Hepatocellular carcinoma (HCC) is one of the most common causes of cancer morbidity and mortality in Southeast Asia and China1, 2 and in the United States, where its incidence has increased by over 90% over the past 3 decades.3 Curative surgery offers the only hope for long-term survival for individuals diagnosed with HCC. However, by the time of clinical presentation, most patients are at the advanced metastatic stage, which rendered liver resection amenable to only a minority of patients. For the majority of patients, the median survival is ∼6 months.4 Chemotherapy is one of the mainstay treatments for patients with inoperable disease. Unfortunately, the response to chemotherapy has been poor; even with the most effective class of anthracycline-based therapies such as doxorubicin, the reported response rate has been low (<20%).5–7 Moreover, the administration of doxorubicin is often associated with the coexistence of drug resistance and high treatment-related cytotoxicity. The overall dismal outcome thus underscores the urgent need for reliable biomarkers in the diagnosis and prognosis of HCC and improved molecular targets for the development of more efficacious therapies.
In breast cancer, specific amplifications and overexpressions of the HER2/neu oncogene hold predictive value of disease outcome and represent an ideal target for anti-cancer therapy.8, 9 In cancer drug resistance, overexpression of transporters proteins, P-glycoprotein (P-gp)10 and multidrug resistance protein (MRP1),11 have long been implicated in the enhanced cellular drug efflux from the increased transporters activities. In HCC, few oncogenes of prognostic and therapeutic values have been described. The ability to expand our knowledge on proto-oncogenes involved in the malignant development and progression of HCC would undoubtedly improve our understandings on liver carcinogenesis. This information would also permit development of novel therapeutic options as current benefits of most molecular therapies such as Imatinib (tyrosine kinase inhibitor) and Gefitinib (anti-EGFR) aimed at attenuating the overexpressed oncogenic activities.
Regional chromosomal gain is a major mechanism in inducing elevated genes expression. In an effort to define vital cytogenetic events from extraneous alterations linked to tumor complexities, our group has previously conducted large-scale informatic analysis on comparative genomic hybridization (CGH) data that were derived from more than 150 HCC tumors.12 The model established from neural network algorithms suggested genomic gains on chromosomes 1q and 8q in the initial development of HCC, and gains of few chromosomes in association with advanced tumor stages and adverse disease outcome. The SORs of these causal events have been further defined to 1q21-q24, 6p12-p22, 7q21-q31, 8q22-q24, 17q21-q25, 19q13 and 20q12-q13.12–15 As studies on correlating CGH data with transcriptome array have shown that 40–60% of genes in amplified regions are indeed overexpressed,16, 17 in this study, we have attempted to define candidate overexpressed genes in the causal regional gains of HCC. We constructed a gene transcript map from a cohort of 22 HCC cell lines and early passages of cultures that presented gains or amplifications of aberrant loci in 40% to 72% of cases (9∼16 cell lines). A significant enrichment of highly expressed genes was suggested from the high-resolution mapping. Distinct expression of proto-oncogene, TOP2A at chr.17q21, further prompted our investigations for its potential value as biomarker in HCC by TMA analysis. The prognostic value of TOP2A in chemotherapy responsiveness and survival was also examined, in comparison to P-gp and MRP1, in prechemotherapy tumor biopsies of 148 HCC patients who had been entered into our Phase III prospective randomized study for doxorubicin-based therapy.7 The therapeutic value of antitopoisomerase therapy in vitro was also explored. Our study represents the first to highlight the clinical importance of TOP2A in HCC.
Eight HCC cell lines, HepG2, Hep3B, PLC/PRF/5, SNU387, SNU398, SNU423, SNU449 and SNU475, were obtained from the American Type Culture Collection (ATCC, Rockville, MD). Huh7 cell line was acquired from the Health Science Research Resources Bank (Osaka, Japan). HepG2, Hep3B and PLC/PRF/5 were cultured according to recommendations in DMEM medium containing 10% fetal bovine serum (Gibco BRL, Grand Island, NY), whereas Huh7, SNU387, SNU398, SNU423, SNU449 and SNU475 were maintained in complete RPMI medium (Gibco BRL, Grand Island, NY). Thirteen HCC cell lines HKCI-1, -2, -3, -4, -5, -6, -7, -8, -9, -10 and HKCI-C1, -C2, -C3 were established from our laboratory, eight of which have been previously reported.18–20 Using the same methodology described, we have newly established 5 additional cell lines (HKCI-6, -7, -8, -9 and -10). The HKCI series of cell lines were maintained in RPMI 1640 glutamax with HEPES buffer supplemented with 10% fetal bovine serum, 10 ng/ml selenium, 10 μg/ml transferrin and 10 μg/ml insulin. The early passages (passages 14–25) of the HKCI cell lines were employed in transcriptional mapping, totaling 22 specimens of homogenous HCC cells being studied. All cell lines have also been subjected to conventional CGH analysis20 to define the regional chromosomal aberrations present. All cultures were maintained in a humidified incubator at 37°C in an atmosphere of 5% CO2.
Tumorous liver and corresponding adjacent non-tumoral liver were collected from 30 patients who underwent curative surgery for HCC at the Prince of Wales Hospital, Hong Kong. Histological examination confirmed the diagnosis of HCC with 76.7% having coexisting cirrhosis. Patients were mostly chronic viral hepatitis B carriers (93.3%).
Randomized phase III clinical study
One hundred and forty-eight patients who entered our reported randomized Phase III clinical study on doxorubicin-based chemotherapy was studied.7 The administration of chemotherapeutic agents was as described.7 Before study entry, the diagnosis of HCC was histologicaly confirmed with tumor biopsy. Patients were separately consented to participate in the present correlative study, which utilized their biopsied samples. The protocol was approved by the Clinical Research Ethics Committee of the Chinese University of Hong Kong. The clinical and demographic characteristics at baseline of patients recruited are shown in Table I. Evaluation of response was conducted by radiological methods after the third and sixth cycles. Classification of response was based on the World Health Organization (WHO) criteria.21 Complete response was defined as no evidence of tumor by clinical and radiological assessments. Partial response was defined as a decrease of ≥50% in cross-perpendicular dimensions of the largest tumor nodule for at least 30 days without the appearance of new lesions or progression of other measurable or evaluable lesions. Stable disease was defined as any response less than a partial response for the largest tumor nodule or an increase in cross-perpendicular dimensions of <25% of any measurable or evaluable lesions, without the appearance of new lesions. Progressive disease was defined as an increase of ≥25% of any measurable or evaluable lesions or the appearance of new lesions.
Table I. Clinical and Demographic Characteristics at Baseline of 148 Patients with Unresectable Hepatocellular Carcinoma
No. of patients
AFP, alfa feto-protein; HBsAg, hepatitis B surface antigen.
Chemotherapy “nonresistant” group = partial response + stable disease; chemotherapy “resistant” group = progressive disease.
For the purpose of the present study, patients were categorized into 2 groups. Those whom did not obtain clinical benefit and developed progressive disease were classified as being “chemotherapy resistant” group. Those who obtained clinical benefit by achieving complete response, partial response and stable disease were classified as being “chemotherapy nonresistant” group. The primary endpoint was overall survival from date of randomization. The secondary endpoints were overall response and toxicity.
A gene transcript map from array profiling was constructed according to our method previously described.19 Briefly, reverse-transcribed RNA from cell lines and normal liver pool were fluorescent-labeled with Cy5-dCTP or Cy3-dCTP. Labeled cDNAs were hybridized onto cDNA arrays (Ontario Cancer Institute, Canada) containing ∼19,000 sequence-verified human genes. The average resolution throughout the genome was ∼100 to 200 kb. Hybridized signals captured by the ScanArray 5000 (Packard BioScience, UK) were analyzed by the GenePix Pro 4.0 (Axon, CA). Data manipulation was facilitated by the Normalization Suite v2.5 (http://www.utoronto.ca/cancyto/). Results from duplicate spots and dye swap experiments were averaged, and signal ratios integrated with physical map location of each cDNA clone. “Test to reference” ratios ≥2-fold were considered increased genes expression.
First strand cDNA was prepared from HCC tumors, adjacent non-tumoral livers and cell lines by random hexanucleotide (Applied Biosystems, Foster City, CA). cDNA preparation mixed with 0.2 μM primers, 4 mM MgCl2, 0.25 mM dNTPs mix and 0.025U AmpliTaq Gold® DNA polymerase (Applied Biosystem, Foster City, CA) was amplified at 94°C for 30 sec, 60°C for 10 sec and 72°C for 10 sec. Primer sequences for TOP2A amplification were forward, 5′-TTCTTGATATGCCCCTTTGG-3′ and reverse, 5′-GCTTCAACAGCCTCCAATTC-3′. The emission intensity was detected by the iCycler detection system (BioRad Laboratories, CA). The amount of TOP2A expression was determined by normalization to endogenous reference (18S RNA). Threshold cycles averaged from triplicate reactions were expressed as a relative ratio between cell lines and normal human liver controls, and between tumors and adjacent nontumoral liver. A value ≥2-fold was considered up-regulation.
Fluorescence in-situ hybridization
TOP2A copy number was studied by dual-color FISH according to method described.19 Human BAC clone RP11-513C18 (Invitrogen, CA) that embodied the TOP2A gene and reference clone RP11-110A23 (Invitrogen, CA) on chr1p31.3 were differentially labeled with biotin-dUTP and digoxigenin-dUTP, respectively. The selection of reference probe was based on our previous CGH study that suggested regional 1p31.3 rarely displayed imbalances in HCC.15 Hybridized probes were visualized by fluorescent-conjugated antibodies and scored for test to reference signal ratios in at least 50 interphase nuclei following nuclear counterstained in DAPI.
Formalin-fixed paraffin-embedded archive tissue of 172 paired HCC tumor and adjacent non-tumoral liver were arranged in tissue array blocks. Five-micrometer sections taken from each array block were stained haematoxylin & eosin. Primary antibody against TOP2A (Calbiochem, San Diego, CA) was applied at dilution of 1:60. The specificity of TOP2A antibody was confirmed by blocking analysis using purified human Topoisomerase 2α, 170 kDa (TopoGEN, Port Orange, FL). Chromogen development was performed using the universal HRP Multimer ultraview kit (Ventana medical system, Australia). The scoring of TOP2A expressions was classified as no staining (Grade 1), 1–9% of cells (Grade 2) and >10% of cells (Grade 3). The clinicopathological information of tumor pairs studied is shown in Table II.
Table II. Clinicopathological Parameters of HCC Cases Studied in Tissue Microarray
HBsAg, hepatitis B surface antigen.
Viral HBV infection
Grade of tumor differentiation
Immunohistochemical stainings of TOP2A, P-gp and MRP1
Serial 4 μm sections were obtained from 148 prechemotherapy tumor biopsies embedded in paraffin. Immunohistochemical staining of TOP2A was as described for TMA analysis. Expression of MRP1 (at 1:20 dilution; Alexis Biochemicals, Switzerland) was detected using avidin–biotin-complex method, whereas P-gp (at 1:80 dilution; Alexis Biochemicals, Switzerland) by the SuperPicture polymer detection. Following incubation with primary antibody, sections were treated with biotinylated rabbit anti-mouse immunoglobulins (at 1:200 dilution; DAKO, Carpinteria, CA) before the addition of streptavidin horseradish peroxidase conjugate (at 1:400 dilution; Zymed, San Francisco, CA). Chromogen signals were developed with diaminobenzidine-H2O2 substrate mixture. The percentage of positively stained tumor cells was scored as TMA, protein expressions were classified as Grade 1 (no staining), Grade 2 (1–9%) and Grade 3 (>10% of cells).
In-vitro cytotoxic assay
The cytotoxic effect of Etoposide (ABIC, Israel) as a single agent and in combination with Doxorubicin (Ebewe Pharma, Austria) in 3 HCC cell lines, HKCI-1, -2 and -3, were studied and evaluated by the MTT assay (Sigma, St. Louis, MO). Approximately 4 × 103 cells/well of each cell line were seeded 24 hr before addition of Etoposide. Five replicates were analyzed for each concentration ranging from 0 to 345 μM. Cell viability was measured at 48 hr by the colorimetric product developed at 570 nm. The IC50 and IC20 concentrations were determined by the linear regression analysis using the Graphpad Prism 3.0 software (San Diego, CA).
In the combinatory application of Etoposide with Doxorubicin, cells were treated with IC20 concentration of Etoposide for 4 hr before the addition of sequential dilutions of Doxorubicin ranging from 0 to 367 μM. After 44 hr, MTT assay was performed and the IC50 value determined was compared to treatment with Doxorubicin alone. Values reported were the mean of 3 independent experiments.
TOP2A mRNA expression in HCC tumors and paired adjacent non-tumoral livers were compared by the Student's paired t-test. Correlation of TOP2A mRNA levels with protein expressions was examined by the Spearman's rho test. The Pearson's χ2 test was used to examine the TOP2A protein scores with clinicopathological features including tumor stagings, histological differentiations, presence of macro- or microvascular invasions and patients' demographics. In the Phase III clinical study, overall survival was measured from the date of randomization to the date of death or last contact if patient was still alive. All data were censored on May 31, 2004. Immunohistochemical expressions of TOP2A, P-gp and MPR1 were studied for correlation with chemotherapy responsiveness and survival of patients. A stepwise logistic regression analysis was applied to examine the association of independent marker with response to chemotherapy. Overall survival was analyzed by the Kaplan-Meier method. A p-value <0.05 was considered to be statistically significant. All statistical analysis was performed using SPSS for Windows 10.0, SAS version 8.2 and Graphpad Prism 3.0.
Candidate Prot-oncogenes in causal regional gains
High-resolution array-based transcriptional mapping revealed a number of aberrantly overexpressed genes in the causative genomic regions of HCC. In 22 HCC cell lines studied, modulators or transducers of various signalling systems (14-3-3 zeta; IL11; GNA13; GRB2), transcriptional regulators (C-MYC; SP2, SPIB; SUMO2), protein stabilizers (heat shock protein 70; DNAJC5) and inducers of cell proliferation (CSE1L; HDGF) were suggested (Supp. Info. Table I). Of interest, several cell cycle related genes (TOP2A, CDC27 and TUBG1) were found to cluster on chr. 17q21 (Fig. 1). Also, few genes were found exclusively overexpressed in cases with copy number gains, e.g., the spindle assembler, AURKA, cytoskeletal component, SGCA, regulator of fibrinolysis, SERPINE1 and ABC transporter, CFTR. Other candidates such as the transcriptional regulator, ZNF229, export receptor, CSE1L and matrix adhesion molecule, ITGA3 were found to be up-regulated in both cell lines categories with and without regional gains.
TOP2A expressions in HCC
Distinct overexpression of TOP2A identified from transcriptional mapping has prompted our further investigations on its potential value as biomarker in HCC. Based on microarray, up-regulations of TOP2A were suggested in 14/22 cell lines, among which 9 cell lines displayed regional 17q21-q25 gain. Validative measurements of TOP2A mRNA transcripts by qRT-PCR suggested complete concordance with array findings in 14/14 cell lines (Fig. 2a). Increased expressions of TOP2A in 3 additional cell lines (HKCI-1, Hep3B and Huh7) were also suggested from qRT-PCR, albeit not indicated from array hybridization. The possible role of DNA copy gains in the induction of TOP2A expression was also investigated by FISH. Copy gains of TOP2A were confirmed in 9/9 cell lines with 17q gain (Fig. 2b).
By qRT-PCR, we further examined the TOP2A mRNA levels in HCC and adjacent non-tumoral liver, which is also often considered the premalignant lesion of HCC. A significant level of TOP2A mRNA was suggested in tumors (median 13.46, quartiles 3.14–28.06) relative to non-tumoral liver (median 0.27, 0.15–1.47) (n = 30; paired t-test p = 0.0018) (Fig. 3a). A corresponding translated TOP2A protein was also confirmed by immunohistochemical staining, where a positive correlation with mRNA levels was suggested (n = 23; r = 0.612, p = 0.002) (Fig. 3b).
Large-scale immunohistochemical analysis of TOP2A expression was performed on a Tissue Microarray that contained 172 early resectable HCC tumors and paired adjacent non-tumoral livers (Table II). In agreement with mRNA findings, distinct TOP2A protein stainings were suggested in tumors compared to adjacent non-tumoral livers (Fig. 4). In the estimation of cells positive for the TOP2A nuclear staining, 32% of cases showed absent of expression, whereas Grades 2 and 3 scorings were suggested in 50% and 18% of cases, respectively. Further correlative analysis of TOP2A protein levels with clinicopathological features suggest high TOP2A protein score (Grade 3) to be significantly associated with poorly differentiated histology (p < 0.001), microvascular invasion (p = 0.004) and an early age onset of HCC (≤40 yrs; p = 0.007) (Fig. 5).
Indicator of chemoresistance and survival
Of the 148 advance inoperable HCC patients recruited from the Phase III clinical study, 73 patients developed progressive disease during chemotherapy and were classified as the “chemotherapy resistant” group. There were 23 partial responders and 52 stable diseases; together they were classified as the “chemotherapy nonresistance” group. The overall median survival was 7.9 months (range 6.4–10.8). Patient characteristics are illustrated in Table I, which are similar to the reported Phase III study.7
Immunohistochemical stainings of 148 prechemotherapy biopsies indicated Grade 3 (>10% of cells) expressions of MRP1, P-gp and TOP2A in 65%, 54% and 30% of patients, respectively. Figure 6 shows examples of immunohistochemical results on MRP1, P-gp and TOP2A. Statistical correlations did not suggest expressions for MRP1 and P-gp in association with chemoresponses nor survival. However, a strong association for high TOP2A expressions with chemoresistance (p = 0.0289; odds ratio for resistance = 2.245, 95% CI 1.087–4.635) (Fig. 7) and shorter patient survivals (p < 0.0001) was demonstrated. Figure 8 shows the survival curves of studied population according to TOP2A, MRP1 and P-gp expressions.
Combinatory effect of TOP2A targeting and doxorubicin
Etoposide can interrupt the activity of DNA topoisomerase II by suppressing the enzyme mediated DNA cleavage. Doxorubicin, on the other hand, is a DNA intercalator that is currently the first-line cytotoxic agent for HCC patients. We examined the effect of Etoposide as a single agent and in combination with Doxorubicin on 3 HCC cell lines, HKCI-1, HKCI-2 and HKCI-3. These cell lines displayed TOP2A overexpressions by >10-fold compared to normal liver controls. A dosage-dependent inhibition on cell viability by Etoposide alone was suggested in all 3 cell lines, where the IC20 values determined ranged from 68 to 120 μM. When Etoposide at IC20 concentrations was introduced in conjunction with Doxorubicin, an enhanced combinatory effect was readily observed. A 10-fold reduction in the IC50 value of Doxorubicin was suggested in HKCI-1, whereas ∼3.5-fold decrease was suggested in both HKCI-2 and HKCI-3 (Fig. 9).
Gene expression profilings have successfully provided new insights into the molecular pathogenesis and classifications of human malignancies.22–24 Array-based transcriptional mapping further offers the feasibility to correlate chromosomal aberrations with gene expression data in identifying novel genes.25, 26 Here, we examined 7 regional gains that are of pivotal importance to the causation and maintenance of HCC for affected proto-oncogenes. Transcriptional mapping indicated a number of aberrantly expressed candidates in the causative loci. These included 14-3-3 zeta, PVT1 and C-MYC at 8q22-q24, S100A4 and HDGF on 1q21-q24, and STK6 on 20q12-q13. Of interest was cell cycle related genes on chr. 17q21 (TOP2A, CDC27 and TUBG1) were found to be concurrently up-regulated. This might have implications in the concerted actions of genes induced from the proximal 17q gain in the uncontrolled cell growth. In line with our previous bioinformatic study on CGH events, which suggested gains of 17q in the progression of HCC,12 studies on gastric cancer, prostate carcinoma and neuroblastoma have also long postulated on genes harboring on the proximal 17q12-q21 to play an important role in tumor progressions.27–29 In this respect, distinct overexpression of TOP2A identified at chr. 17q21.2 was further investigated for its potential clinical value in HCC.
The TOP2A gene encodes a DNA topoisomerase, an enzyme that controls and alters the topologic states of DNA during transcription. Despite a role in DNA metabolism, in-vitro studies have indicated that expression of TOP2A in tumors is less dependent on the proliferation state of malignant cells.30 Expression of TOP2A has also been demonstrated in breast, colon, ovarian and small cell lung cancers to be a valuable prognostic marker for tumor advancements, recurrences and predictor of poorer survival.31–34 In HCC, although common up-regulations of TOP2A have been suggested,35, 36 a systematic investigation of TOP2A as a biomarker and target in chemosensitization has not been described. In this study, we found a strong positive relationship between TOP2A mRNA expressions with translated protein levels. This finding has invariably facilitated our further large-scale evaluation on the prognostic significance of TOP2A by immunohistochemical analysis. By Tissue Microarray analysis, we found high protein expressions of TOP2A correlated with advance tumor histology, microvascular invasion and the likely indication of an early age onset. As vascular invasion frequently occurs in HCC and holds prognostic implications for tumor invasions and metastasis,37 our finding may be interpretation as TOP2A expressions correlate with a more aggressive tumor phenotype.
The emergence of drug resistance presents a major obstacle in cancer chemotherapy for HCC. With the accessibility of pretreatment biopsies from HCC patients who had undergone doxorubicin-containing chemotherapy, immunohistochemical staining of biomarkers in pretreatment tumour biopsies has allowed our correlative analysis with disease outcome. We assessed the protein expressions of TOP2A, P-gp and MRP1 in relation to survival of this patient population and responsiveness to chemotherapy. Overexpressions of both P-gp and MRP1 have been indicated as prognostic indicators in the induction of chemotherapy resistance, disease relapse and overall survival.38, 39 However, in this study, we were unable to establish a causative link between the increased protein efflux activities of P-gp and MRP1 with drug resistance and disease outcome. Instead, prognostic significance was indicated for TOP2A, where high protein score (Grade 3) correlated with nonresponsiveness to chemotherapy and early disease-related deaths. Using in-vitro experimental models, increased TOP2A expression has been shown to confer resistance to fractionated radiation and chemotherapeutic agents in small-cell lung cancer cells,40 and in the acquired doxorubicin resistance behavior of HCC cell lines.41 Our present study thus provides further support for the prognostic significance of TOP2A expression in HCC patients.
In addition to its role in regulating cellular replication, TOP2A is also a molecular target for several anti-cancer chemotherapeutic agents and a variety of mutations in this gene has been described in the development of drug resistance.42, 43 It is know that the activity of DNA topoisomerase II can be interrupted by enzyme binders (e.g., Etoposide) or DNA lesions (e.g., abasic sites) to produce topoisomerase II-mediated DNA damage. In-vitro studies have also demonstrated that the forced expressions of topoisomerase by ectopic transfection could enhance the therapeutic effect of Etoposide in drug-resistant breast and brain cancer cell lines.44, 45 In this study, we showed that Etoposide could sensitize HCC cells to doxorubicin-induced cell death. In HCC cell lines expressing increased TOP2A, Etoposide at low concentrations (IC20 doses) reduced the doxorubicin IC50 dosage by ∼3.5-fold to as much as 10-fold compared to doxorubicin alone. Our data suggested both enhanced and synergistic effects could be found with the combinatorial use of Etoposide and doxorubicin. In view of concerns over the toxicities related to the use of doxorubicin,7, 46 our finding might hold value in the further development of combinatory therapies for HCC patients.
In conclusions, the present study demonstrated the effectiveness in the integrated approach of chromosomal aberrations with genes expression mapping in underpinning candidate genes. Our study also highlights the clinical usefulness in the complementary use of TMA in the validation of molecular markers. Preliminary investigation on the relationship of TOP2A copy number and gene expression suggested a strong predisposing role of FISH-derived copy gains in gene up-regulations. Given that TOP2A expressions correlated with more aggressive HCC tumor phenotype and patient responsiveness to doxorubicin-based therapy, future development of FISH testing and/or immunohistochemical staining for TOP2A status may aid clinical practice in highlighting patients whom are likely to be resistant to conventional cytotoxic therapy. Furthermore, in the light of new clinical trials targeting molecular abnormalities in HCC,47 studies on effective combinatory therapy including reduced toxicities may hold benefit in the management of patients with unresectable hepatocellular carcinoma.
The authors would like to thank Prof. Jeremy Squire and Dr. Ben Beheshti for their technical support in microarray development and analytical software.