRNA expression due to cellular stress from folate
deficiency (39) is comparable in magnitude to that reported
here. The ability to distinguish modest changes in gene
expression was likely facilitated by the fact that we employed
sensitive, real-time PCR to profile the miRNA expression. The
biological significance of a 2- to 3-fold change in miRNA
expression is unknown. Because one miRNA may regulate
scores of target genes and one mRNA may be regulated by
multiple miRNAs (40), a 2- to 3-fold change in the expression
of multiple miRNAs may profoundly alter the regulation of
downstream genes.
The only other study to profile miRNA expression in HCC, to
our knowledge, is that of Murakami et al. (41). Of the seven
miRNAs reported as differentially expressed in HCC by
Murakami et al. (41), three were differentially expressed in
our study (Mir-199a, Mir-199a*, and Mir-18; Table 2). The
number of differentially expressed miRNAs in the entire group
of HCC specimens reported in Table 2 is relatively small
compared with the large number of differentially expressed
miRNAs identified by expression profiling in other cancers
(15, 21, 24, 37, 42 – 44). However, three of the miRNAs with
increased expression here were increased in other solid tumors
including Mir-21 in glioblastoma (23), breast cancer (21) and
pancreas cancer (30); Mir-221 in pancreas (30, 45) and thyroid
cancer (22) and Mir-301 in pancreas cancer (30).
MicroRNA Expression in LiverTissues
www.aacrjournals.org
23. Chan JA, KrichevskyAM, Kosik KS. MicroRNA-21is
an antiapoptotic factor in human glioblastoma cells.
Cancer Res 2005;65:6029 ^ 33.
24. Ciafre SA, Galardi S, Mangiola A, et al. Extensive
modulation of a set of microRNAs in primary glioblastoma. Biochem Biophys Res Commun 2005;334:
1351 ^ 8.
25. He L,Thomson JM, Hemann MT, et al. A microRNA
polycistron as a potential human oncogene. Nature
2005;435:828 ^ 33.
26. El-Serag HB, Rudolph KL. Hepatocellular carcinoma: epidemiology and molecular carcinogenesis. Gastroenterology 2007;132:2557 ^ 76.
27. El-Serag HB, Davila JA, Petersen NJ, et al. The continuing increase in the incidence of hepatocellular carcinoma in the United States: an update. Ann Intern
Med 2003;139:817 ^ 23.
28. SchmittgenTD, JiangJ, Liu Q, et al. A high-throughput method to monitor the expression of microRNA
precursors. Nucleic Acids Res 2004;32:E43.
29. Jiang J, Lee EJ, GusevY, et al. Real-time expression
profiling of microRNA precursors in human cancer cell
lines. Nucleic Acids Res 2005;33:5394 ^ 403.
30. Lee EJ, Gusev Y, Jiang J, et al. Expression profiling
identifies microRNA signature in pancreatic cancer. Int
J Cancer 2007;120:1046 ^ 54.
31. Wang X, Seed B. A PCR primer bank for quantitative gene expression analysis. Nucleic Acids Res
2003;31:e154.
32. Bair E, Tibshirani R. Semi-supervised methods to
predict patient survival from gene expression data.
PLoS Biol 2004;2:E108.
33. Gusev Y, Schmittgen TD, Lerner MR, et al. Computational analysis of biological functions and pathways
collectively targeted by co-expressed microRNAs in
cancer. BMC Bioinformatics 2007;8:S16.
34. Jiang J, Lee EJ, Schmittgen TD. Increased expression of microRNA-155 in Epstein-Barr virus
transformed lymphoblastoid cell lines. Genes Chromosomes Cancer 2006;45:103 ^ 6.
427
35. Meng F, Henson R, Lang M, et al. Involvement of
human micro-RNA in growth and response to chemotherapy in human cholangiocarcinoma cell lines. Gastroenterology 2006;130:2113 ^ 29.
36. Liu CG, Calin GA, Meloon B, et al. An oligonucleotide microchip for genome-wide microRNA profiling in
human and mouse tissues. Proc Natl Acad Sci U S A
2004;101:9740 ^ 4.
37. Yanaihara N, Caplen N, Bowman E, et al. Unique
microRNA molecular profiles in lung cancer diagnosis
and prognosis. Cancer Cell 2006;9:189 ^ 98.
38. Chen C, Ridzon DA, Broomer AJ, et al. Real-time
quantification of microRNAs by stem-loop RT-PCR.
Nucleic Acids Res 2005;33:e179.
39. Marsit CJ, Eddy K, Kelsey KT. MicroRNA responses
to cellular stress. Cancer Res 2006;66:10843 ^ 8.
40. Miranda KC, Huynh T, Tay Y, et al. A pattern-based
method for the identification of MicroRNA binding
sites and their corresponding heteroduplexes. Cell
2006;126:1203 ^ 17.
41. Murakami Y,YasudaT, Saigo K, et al. Comprehensive
analysis of microRNA expression patterns in hepatocellular carcinoma and non-tumorous tissues. Oncogene 2005;25:2537 ^ 45.
42. Volinia S, Calin GA, Liu CG, et al. A microRNA expression signature of human solid tumors defines cancer gene targets. Proc Natl Acad Sci U S A 2006;103:
2257 ^ 61.
43. Lu J, Getz G, Miska EA, et al. MicroRNA expression
profiles classify human cancers. Nature 2005;435:
834 ^ 8.
44. Roldo C, Missiaglia E, Hagan JP, et al. MicroRNA
expression abnormalities in pancreatic endocrine and
acinar tumors are associated with distinctive pathologic features and clinical behavior. J Clin Oncol
2006;24:4677 ^ 84.
45. Bloomston M, Frankel WL, Petrocca F, et al. MicroRNA expression patterns to differentiate pancreatic
adenocarcinoma from normal pancreas and chronic
pancreatitis. JAMA 2007;297:1901 ^ 8.
Clin Cancer Res 2008;14(2) January 15, 2008
Downloaded from http://aacrjournals.org/clincancerres/article-pdf/14/2/419/1977120/419.pdf by guest on 17 June 2022
10. Bartel DP. MicroRNAs: genomics, biogenesis,
mechanism, and function. Cell 2004;116:281 ^ 97.
11. Hobert O. Common logic of transcription factor and
microRNA action.TrendsBiochem Sci 2004;29:462 ^ 8.
12. Poy MN, Eliasson L, Krutzfeldt J, et al. A pancreatic
islet-specific microRNA regulates insulin secretion.
Nature 2004;432:226 ^ 30.
13. Esau C, Kang X, Peralta E, et al. MicroRNA-143 regulates adipocyte differentiation. J Biol Chem 2004;
279:52361 ^ 5.
14. Houbaviy HB, Murray MF, Sharp PA. Embryonic
stem cell-specific microRNAs. Dev Cell 2003;5:
351 ^ 8.
15. Calin GA, Liu CG, Sevignani C, et al. MicroRNA profiling reveals distinct signatures in B cell chronic lymphocytic leukemias. Proc Natl Acad Sci U S A 2004;
101:11755 ^ 60.
16. Calin GA, Sevignani C, Dumitru CD, et al. Human
microRNA genes are frequently located at fragile sites
and genomic regions involved in cancers. Proc Natl
Acad Sci U S A 2004;101:2999 ^ 3004.
17. Michael MZ, SM OC, van Holst Pellekaan NG, et al.
Reduced accumulation of specific microRNAs in colorectal neoplasia. Mol Cancer Res 2003;1:882 ^ 91.
18. Johnson S, Grosshans H, Shingara J, et al. RAS is
regulated by the let-7 microRNA family. Cell 2005;
120:635 ^ 47.
19. Takamizawa J, Konishi H, Yanagisawa K, et al. Reduced expression of the let-7 microRNAs in human
lung cancers in association with shortened postoperative survival. Cancer Res 2004;64:3753 ^ 6.
20. Eis PS, Tam W, Sun L, et al. Accumulation of miR155 and BIC RNA in human B cell lymphomas. Proc
Natl Acad Sci U S A 2005;102:3627 ^ 32.
21. Iorio MV, Ferracin M, Liu CG, et al. MicroRNA gene
expression deregulation in human breast cancer. Cancer Res 2005;65:7065 ^ 70.
22. He H, Jazdzewski K, Li W, et al. The role of microRNA genes in papillary thyroid carcinoma. Proc Natl
Acad Sci U S A 2005;102:19075 ^ 80.