Sorafenib is the standard treatment for patients with advanced hepatocellular carcinoma (HCC). However, the median overall survival (OS) benefit is only ~3 months, and sufficient biomarkers predicting treatment response are not available. The aim of the present study was to evaluate miRNA expression patterns from HCC tissue biopsies as potential biomarkers in patients under sorafenib treatment. Nineteen patients with advanced HCC treated with sorafenib were included. RNA was extracted from formalin-fixed paraffin-embedded (FFPE) liver biopsies. miRNA expression profiling of 818 mature miRNAs was performed using GeneChip® miRNA Array 2.0 (Affymetrix). Global miRNA patterns were assessed using unsupervised hierarchical clustering analysis (UCA), and specific miRNAs with correlation with disease control rate (DCR) or good OS were evaluated by pairwise supervised analyses. UCA divided the patients into three distinct groups by their miRNA expression patterns. However, DCR or OS did not correlate with these sub-groups. We have identified several miRNAs that correlated with either DCR or OS (P<0.05). However, with correction for multiple testing, these results did not reach statistical significance in this small cohort. Global miRNA analysis from very low input RNA deriving from liver biopsies showed distinctive clustering of molecular sub-groups in patients with intermediate and advanced HCC. Clinical response including OS under sorafenib did not correlate with global miRNA expression patterns, but we have identified candidate miRNAs for the prediction of DCR and OS to be evaluated in prospective studies and larger patient cohorts.

CLINICAL PERSPECTIVES

  • Sorafenib is the standard of care in patients with advanced HCC. However, the median OS benefit is approximately only 3 months, and sufficient biomarkers predicting treatment response are not available. The aim of the present study was to evaluate miRNA expression patterns from HCC tissue biopsies as potential biomarkers in patients under sorafenib treatment.

  • We show the feasibility of global miRNA expression analyses from small FFPE liver biopsy specimens, and that the experimental protocol allows increased availability of tissue for future miRNA studies in HCC. Three different miRNA expression clusters were identified in patients with intermediate and advanced HCC in a Caucasian patient cohort, but were not linked to the sorafenib treatment response.

  • This study demonstrates that miRNA analyses are feasible from fine needle biopsy specimens and is the first to correlate the response under sorafenib therapy with miRNA expression signatures in HCC. The data do not support that miRNA profiles function as reliable biomarkers for sorafenib- response-prediction. Nevertheless, it might be worthwhile including the candidate miRNAs identified in the present study in biomarker protocols in ongoing or future studies on anti-angiogenic agents beyond sorafenib.

INTRODUCTION

Hepatocellular carcinoma (HCC) is a major complication of end-stage liver disease [14]. The worldwide annual incidence of HCC is estimated to be 600 000 cases. In Western countries and Japan, the incidence of hepatitis C virus (HCV)-related liver cirrhosis and liver cancer is rising due to the HCV epidemic [5,6]. A further increase is assumed, since obesity, diabetes mellitus and non-alcoholic steatohepatitis have been identified as risk factors of HCC, especially in Western countries [7]. Other important risk factors for HCC development are chronic hepatitis B virus (HBV) infection, haemochromatosis and alcohol-related liver cirrhosis [4,8].

According to the Barcelona Clinic Liver Cancer (BCLC) system, HCC is differentiated into five stages [3,9]. Curative treatment options are available in (very) early-stage HCC (BCLC-0, BCLC stage A) only, whereas treatment in BCLC stages B–D is non-curative and palliative, associated with median survival rates of less than 2 years [2,3]. A major step in HCC pathogenesis, in general, is the switch to arterial hypervascularization, which principally occurs in nodules exceeding a diameter of ~2 cm [10,11]. Anti-angiogenic treatment of HCC either by transarterial chemoembolization (TACE) or sorafenib, an oral multikinase inhibitor with activity against vascular endothelial growth factor receptor 2 (VEGFR2), platelet-derived growth factor receptor (PDGFR), c-Kit receptors, b-RAF and p38, is the current basic principle in HCC treatment in non-curable stages [12]. Sorafenib is currently the sole agent approved for HCC and has become the backbone in anti-angiogenic therapy in patients with HCC BCLC stage C and selected patients with HCC BCLC stage B [13].

In the pivotal study with sorafenib performed in a Western cohort, the objective response rate was less than 5%, and the disease control rate (DCR) was 43% in patients treated with sorafenib (n=299) [14]. Moreover, the median overall survival (OS) benefit was 2.8 months compared with best supportive care [14]. Comparable results were seen in a large Asian-Pacific population [15]. Hence, the main therapeutic effect of sorafenib is disease stabilization [14]. However, significant tumour size reduction and long-lasting response under sorafenib therapy have been reported in single patients [16,17].

Numerous drug-related events were observed within these trials as well as in clinical practice, for example hand–foot skin reaction, fatigue, diarrhoea and weight loss [14,15,18,19]. Reliable biomarkers predicting treatment response to sorafenib in HCC are not available yet, although an α-fetoprotein (AFP) decline has been suggested as a predictive surrogate marker for efficacy in patients with baseline AFP >20 ng/ml before initiation of sorafenib [20]. Dysregulation, either over- or under-expression, of small regulatory non-coding RNA molecules (miRNAs) has been reported in various malignant entities, and dysregulation of miRNA expression is assumed to play a major role in cancer pathogenesis [21,22]. Moreover, a classification of HCC sub-populations by distinct miRNA patterns has been reported previously [23].

The aim of the present study was, therefore, to evaluate the feasibility of global miRNA expression analyses from formalin-fixed paraffin-embedded (FFPE) liver biopsy specimens. Furthermore, we aimed at analysing whether miRNA expression patterns in advanced HCC are linked to treatment response to sorafenib.

MATERIALS AND METHODS

Patient selection and treatment/ethics statement

The present study is a retrospective analysis, which was approved by the local Ethics Committee of the University Hospital Frankfurt, Germany (Ethik-Kommission des Fachbereichs Medizin der Johann Wolfgang Goethe-Universität). Written informed consent was obtained from all patients. In total, 19 patients with mainly intermediate and advanced liver cancer were treated with sorafenib (target dose, 400 mg twice daily) in a tertiary German liver centre (Table 1). Patients with intermediate HCC received sorafenib as second-line treatment or had contraindications to other treatment options.

Table 1
Patient characteristics

Values are means±S.D., medians (minimum, maximum) or numbers (percentage). *At first diagnosis of HCC; at start of sorafenib. ECOG, Eastern Co-operative Oncology Group; MELD, Model for End-Stage Liver Disease.

ParameterValue
Age (years)* 64±14 
Gender (n 
 Female 4 (21%) 
 Male 15 (79%) 
Ancestry region (n 
 Europe 19 
Cause of liver cirrhosis (n 
 Viral (HBV or HCV) 5 (26%) 
 Non-viral 14 (74%) 
Body mass index (kg/m2)† 29.3±3.6 
ECOG performance status (n)†  
 0 15 (79%) 
 1 4 (21%) 
Child–Pugh class (n)†  
 A 18 (95%) 
 B 1 (5%) 
MELD score* 7 (6, 17) 
AFP serum level (ng/ml)† 17 (1.6, 79579) 
BCLC stage (n)†  
 A 1 (5%) 
 B 11 (58%) 
 C 7 (37%) 
Histological grading (G) (n)*  
 G1 2 (11%) 
 G2 15 (79%) 
 G3 1 (5%) 
 G4 1 (5%) 
Second-line therapy after sorafenib (n4 (21%) 
Third-line therapy after sorafenib (n2 (11%) 
ParameterValue
Age (years)* 64±14 
Gender (n 
 Female 4 (21%) 
 Male 15 (79%) 
Ancestry region (n 
 Europe 19 
Cause of liver cirrhosis (n 
 Viral (HBV or HCV) 5 (26%) 
 Non-viral 14 (74%) 
Body mass index (kg/m2)† 29.3±3.6 
ECOG performance status (n)†  
 0 15 (79%) 
 1 4 (21%) 
Child–Pugh class (n)†  
 A 18 (95%) 
 B 1 (5%) 
MELD score* 7 (6, 17) 
AFP serum level (ng/ml)† 17 (1.6, 79579) 
BCLC stage (n)†  
 A 1 (5%) 
 B 11 (58%) 
 C 7 (37%) 
Histological grading (G) (n)*  
 G1 2 (11%) 
 G2 15 (79%) 
 G3 1 (5%) 
 G4 1 (5%) 
Second-line therapy after sorafenib (n4 (21%) 
Third-line therapy after sorafenib (n2 (11%) 

Evaluation of treatment response

Survival was recorded in all patients and classified as good (longer than median survival of the cohort) or poor (shorter than median survival of the cohort). Furthermore, treatment response was assessed radiologically using either contrast-enhanced computed tomography (CT) or dynamic MRI according to the Response Evaluation Criteria In Solid Tumours (RECIST) [24]. DCR was defined as the percentage of patients who achieved complete response, partial response, or stable disease at first staging.

Liver biopsy and histological assessment of liver cancer

Biopsy under ultrasound guidance was performed with an 18 gauge cutting needle to allow accurate histological evaluation. Liver biopsy specimens were fixed in formalin and embedded in paraffin for routine staining and immunohistochemistry. Diagnosis of HCC was confirmed by an experienced pathologist. Grading was assessed according to the current TNM (tumour node metastasis) classification of malignant tumours. Immunohistochemical staining for glypican 3, glutamine synthetase and heat-shock protein 70 (HSP70) was used to support diagnosis and grading evaluation. All liver biopsies were micro-scopically reviewed by a pathologist, and tissue areas with a tumour cell content of more than 95% were macro-dissected and used for further analyses.

Isolation and expression profiling analysis of miRNA

Total RNA was extracted from tumour biopsy FFPE specimen using the RNeasy FFPE Kit (Qiagen) following the manufacturer's protocol. RNA quantification was performed using the Nanodrop 2000 spectrometer (Thermo Fisher Scientific) using 50 ng of total RNA for further analyses.

Material preparation for validation experiments

To test the performance of the procedure in low tissue amounts, technical and biological replicates were analysed. For technical validation, the array analysis was repeatedly performed (duplicate) from the same RNA extract. To evaluate biological variance, one HCC FFPE biopsy was separated into two fragments using a scalpel (blade no. 11), and RNA extraction as well as further analysis [microarray or quantitative real-time reverse transcription PCR (Q-RT-PCR)] were performed independently for the same patient.

Microarray analysis

Global miRNA expression profiling analysis of 818 mature miRNAs using GeneChipR miRNA Array v1.0 (Affymetrix) was performed following the manufacturer's protocol with the exception of the hybridization duration. Hybridization time was elongated from 18 to 40 h to compensate for low RNA input amount. The data discussed in the present study have been deposited in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus [25] and are accessible through GEO series accession number GSE56059 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE56059).

Real-time PCR miRNA expression analysis

Quantification of miRNA was performed by standardized TaqMan® MicroRNA Q-RT-PCR assays (Applied Biosystems) according to the manufacturer's protocol. Expression was analysed for three miRNAs (hsa-miR-21, hsa-miR-126 and hsa-miR-150) and one endogenous control (U6). Samples were analysed in triplicate, and ΔCt values were calculated using the endogenous control.

Bioinformatics and statistical analyses

Clinical and biochemical characteristics of patients are expressed as means±S.D. or medians and range as appropriate. Unless indicated otherwise, all tests were two tailed and P values < 0.05 were considered significant. Survival was assessed by Kaplan–Meier's analysis. Pearson's correlation or Spearman's test was applied as appropriate for calculation of correlations between two variables. Wilcoxon Mann–Whitney's U test or Wilcoxon-paired sample test was applied for comparisons of two groups with either unrelated or related samples respectively. Kruskal–Wallis's test or Friedman's test was used for group comparisons with more than two groups involved with either unrelated or related samples respectively.

Statistical analysis of miRNA profiles was performed using the statistical computing environment R. Additional software packages (geneplotter, gplots) were taken from the Bioconductor project. Replicate correlation was calculated using Pearson's correlation coefficient and depicted as a scatterplot. Unsupervised hierarchical clustering analysis (UCA) was performed for miRNAs with an S.D. ≥1.5 across all samples using the Manhattan distance method and the average linkage method. For the supervised analysis, an expression intensity and variance filter was used to reduce the dimension of microarray data. Data were filtered by an intensity filter (gene intensity >100 in at least 25% of the samples, if the group size is equal) and a variance filter (the interquartile range of log 2 intensities >0.5, if the group size is equal). After the filtering process P values were calculated with two sample t tests (variance=equal) to identify differentially expressed miRNAs between two groups. For multiple testing problems, a false discovery rate (FDR) was used [26]. Also fold change (FC) between the two groups was calculated for each gene. The lists of differentially expressed genes were then filtered for P<0.05 and FC > 1.5. DIANA miRPath v.2.0 was used as a computational predictive model to calculate potentially targeted genes and pathways using microT-CDS database (P value threshold 0.05, MicroT threshold 0.8, FDR correction and conservative statistics applied) [27]. Depicted pathways are derived from the KEGG database [25].

RESULTS

Efficacy of sorafenib therapy

Mean±S.E.M. and median OS were 1247±142 days and 1482 days respectively. First staging was performed at 102 (57–208) days (median, range) after start of sorafenib therapy. In patients with available radiological response assessment (n=15), median progression-free survival was 256 days (minimum 84, maximum 1824). Overall, seven out of 19 (37%) patients showed a good response, defined as longer than mean survival, compared with 12 out of 19 (63%) patients with poor response. The DCR was 15 out of 19 (79%) patients. Tumour biopsies were acquired at 57 (6–159) days (median, range) before start of the sorafenib therapy.

Technical validation of miRNA analysis

The use of small liver biopsy specimen for array analysis yielded very low input RNA (50–100 ng of total RNA). To test the performance of the miRNA analysis in low tissue amounts, technical and biological replicates were analysed using two complimentary methods. For technical validation the array analysis was repeatedly performed (duplicate) from the same RNA extract [Pearson's correlation coefficient (R)=0.915, Figure 1A]. To evaluate biological variance within the tumour, one HCC FFPE biopsy was separated in two fragments and further analysis (microarray or Q-RT-PCR) was performed independently for the same patient. The microarray analysis showed a close correlation for biological replicates (R=0.69, Figure 1B). In addition, Q-RT-PCR was performed as a complimentary validation method. Expression of three randomly chosen miRNAs (miR-21, miR-126 and miR-150) correlated well between biological replicates, and quantitative expression differences were comparable with semi-quantitative assessment by microarray experiments (Figure 1C).

Technical validation of replicates

Figure 1
Technical validation of replicates

(A) Array analysis was repeatedly performed (duplicate) to evaluate assay reproducibility. A scatterplot is shown depicting correlation of miRNA expression between two replicates [measurement 1 (y-axis) and measurement 2 (x-axis)] of the same RNA extract [Pearson's correlation coefficient (R)=0.915). (B) One HCC biopsy was separated into two and RNA extraction as well as microarray analysis were performed independently for the same patient (R1 compared with R1*). A scatterplot is shown depicting correlation of miRNA expression between two replicates [R1* (y-axis) and R1 (x-axis)] of the same individual [Pearson's correlation coefficient (R)=0.692]. (C) Comparison of biological replicates using Q-RT-PCR. miRNA expression was analysed for three randomly chosen miRNAs (hsa-miR-21, hsa-miR-126 and hsa-miR-150) and one endogenous control (U6). Samples were analysed in triplicate, and ΔCt values were calculated using the endogenous control.

Figure 1
Technical validation of replicates

(A) Array analysis was repeatedly performed (duplicate) to evaluate assay reproducibility. A scatterplot is shown depicting correlation of miRNA expression between two replicates [measurement 1 (y-axis) and measurement 2 (x-axis)] of the same RNA extract [Pearson's correlation coefficient (R)=0.915). (B) One HCC biopsy was separated into two and RNA extraction as well as microarray analysis were performed independently for the same patient (R1 compared with R1*). A scatterplot is shown depicting correlation of miRNA expression between two replicates [R1* (y-axis) and R1 (x-axis)] of the same individual [Pearson's correlation coefficient (R)=0.692]. (C) Comparison of biological replicates using Q-RT-PCR. miRNA expression was analysed for three randomly chosen miRNAs (hsa-miR-21, hsa-miR-126 and hsa-miR-150) and one endogenous control (U6). Samples were analysed in triplicate, and ΔCt values were calculated using the endogenous control.

Inter-individual correlation of miRNA expression profiles in advanced HCC

After global filtering, a panel of 84 miRNAs was used to perform UCA. UCA divided the patients into three distinct sub-groups by their respective global miRNA expression profile (Figure 2). The cluster features two equally large groups (group A, n=8, and group B, n=8) and a small cluster of three patients (group C). The small group C shows a general down-regulation of nearly all of the featured miRNAs. In particular the let-7 group of miRNAs is strongly down-regulated with up to 45-fold down-regulation compared with the other groups (see Supplementary Table S1 at http://www.clinsci.org/cs/128/cs1280029add.htm). In the direct comparison of groups A and B, a relatively low general expression of miRNAs in group B with particularly strong down-regulation of some known tumour suppressor miRNAs (e.g. miR-29a, miR-126 and miR-26a) was observed, whereas group A featured a relative down-regulation of miR-99/100, miR-1826 and miR-768-5p. The HCC miRNA cluster groups did not correlate with HCC aetiology, AFP level, histological grading, immunohistochemical markers of differentiation, radiological tumour stage (results not shown) or overall patient survival (P=non-significant for all; Figure 3). DIANA-miRPath v2.0 enrichment analysis of predicted miRNA target genes was used to identify pathways potentially influenced by the miRNA profiles for the respective cluster groups. For each group, miRNAs with differential expression and highest fold-change compared with the respective two other groups were selected as basis for the analysis (FC > 2×S.D.; P<10−4, see Supplementary Table S2 at http://www.clinsci.org/cs/128/cs1280029add.htm). All three groups were enriched for well-known oncogenic signalling pathways. Group A was highly enriched for miRNAs with a role in the phosphoinositide 3-kinase (PI3K)/Akt pathway (P=2.1×10−13). Group B was predicted to be associated with ErbB and the mitogen-activated protein kinase (MAPK) signalling cascades respectively (P=9.2×10−16 and P=3.0×10−13). miRNAs differentially expressed in group C were enriched for miRNA target genes involved in the transforming growth factor β (TGFβ) pathway (P=4.2×10−7). PI3K/Akt, ErbB and TGFβ pathways with potentially affected genes are shown in Supplementary Figure S1 (http://www.clinsci.org/cs/128/cs1280029add.htm).

UCA of miRNA expression profiles generated from 19 HCC tumour samples using GeneChipR miRNA Array (version 1.0)

Figure 2
UCA of miRNA expression profiles generated from 19 HCC tumour samples using GeneChipR miRNA Array (version 1.0)

UCA was performed for miRNAs with an S.D. ≥1.5 across all samples using the Manhattan distance method and the average linkage method. High expression is given in red and low expression in green.

Figure 2
UCA of miRNA expression profiles generated from 19 HCC tumour samples using GeneChipR miRNA Array (version 1.0)

UCA was performed for miRNAs with an S.D. ≥1.5 across all samples using the Manhattan distance method and the average linkage method. High expression is given in red and low expression in green.

Kaplan–Meier's survival analysis of 19 patients since the start of sorafenib treatment

Figure 3
Kaplan–Meier's survival analysis of 19 patients since the start of sorafenib treatment

Patients were stratified for either of three HCC miRNA clusters groups deriving from the UCA (see Figure 2; dashed: group A, n=8; dash-dotted: group B, n=8; joined: group C, n=3). OS was not associated with groups of miRNA expression patterns.

Figure 3
Kaplan–Meier's survival analysis of 19 patients since the start of sorafenib treatment

Patients were stratified for either of three HCC miRNA clusters groups deriving from the UCA (see Figure 2; dashed: group A, n=8; dash-dotted: group B, n=8; joined: group C, n=3). OS was not associated with groups of miRNA expression patterns.

Association of single miRNAs with patient characteristics

Several specific miRNAs showed correlations with clinical parameters of the included HCC patients in the supervised expression analysis. HCC arising in patients with non-viral liver disease (HBV or HCV) showed the up-regulation of miR-494, miR-1826, miR-923 and miR-768-5p, whereas AFP elevation (>400 ng/ml) was associated with the up-regulation of miR-342-3p, miR-130a, miR-150, miR-574-3p and miR-15b and the down-regulation of miR-30c (see Supplementary Table S3 at http://www.clinsci.org/cs/128/cs1280029add.htm; FC≥3 and P<0.05 for all). Moreover, tumour expression levels of specific miRNAs were identified to correlate with the response to therapy and survival (Table 2). Down-regulation of miR-99b was associated with DCR (P<0.05), whereas up-regulation of miR-140-3p, miR-100, miR-125 and miR-22 was associated with a good OS (P<0.05 for all). Statistical significance of the above-mentioned associations, however, was not maintained after correction for multiple testing (FDR>0.05).

Table 2
Specific miRNAs associated with DCR or OS
(A) miRNA associated with DCR
miRNAFCP valueFDR
hsa-miR-99b −2.24 0.074 0.75 
(B) miRNAs associated with OS
miRNA FC P value FDR 
hsa-miR-140-3p 2.5 0.015 0.38 
hsa-miR-100 3.5 0.023 0.38 
hsa-miR-125b 2.9 0.026 0.38 
hsa-miR-22 2.8 0.046 0.38 
(A) miRNA associated with DCR
miRNAFCP valueFDR
hsa-miR-99b −2.24 0.074 0.75 
(B) miRNAs associated with OS
miRNA FC P value FDR 
hsa-miR-140-3p 2.5 0.015 0.38 
hsa-miR-100 3.5 0.023 0.38 
hsa-miR-125b 2.9 0.026 0.38 
hsa-miR-22 2.8 0.046 0.38 

DISCUSSION

Worldwide, HCC is the sixth most common cancer [1]. Most patients are diagnosed with intermediate or advanced stages, and thus overall therapeutic options are limited and prognosis is still poor [9]. Moreover, the clinical course and molecular alterations show a high variability [28,29]. Currently, no clinically reliable biomarker is known to predict the individual treatment response, mainly to sorafenib, in patients with intermediate or advanced HCC. miRNAs are involved in regulation of gene expression and thereby affect virtually all cell functions [30]. Consequently, miRNAs are promising putative biomarkers in HCC as their up-regulation and down-regulation have been associated with pathogenesis and clinical outcome in HCC [31–35]. For instance, miR-199a/b-3p has been found to be consistently down-regulated compared with tumour-adjacent non-tumourous tissue in 255 out of 294 samples (87%) and low miR-199a/b-3p expression correlated with poor OS [36]. Moreover, a classification of HCC based on miRNA expression levels has been proposed previously [23]. Nevertheless, it is yet unknown, whether miRNA patterns could be used as a biomarker to distinguish between HCC patients with better or worse course in advanced stages (BCLC stages B/C). In the present study, we have investigated whether miRNAs obtained from percutaneous liver biopsy specimen could be used to classify intermediate and advanced stage HCC and, moreover, predict treatment response to sorafenib.

Toffanin et al. [23] and most others [33,3740] performing global miRNA expression analyses have used surgically resected HCC specimens. In daily clinical practice and according to current guidelines [11,41], surgical treatment–either resection or liver transplantation–is suitable for early HCC stages only. Thus, studies investigating miRNA expression profiles have derived mainly from early liver cancer specimen. One group previously used fresh-tissue fine-needle HCC biopsies to perform an array-based miRNA analysis screening for only a small fragment of currently known miRNAs [42]. A custom microarray was used containing probes for 180 mature human miRNAs, and five down-regulated and three up-regulated miRNAs were discovered compared with non-tumourous liver tissue. Previously, other groups investigated miRNA expression from hepatic FFPE material using real-time PCR from different samples or demonstrating the feasibility of miRNA array experiments from tissue sections [43,44]. Also, FFPE tissue from unguided biopsies from non-cirrhotic non-tumorous liver has been successfully used for array-based miRNA analyses [45]. The novelty in the present study lies in the use of FFPE material from very small ultrasound-guided HCC biopsies of patients under sorafenib therapy, which, to our knowledge, has not been published to date.

Liver biopsies can be performed in most patients with intermediate or advanced HCC, and archived FFPE biopsy tissue is readily available. Possible disadvantages are fixation-associated decrease in RNA quality and the much smaller amount of tissue available for analysis. With small modifications to the hybridization routine of the microarray platform, we were able to achieve a high-quality quantification of global miRNA expression from as low as 50 ng of input RNA deriving from FFPE tissue. Technical validation showed very good reproducibility, and UCA was comparable with previously published miRNA UCA patterns [23]. As expected the correlation coefficient of biological replicates was lower compared with technical replicates, probably due to intra-tumoural heterogeneity. However, the reproducibility was still very high, allowing for good quality miRNA screening from FFPE fine-needle biopsy specimen.

In the present study, RNA extraction was performed from HCC areas only (tumour cell content >95%). Thus, observed distinct miRNA cluster profiles are based on molecular differences in HCC itself rather than on differences of cirrhotic tissue. The data suggest that intermediate and advanced HCC can be classified into three groups by tumour HCC miRNA expression clusters, which is a similar finding to previous studies that included mostly early HCC stages [23]. One cluster (group C) featured a generally low expression of all miRNAs with marked down-regulation of let-7g, which has known tumour suppressive properties and was found to be down-regulated in HCC tissue by others [46,47]. Nevertheless, down-regulation of let-7g was not associated with the clinical or histopathological tumour phenotype in the present study. Group B also showed a general down-regulation of miRNAs when compared with group A, including suppression of miR-26a, miR-29a and miR-126, all miRNAs whose down-regulation has been associated previously with HCC pathogenesis [4850]. miR-99a and miR-100 are down-regulated in group A. Also, suppression of those two miRNAs has been implicated in HCC pathogenesis and prognosis before [51,52]. Furthermore, pathway analysis showed an enrichment of genes potentially influenced by detected miRNAs within the PI3K/Akt, ErbB and TGFβ pathways for groups A, B and C respectively. Those pathways have been associated with cancer development in general and were frequently found previously to be dysregulated in HCC [5357]. However, the current pathways analyses have to be treated with caution as they are based on an in silico approach only and are not yet supported by mRNA gene expression data.

We have found several miRNAs to be differentially expressed between viral and non-viral aetiologies of HCC underlying liver disease. In particular miR-494, which was up-regulated in HCC of non-viral origin, has been recently implicated in HCC carcinogenesis [58]. Among miRNAs that were associated with AFP elevation, miR-15b was found to be associated with HCC prognosis in previous studies [59,60].

Next, we have investigated whether miRNA profiles or dysregulation of distinct miRNA species may be associated with the treatment response to sorafenib. No association between the three identified tumour miRNA expression patterns and survival in sorafenib-treated patients was observed. Furthermore, we have analysed putative correlations of individual miRNAs with DCR or OS in patients with sorafenib-treated HCC. Indeed, several miRNAs were associated with either DCR or OS. However, this was not statistically significant after correction for multiple testing, which–in part–could be explained by the limited number of patients included in the present study. miRNAs with previously described correlation with the prognosis in surgically resected HCC, such as miR-199a/b, did not correlate with DCR or OS in the present study [6163]. A possible explanation for this might be the fact that miR-199a/b is only predictive for surgical therapy of less advanced HCC, as our cohort contained only advanced non-resectable HCC.

AUTHOR CONTRIBUTION

Jan Peveling-Oberhag designed and performed the research, revised and analysed the clinical data and wrote the paper; Claudia Döring performed the bioinformatics and statistical analysis; Sylvia Hartmann provided advice, and collected and characterized the histological samples; Natalie Fillmann performed the statistical analysis; Angelika Mertens performed the research and collected histological samples; Albrecht Piiper provided advice and assisted in the correction of the paper; Eva Herrmann performed the statistical analysis, provided advice and revised the paper; Martin-Leo Hansmann designed the research, collected and characterized histological samples and revised the paper; Stefan Zeuzem designed the research, provided advice and revised the paper; Jörg Trojan designed the research, provided advice and revised the paper; Martin-Walter Welker designed and performed the research, revised and analysed clinical data, and wrote the paper.

We thank Ekaterini Hadzoglou and Sabine Albrecht for excellent technical assistance.

FUNDING

This work was supported by a ‘Patenschaftsmodell’ grant from the Medical Faculty of the J.W. Goethe-University Hospital and the German Federal Ministry of Education and Research (BMBF, BioChancePlus [grant number #0316043E]).

Abbreviations

     
  • AFP

    α-fetoprotein

  •  
  • BCLC

    Barcelona Clinic Liver Cancer

  •  
  • DCR

    disease control rate

  •  
  • FC

    fold change

  •  
  • FDR

    false discovery rate

  •  
  • FFPE

    formalin-fixed paraffin-embedded

  •  
  • HBV

    hepatitis B virus

  •  
  • HCC

    hepatocellular carcinoma

  •  
  • HCV

    hepatitis C virus

  •  
  • OS

    overall survival

  •  
  • PI3K

    phosphoinositide 3-kinase

  •  
  • Q-RT-PCR

    quantitative real-time reverse transcription PCR

  •  
  • TGFβ

    transforming growth factor β

  •  
  • UCA

    unsupervised hierarchical clustering analysis

References

References
1
Parkin
 
D. M.
Bray
 
F.
Ferlay
 
J.
Pisani
 
P.
 
Global cancer statistics, 2002
CA Cancer J. Clin.
2005
, vol. 
55
 (pg. 
74
-
108
)
[PubMed]
2
Cabibbo
 
G.
Enea
 
M.
Attanasio
 
M.
Bruix
 
J.
Craxi
 
A.
Camma
 
C.
 
A meta-analysis of survival rates of untreated patients in randomized clinical trials of hepatocellular carcinoma
Hepatology
2010
, vol. 
51
 (pg. 
1274
-
1283
)
[PubMed]
3
Llovet
 
J. M.
Burroughs
 
A.
Bruix
 
J.
 
Hepatocellular carcinoma
Lancet
2003
, vol. 
362
 (pg. 
1907
-
1917
)
[PubMed]
4
El Serag
 
H. B.
Rudolph
 
K. L.
 
Hepatocellular carcinoma: epidemiology and molecular carcinogenesis
Gastroenterology
2007
, vol. 
132
 (pg. 
2557
-
2576
)
[PubMed]
5
El Serag
 
H. B.
Mason
 
A. C.
 
Rising incidence of hepatocellular carcinoma in the United States
N. Engl. J. Med.
1999
, vol. 
340
 (pg. 
745
-
750
)
[PubMed]
6
Taura
 
N.
Yatsuhashi
 
H.
Hamasaki
 
K.
Nakao
 
K.
Daikoku
 
M.
Ueki
 
T.
Yano
 
K.
Matsumoto
 
T.
Ishibashi
 
H.
Eguchi
 
K.
 
Increasing hepatitis C virus-associated hepatocellular carcinoma mortality and aging: long term trends in Japan
Hepatol. Res.
2006
, vol. 
34
 (pg. 
130
-
134
)
[PubMed]
7
Ekstedt
 
M.
Franzen
 
L. E.
Mathiesen
 
U. L.
Thorelius
 
L.
Holmqvist
 
M.
Bodemar
 
G.
Kechagias
 
S.
 
Long-term follow-up of patients with NAFLD and elevated liver enzymes
Hepatology
2006
, vol. 
44
 (pg. 
865
-
873
)
[PubMed]
8
Fattovich
 
G.
Stroffolini
 
T.
Zagni
 
I.
Donato
 
F.
 
Hepatocellular carcinoma in cirrhosis: incidence and risk factors
Gastroenterology
2004
, vol. 
127
 (pg. 
S35
-
S50
)
[PubMed]
9
Llovet
 
J. M.
Bru
 
C.
Bruix
 
J.
 
Prognosis of hepatocellular carcinoma: the BCLC staging classification
Semin. Liver Dis.
1999
, vol. 
19
 (pg. 
329
-
338
)
[PubMed]
10
Forner
 
A.
Vilana
 
R.
Ayuso
 
C.
Bianchi
 
L.
Sole
 
M.
Ayuso
 
J. R.
Boix
 
L.
Sala
 
M.
Varela
 
M.
Llovet
 
J. M.
, et al 
Diagnosis of hepatic nodules 20 mm or smaller in cirrhosis: prospective validation of the noninvasive diagnostic criteria for hepatocellular carcinoma
Hepatology
2008
, vol. 
47
 (pg. 
97
-
104
)
[PubMed]
11
Bruix
 
J.
Sherman
 
M.
 
Management of hepatocellular carcinoma: an update
Hepatology
2011
, vol. 
53
 (pg. 
1020
-
1022
)
[PubMed]
12
Wilhelm
 
S.
Carter
 
C.
Lynch
 
M.
Lowinger
 
T.
Dumas
 
J.
Smith
 
R. A.
Schwartz
 
B.
Simantov
 
R.
Kelley
 
S.
 
Discovery and development of sorafenib: a multikinase inhibitor for treating cancer
Nat. Rev. Drug Discov.
2006
, vol. 
5
 (pg. 
835
-
844
)
[PubMed]
13
Welker
 
M. W.
Trojan
 
J.
 
Anti-angiogenesis in hepatocellular carcinoma treatment: current evidence and future perspectives
World J. Gastroenterol.
2011
, vol. 
17
 (pg. 
3075
-
3081
)
[PubMed]
14
Llovet
 
J. M.
Ricci
 
S.
Mazzaferro
 
V.
Hilgard
 
P.
Gane
 
E.
Blanc
 
J. F.
de Oliveira
 
A. C.
Santoro
 
A.
Raoul
 
J. L.
Forner
 
A.
, et al 
Sorafenib in advanced hepatocellular carcinoma
N. Engl. J. Med.
2008
, vol. 
359
 (pg. 
378
-
390
)
[PubMed]
15
Cheng
 
A. L.
Kang
 
Y. K.
Chen
 
Z.
Tsao
 
C. J.
Qin
 
S.
Kim
 
J. S.
Luo
 
R.
Feng
 
J.
Ye
 
S.
Yang
 
T. S.
, et al 
Efficacy and safety of sorafenib in patients in the Asia-Pacific region with advanced hepatocellular carcinoma: a phase III randomised, double-blind, placebo-controlled trial
Lancet Oncol.
2009
, vol. 
10
 (pg. 
25
-
34
)
[PubMed]
16
Abbadessa
 
G.
Rimassa
 
L.
Pressiani
 
T.
Carrillo-Infante
 
C.
Cucchi
 
E.
Santoro
 
A.
 
Optimized management of advanced hepatocellular carcinoma: four long-lasting responses to sorafenib
World J. Gastroenterol.
2011
, vol. 
17
 (pg. 
2450
-
2453
)
[PubMed]
17
Gerardi
 
A. M.
Stoppino
 
L. P.
Liso
 
A.
Macarini
 
L.
Landriscina
 
M.
 
Rapid long-lasting biochemical and radiological response to sorafenib in a case of advanced hepatocellular carcinoma
Oncol. Lett.
2013
, vol. 
5
 (pg. 
975
-
977
)
[PubMed]
18
Worns
 
M. A.
Weinmann
 
A.
Pfingst
 
K.
Schulte-Sasse
 
C.
Messow
 
C. M.
Schulze-Bergkamen
 
H.
Teufel
 
A.
Schuchmann
 
M.
Kanzler
 
S.
Duber
 
C.
, et al 
Safety and efficacy of sorafenib in patients with advanced hepatocellular carcinoma in consideration of concomitant stage of liver cirrhosis
J. Clin. Gastroenterol.
2009
, vol. 
43
 (pg. 
489
-
495
)
[PubMed]
19
Welker
 
M. W.
Lubomierski
 
N.
Gog
 
C.
Herrmann
 
E.
Engels
 
K.
Vogl
 
T.
Bechstein
 
W.
Zeuzem
 
S.
Trojan
 
J.
 
Use of sorafenib in advanced hepatocellular carcinoma–a case study of a German Liver Center
Onkologie
2010
, vol. 
33
 (pg. 
86
-
86
)
20
Personeni
 
N.
Bozzarelli
 
S.
Pressiani
 
T.
Rimassa
 
L.
Tronconi
 
M. C.
Sclafani
 
F.
Carnaghi
 
C.
Pedicini
 
V.
Giordano
 
L.
Santoro
 
A.
 
Usefulness of alpha-fetoprotein response in patients treated with sorafenib for advanced hepatocellular carcinoma
J. Hepatol.
2012
, vol. 
57
 (pg. 
101
-
107
)
[PubMed]
21
Nana-Sinkam
 
S. P.
Croce
 
C. M.
 
MicroRNA dysregulation in cancer: opportunities for the development of microRNA-based drugs
IDrugs
2010
, vol. 
13
 (pg. 
843
-
846
)
[PubMed]
22
Sotiropoulou
 
G.
Pampalakis
 
G.
Lianidou
 
E.
Mourelatos
 
Z.
 
Emerging roles of microRNAs as molecular switches in the integrated circuit of the cancer cell
RNA
2009
, vol. 
15
 (pg. 
1443
-
1461
)
[PubMed]
23
Toffanin
 
S.
Hoshida
 
Y.
Lachenmayer
 
A.
Villanueva
 
A.
Cabellos
 
L.
Minguez
 
B.
Savic
 
R.
Ward
 
S. C.
Thung
 
S.
Chiang
 
D. Y.
, et al 
MicroRNA-based classification of hepatocellular carcinoma and oncogenic role of miR-517a
Gastroenterology
2011
, vol. 
140
 (pg. 
1618
-
1628
)
[PubMed]
24
Goldberg
 
S. N.
Charboneau
 
J. W.
Dodd
 
G. D.
Dupuy
 
D. E.
Gervais
 
D. A.
Gillams
 
A. R.
Kane
 
R. A.
Lee
 
F. T.
Livraghi
 
T.
McGahan
 
J. P.
, et al 
Image-guided tumor ablation: proposal for standardization of terms and reporting criteria
Radiology
2003
, vol. 
228
 (pg. 
335
-
345
)
[PubMed]
25
Ogata
 
H.
Goto
 
S.
Sato
 
K.
Fujibuchi
 
W.
Bono
 
H.
Kanehisa
 
M.
 
KEGG: Kyoto encyclopedia of genes and genomes
Nucleic Acids Res.
1999
, vol. 
27
 (pg. 
29
-
34
)
[PubMed]
26
Benjamini
 
Y.
Hochberg
 
Y.
 
Controlling the false discovery rate: a practical and powerful approach to multiple testing
J.R. Statist. Soc. B
1995
, vol. 
57
 (pg. 
289
-
300
)
27
Vlachos
 
I. S.
Kostoulas
 
N.
Vergoulis
 
T.
Georgakilas
 
G.
Reczko
 
M.
Maragkakis
 
M.
Paraskevopoulou
 
M. D.
Prionidis
 
K.
Dalamagas
 
T.
Hatzigeorgiou
 
A. G.
 
DIANA miRPath v.2.0: investigating the combinatorial effect of microRNAs in pathways
Nucleic Acids Res.
2012
, vol. 
40
 (pg. 
W498
-
W504
)
[PubMed]
28
Villanueva
 
A.
Hoshida
 
Y.
Battiston
 
C.
Tovar
 
V.
Sia
 
D.
Alsinet
 
C.
Cornella
 
H.
Liberzon
 
A.
Kobayashi
 
M.
Kumada
 
H.
, et al 
Combining clinical, pathology, and gene expression data to predict recurrence of hepatocellular carcinoma
Gastroenterology
2011
, vol. 
140
 (pg. 
1501
-
1512
)
[PubMed]
29
Zucnan-Rossi
 
J.
Laurent-Puig
 
P.
 
Genetic diversity of hepatocellular carcinomas and its potential impact on targeted therapies
Pharmacogenomics
2007
, vol. 
8
 (pg. 
997
-
1003
)
[PubMed]
30
Ambros
 
V.
 
The functions of animal microRNAs
Nature
2004
, vol. 
431
 (pg. 
350
-
355
)
[PubMed]
31
Wong
 
C. C.
Wong
 
C. M.
Tung
 
E. K.
Au
 
S. L.
Lee
 
J. M.
Poon
 
R. T.
Man
 
K.
Ng
 
I. O.
 
The microRNA miR-139 suppresses metastasis and progression of hepatocellular carcinoma by down-regulating Rho-kinase 2
Gastroenterology
2011
, vol. 
140
 (pg. 
322
-
331
)
[PubMed]
32
Xiong
 
Y.
Fang
 
J. H.
Yun
 
J. P.
Yang
 
J.
Zhang
 
Y.
Jia
 
W. H.
Zhuang
 
S. M.
 
Effects of microRNA-29 on apoptosis, tumorigenicity, and prognosis of hepatocellular carcinoma
Hepatology
2010
, vol. 
51
 (pg. 
836
-
845
)
[PubMed]
33
Wei
 
R.
Huang
 
G. L.
Zhang
 
M. Y.
Li
 
B. K.
Zhang
 
H. Z.
Shi
 
M.
Chen
 
X. Q.
Huang
 
L.
Zhou
 
Q. M.
Jia
 
W. H.
, et al 
Clinical significance and prognostic value of microRNA expression signatures in hepatocellular carcinoma
Clin. Cancer Res.
2013
, vol. 
19
 (pg. 
4780
-
4791
)
[PubMed]
34
Xia
 
H.
Ooi
 
L. L.
Hui
 
K. M.
 
MicroRNA-216a/217-induced epithelial-mesenchymal transition targets PTEN and SMAD7 to promote drug resistance and recurrence of liver cancer
Hepatology
2013
, vol. 
58
 (pg. 
629
-
641
)
[PubMed]
35
Zhu
 
K.
Pan
 
Q.
Zhang
 
X.
Kong
 
L. Q.
Fan
 
J.
Dai
 
Z.
Wang
 
L.
Yang
 
X. R.
Hu
 
J.
Wan
 
J. L.
, et al 
MiR-146a enhances angiogenic activity of endothelial cells in hepatocellular carcinoma by promoting PDGFRA expression
Carcinogenesis
2013
, vol. 
34
 (pg. 
2071
-
2079
)
[PubMed]
36
Hou
 
J.
Lin
 
L.
Zhou
 
W.
Wang
 
Z.
Ding
 
G.
Dong
 
Q.
Qin
 
L.
Wu
 
X.
Zheng
 
Y.
Yang
 
Y.
, et al 
Identification of miRNomes in human liver and hepatocellular carcinoma reveals miR-199a/b-3p as therapeutic target for hepatocellular carcinoma
Cancer Cell
2011
, vol. 
19
 (pg. 
232
-
243
)
[PubMed]
37
Jiang
 
J.
Gusev
 
Y.
Aderca
 
I.
Mettler
 
T. A.
Nagorney
 
D. M.
Brackett
 
D. J.
Roberts
 
L. R.
Schmittgen
 
T. D.
 
Association of MicroRNA expression in hepatocellular carcinomas with hepatitis infection, cirrhosis, and patient survival
Clin. Cancer Res.
2008
, vol. 
14
 (pg. 
419
-
427
)
[PubMed]
38
Connolly
 
E.
Melegari
 
M.
Landgraf
 
P.
Tchaikovskaya
 
T.
Tennant
 
B. C.
Slagle
 
B. L.
Rogler
 
L. E.
Zavolan
 
M.
Tuschl
 
T.
Rogler
 
C. E.
 
Elevated expression of the miR-17-92 polycistron and miR-21 in hepadnavirus-associated hepatocellular carcinoma contributes to the malignant phenotype
Am. J. Pathol.
2008
, vol. 
173
 (pg. 
856
-
864
)
[PubMed]
39
Gramantieri
 
L.
Ferracin
 
M.
Fornari
 
F.
Veronese
 
A.
Sabbioni
 
S.
Liu
 
C. G.
Calin
 
G. A.
Giovannini
 
C.
Ferrazzi
 
E.
Grazi
 
G. L.
, et al 
Cyclin G1 is a target of miR-122a, a microRNA frequently down-regulated in human hepatocellular carcinoma
Cancer Res.
2007
, vol. 
67
 (pg. 
6092
-
6099
)
[PubMed]
40
Huang
 
X. H.
Wang
 
Q.
Chen
 
J. S.
Fu
 
X. H.
Chen
 
X. L.
Chen
 
L. Z.
Li
 
W.
Bi
 
J.
Zhang
 
L. J.
Fu
 
Q.
, et al 
Bead-based microarray analysis of microRNA expression in hepatocellular carcinoma: miR-338 is downregulated
Hepatol. Res.
2009
, vol. 
39
 (pg. 
786
-
794
)
[PubMed]
41
Bruix
 
J.
Sherman
 
M.
Llovet
 
J. M.
Beaugrand
 
M.
Lencioni
 
R.
Burroughs
 
A. K.
Christensen
 
E.
Pagliaro
 
L.
Colombo
 
M.
Rodes
 
J.
 
Clinical management of hepatocellular carcinoma. Conclusions of the Barcelona-2000 EASL conference. European Association for the Study of the Liver
J. Hepatol.
2001
, vol. 
35
 (pg. 
421
-
430
)
[PubMed]
42
Murakami
 
Y.
Yasuda
 
T.
Saigo
 
K.
Urashima
 
T.
Toyoda
 
H.
Okanoue
 
T.
Shimotohno
 
K.
 
Comprehensive analysis of microRNA expression patterns in hepatocellular carcinoma and non-tumorous tissues
Oncogene
2006
, vol. 
25
 (pg. 
2537
-
2545
)
[PubMed]
43
Doleshal
 
M.
Magotra
 
A. A.
Choudhury
 
B.
Cannon
 
B. D.
Labourier
 
E.
Szafranska
 
A. E.
 
Evaluation and validation of total RNA extraction methods for microRNA expression analyses in formalin-fixed, paraffin-embedded tissues
J. Mol. Diagn.
2008
, vol. 
10
 (pg. 
203
-
211
)
[PubMed]
44
Liu
 
A.
Xu
 
X.
 
MicroRNA isolation from formalin-fixed, paraffin-embedded tissues
Methods Mol. Biol.
2011
, vol. 
724
 (pg. 
259
-
267
)
[PubMed]
45
Zhang
 
X.
Chen
 
C.
Wu
 
M.
Chen
 
L.
Zhang
 
J.
Zhang
 
X.
Zhang
 
Z.
Wu
 
J.
Wang
 
J.
Chen
 
X.
, et al 
Plasma microRNA profile as a predictor of early virological response to interferon treatment in chronic hepatitis B patients
Antivir. Ther.
2012
, vol. 
17
 (pg. 
1243
-
1253
)
[PubMed]
46
Gramantieri
 
L.
Ferracin
 
M.
Fornari
 
F.
Veronese
 
A.
Sabbioni
 
S.
Liu
 
C. G.
Calin
 
G. A.
Giovannini
 
C.
Ferrazzi
 
E.
Grazi
 
G. L.
, et al 
Cyclin G1 is a target of miR-122a, a microRNA frequently down-regulated in human hepatocellular carcinoma
Cancer Res.
2007
, vol. 
67
 (pg. 
6092
-
6099
)
[PubMed]
47
Lan
 
F. F.
Wang
 
H.
Chen
 
Y. C.
Chan
 
C. Y.
Ng
 
S. S.
Li
 
K.
Xie
 
D.
He
 
M. L.
Lin
 
M. C.
Kung
 
H. F.
 
Hsa-let-7g inhibits proliferation of hepatocellular carcinoma cells by downregulation of c-Myc and upregulation of p16(INK4A)
Int. J. Cancer
2011
, vol. 
128
 (pg. 
319
-
331
)
[PubMed]
48
Felli
 
N.
Felicetti
 
F.
Lustri
 
A. M.
Errico
 
M. C.
Bottero
 
L.
Cannistraci
 
A.
De Feo
 
A.
Petrini
 
M.
Pedini
 
F.
Biffoni
 
M.
, et al 
miR-126&126* restored expressions play a tumor suppressor role by directly regulating ADAM9 and MMP7 in melanoma
PloS One
2013
, vol. 
8
 pg. 
e56824
 
[PubMed]
49
Kota
 
J.
Chivukula
 
R. R.
O’Donnell
 
K. A.
Wentzel
 
E. A.
Montgomery
 
C. L.
Hwang
 
H. W.
Chang
 
T. C.
Vivekanandan
 
P.
Torbenson
 
M.
Clark
 
K. R.
, et al 
Therapeutic microRNA delivery suppresses tumorigenesis in a murine liver cancer model
Cell
2009
, vol. 
137
 (pg. 
1005
-
1017
)
[PubMed]
50
Braconi
 
C.
Kogure
 
T.
Valeri
 
N.
Huang
 
N.
Nuovo
 
G.
Costinean
 
S.
Negrini
 
M.
Miotto
 
E.
Croce
 
C. M.
Patel
 
T.
 
microRNA-29 can regulate expression of the long non-coding RNA gene MEG3 in hepatocellular cancer
Oncogene
2011
, vol. 
30
 (pg. 
4750
-
4756
)
[PubMed]
51
Li
 
D.
Liu
 
X.
Lin
 
L.
Hou
 
J.
Li
 
N.
Wang
 
C.
Wang
 
P.
Zhang
 
Q.
Zhang
 
P.
Zhou
 
W.
, et al 
MicroRNA-99a inhibits hepatocellular carcinoma growth and correlates with prognosis of patients with hepatocellular carcinoma
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
36677
-
36685
)
[PubMed]
52
Petrelli
 
A.
Perra
 
A.
Schernhuber
 
K.
Cargnelutti
 
M.
Salvi
 
A.
Migliore
 
C.
Ghiso
 
E.
Benetti
 
A.
Barlati
 
S.
Ledda-Columbano
 
G. M.
, et al 
Sequential analysis of multistage hepatocarcinogenesis reveals that miR-100 and PLK1 dysregulation is an early event maintained along tumor progression
Oncogene
2012
, vol. 
31
 (pg. 
4517
-
4526
)
[PubMed]
53
Bissell
 
D. M.
Roulot
 
D.
George
 
J.
 
Transforming growth factor beta and the liver
Hepatology
2001
, vol. 
34
 (pg. 
859
-
867
)
[PubMed]
54
Breuhahn
 
K.
Longerich
 
T.
Schirmacher
 
P.
 
Dysregulation of growth factor signaling in human hepatocellular carcinoma
Oncogene
2006
, vol. 
25
 (pg. 
3787
-
3800
)
[PubMed]
55
Grabinski
 
N.
Ewald
 
F.
Hofmann
 
B. T.
Staufer
 
K.
Schumacher
 
U.
Nashan
 
B.
Jucker
 
M.
 
Combined targeting of AKT and mTOR synergistically inhibits proliferation of hepatocellular carcinoma cells
Mol. Cancer
2012
, vol. 
11
 pg. 
85
 
[PubMed]
56
Ito
 
Y.
Takeda
 
T.
Sakon
 
M.
Tsujimoto
 
M.
Higashiyama
 
S.
Noda
 
K.
Miyoshi
 
E.
Monden
 
M.
Matsuura
 
N.
 
Expression and clinical significance of erb-B receptor family in hepatocellular carcinoma
Br. J. Cancer
2001
, vol. 
84
 (pg. 
1377
-
1383
)
[PubMed]
57
Vivanco
 
I.
Sawyers
 
C. L.
 
The phosphatidylinositol 3-kinase AKT pathway in human cancer
Nat. Rev. Cancer
2002
, vol. 
2
 (pg. 
489
-
501
)
[PubMed]
58
Lim
 
L.
Balakrishnan
 
A.
Huskey
 
N.
Jones
 
K. D.
Jodari
 
M.
Ng
 
R.
Song
 
G.
Riordan
 
J.
Anderton
 
B.
Cheung
 
S. T.
, et al 
MicroRNA-494 within an oncogenic microRNA megacluster regulates G1/S transition in liver tumorigenesis through suppression of mutated in colorectal cancer
Hepatology
2014
, vol. 
59
 (pg. 
202
-
215
)
[PubMed]
59
Chung
 
G. E.
Yoon
 
J. H.
Myung
 
S. J.
Lee
 
J. H.
Lee
 
S. H.
Lee
 
S. M.
Kim
 
S. J.
Hwang
 
S. Y.
Lee
 
H. S.
Kim
 
C. Y.
 
High expression of microRNA-15b predicts a low risk of tumor recurrence following curative resection of hepatocellular carcinoma
Oncol. Rep.
2010
, vol. 
23
 (pg. 
113
-
119
)
[PubMed]
60
Huang
 
Y. H.
Lin
 
K. H.
Chen
 
H. C.
Chang
 
M. L.
Hsu
 
C. W.
Lai
 
M. W.
Chen
 
T. C.
Lee
 
W. C.
Tseng
 
Y. H.
Yeh
 
C. T.
 
Identification of postoperative prognostic microRNA predictors in hepatocellular carcinoma
PLoS One
2012
, vol. 
7
 pg. 
e37188
 
[PubMed]
61
Henry
 
J. C.
Park
 
J. K.
Jiang
 
J.
Kim
 
J. H.
Nagorney
 
D. M.
Roberts
 
L. R.
Banerjee
 
S.
Schmittgen
 
T. D.
 
miR-199a-3p targets CD44 and reduces proliferation of CD44 positive hepatocellular carcinoma cell lines
Biochem. Biophys. Res. Commun.
2010
, vol. 
403
 (pg. 
120
-
125
)
[PubMed]
62
Fornari
 
F.
Milazzo
 
M.
Chieco
 
P.
Negrini
 
M.
Calin
 
G. A.
Grazi
 
G. L.
Pollutri
 
D.
Croce
 
C. M.
Bolondi
 
L.
Gramantieri
 
L.
 
MiR-199a-3p regulates mTOR and c-Met to influence the doxorubicin sensitivity of human hepatocarcinoma cells
Cancer Res.
2010
, vol. 
70
 (pg. 
5184
-
5193
)
[PubMed]
63
Wang
 
C.
Song
 
B.
Song
 
W.
Liu
 
J.
Sun
 
A.
Wu
 
D.
Yu
 
H.
Lian
 
J.
Chen
 
L.
Han
 
J.
 
Underexpressed microRNA-199b-5p targets hypoxia-inducible factor-1alpha in hepatocellular carcinoma and predicts prognosis of hepatocellular carcinoma patients
J. Gastroenterol. Hepatol.
2011
, vol. 
26
 (pg. 
1630
-
1637
)
[PubMed]

Author notes

1

These authors contributed equally to this work.

Supplementary data