Topic Highlight Open Access
Copyright ©2008 The WJG Press and Baishideng. All rights reserved.
World J Gastroenterol. Dec 7, 2008; 14(45): 6915-6923
Published online Dec 7, 2008. doi: 10.3748/wjg.14.6915
Mouse models in liver cancer research: A review of current literature
Martijn WH Leenders, Maarten W Nijkamp, Inne HM Borel Rinkes
Martijn WH Leenders, Maarten W Nijkamp, Inne HM Borel Rinkes, Department of Surgery, University Medical Center Utrecht, Utrecht GA 3508, The Netherlands
Author contributions: Leenders MWH and Nijkamp MW contributed equally to this work; Leenders MWH and Nijkamp MW performed the literature search and wrote the article; Borel Rinkes IHM is chief investigator, guided the research work and revised the article.
Correspondence to: Dr. Inne HM Borel Rinkes, MD, PhD, Professor, Department of Surgery, University Medical Center Utrecht, PO box 85500, Utrecht GA 3508, The Netherlands.
Telephone: +31-88-7558074 Fax: +31-30-2541944
Received: May 26, 2008
Revised: August 7, 2008
Accepted: August 14, 2008
Published online: December 7, 2008


Primary liver cancer remains one of the most lethal malignancies worldwide. Due to differences in prevalence of etiological factors the incidence of primary liver cancer varies among the world, with a peak in East-Asia. As this disease is still lethal in most of the cases, research has to be done to improve our understanding of the disease, offering insights for possible treatment options. For this purpose, animal models are widely used, especially mouse models. In this review, we describe the different types of mouse models used in liver cancer research, with emphasis on genetically engineered mice used in this field. We focus on hepatocellular carcinoma (HCC), as this is by far the most common type of primary liver cancer, accounting for 70%-85% of cases.

Key Words: Primary liver cancer, Hepatocellular carcinoma, Mouse model, Genetically engineered mice

Table 1 Transgenic mouse models for HCC.
TransgenePromoterMouse strainPercentage HCCsReference
c-mycAlbC57BL/6 × CBA/J65% in males at 20 mo40
TGF-αMTCD150% in males > 12 mo77, 78
c-myc/TGF-αAlb, MTC57BL/6 × CBA/J × CD1100% in males at 8 mo40
SV40 T-AgATIIIC57BL/6 × DBA2100% at 8 mo36
E2F-1AlbC57BL/6 × CBA/J33%-60% at 12 mo79, 80
c-myc/E2F-1AlbC57BL/6 × CBA/J100% at 9 mo80

The American Cancer Society has estimated that, in 2007, there were over 700 000 new cases of primary liver cancer worldwide. It is the fifth most common malignancy in men and the eighth in women. Liver cancer is among the most lethal cancers (five-year survival rates under 11%), which makes it the third most frequent cause of cancer death in men and the sixth in women[1]. Liver cancer consists of several histologically different primary hepatic malignancies, such as cholangiocarcinoma, hepatoblastoma and haemangiosarcoma, but hepatocellular carcinoma (HCC) is by far the most common type, accounting for 70%-85% of cases[1,2].

Cirrhosis (due to for instance hemochromatosis), chronic hepatitis B and C viral infections, chronic alcohol consumption, aflatoxin-B1 intake (from contaminated food) are the most important of the well-defined risk factors for HCC. Variations in the prevalence of these etiological factors mirror the geographical distribution of the incidence of HCC. The worldwide (age-adjusted) incidence per 100 000 persons is 14.9/5.5 (men/women), varying from 2.6/1.3 in Northern Europe to 35.4/12.6 in East Asia[2,3].

Animal models for human HCC can be helpful to our understanding of the (molecular) mechanisms underlying the pathogenesis of HCC. The laboratory mouse remains one of the best models to study cancer in vivo due to various features, such as the small size, the similarities to humans and the entirely sequenced genome and the similarities to humans[4].

For instance, the risk of HCC in males is approximately 2-5 times greater than in females[5,6]. This gender difference is seen in mice as well. Recently, Naugler et al[5] attributed this disparity to a higher serum interleukin-6 (IL-6) concentration in male than in female mice after administration of the chemical hepatocarcinogen diethylnitrosamine. IL-6 is secreted by Kupffer cells in response to, for example, necrotic hepatocytes. They demonstrated that estrogen inhibits the secretion of IL-6.

Since the spontaneous incidence of liver tumors in the most frequently used strains of mice is < 4%, mouse models have been developed to induce HCC-formation[7,8].

None of the currently available mouse models meet all criteria of the ideal animal model, which include biologic, genetic, etiologic and therapeutic criteria[9]. Therefore, the most appropriate model for a particular experimental question should be used to answer the specific research question.

This review highlights the currently used mouse models for HCC, with emphasis on genetically engineered models.


An etiological role for external agents in contributing to human HCC, also referred to as hepatocarcinogens, has been established by primarily epidemiological studies[9]. Examples of known human hepatocarcinogens include aflatoxins, ethanol and combined oral contraceptives[10]. Despite the fact that some of these chemicals are carcinogenic to the mouse liver as well, this is not true for all human hepatocarcinogens. For example, cirrhosis and liver cancer have not been observed in mice subjected to ethanol solely[10-12]. At the same time, hardly any of the mouse hepatocarcinogens have shown to be carcinogenic for humans (aflatoxin B1 and oral contraceptives belong to the exceptions). This discrepancy in response to carcinogens can probably be explained by species differences[9-11]. Nonetheless, carcinogen-induced mouse models are still frequently used for HCC research.

The (hepato) carcinogens are subdivided into two classes, namely genotoxic and non-genotoxic (or epigenetic) carcinogens. The genotoxic carcinogens can presumably cause cancer by forming DNA adducts, which lead to genetic changes of the target cell. These changes can direct normal cells to a preneoplastic state (initiation). The non-genotoxic carcinogens do not modify DNA structure, but generally stimulate the preneoplastic or initiated cells to evolve into a malignant neoplasm by controlling cell proliferation, apoptosis and cell differentiation[10,11,13]. Because of the high incidence of altered (or preneoplastic) hepatocytes in certain mouse strains (especially C3H and B6C3F1 mice, the latter corresponding to the progeny of C3H mice coupled to C57BL/6 mice), epigenetic carcinogens alone can be used to induce HCC-formation in these mice[8,11].

Many chemicals have been shown to induce HCCs in the mouse liver[10,14]. To date, the most frequently used hepatocarcinogens in mice are diethylnitrosamine (DEN) and phenobarbital (PB), although the carcinogenic effect of PB is controversial[8]. Furthermore, attention has been paid to the hepatocarcinogenic effect of peroxisome proliferators such as clofibrate and the experimental Wy-14.643[13].

The single most frequently used chemical for induction of HCCs in mice is DEN, a genotoxic carcinogen. DEN is typically administered to mice between 12 and 15 d of age by a single intraperitoneal injection (5 μg/g body weight). Using this protocol, originally described by Vesselinovitch and Mihailovich, 100% of B6C3F1 male mice developed HCCs, on average, 44 wk after intraperitoneal injection of DEN[15-17]. Less frequently used protocols for DEN-administration include intraperitoneal injection of a higher dose of DEN (for example 80-90 μg/g body weight) to older mice (4-5 wk of age) and intraperitoneal DEN-injection 36 h after a partial hepatectomy (mice 5-6 wk of age). These protocols are less efficient in producing HCCs than those previously mentioned[8]. DEN-induced mouse models of HCC are predominantly used to study the molecular mechanisms of (chemical) hepatocarcinogenesis. Furthermore, the influence of (trans) genes and chemicals that might prevent HCC development can be evaluated by means of this model.

In the background of carcinogen-induced mouse models for HCC, identifying chemicals that might be carcinogenic to humans is the major application of laboratory mice. For this purpose, chemicals are either administered to newborn mice in order to determine genotoxicity, or compounds are administered for longer periods (usually 2 years) to assess epigenetic carcinogenity[6,14]. However, because of the significant inconsistencies between mouse and human carcinogens, extrapolating the results of these mouse studies to the human situation remains difficult[18].

Carcinogen-induced mouse models for HCC are useful for establishing a relationship between carcinogen exposure and specific genetic changes[19]. However, the influences of sex, age and genetic background of the mice on the predictability of HCC-development remain disadvantages of these models.


Implantation models are among the most widely used models to accomplish HCC formation in mice, because of their suitability in studies for preclinical evaluation of anticancer agents. In implantation models that are currently used to induce HCC formation in mice, HCC cell lines or tumor tissue fragments are implanted in recipient mice. Here, we will give an impression of the approaches that can be used to obtain HCCs in these mice.

The earliest implantation model is the syngeneic transplantable tumor model, in which a HCC cell line or tissue fragment is implanted in mice of the strain from which the implant originates. To date, this model is less frequently used, because of the discovery that human tumor cell lines and tissue fragments can be implanted into immunodeficient mice. Nevertheless, this model is still required when anticancer agents are tested that work by activating the immune system (immune therapy)[20-22]. The unpredictability of the effectiveness of studied anticancer agents in human HCC and the small number of available murine cell lines are the most important disadvantages of this implantation model[20].

As mentioned, nowadays, primary human HCC cell lines or tissue fragments are implanted into immunocompromised mice (xenograft models).

The most widely used mouse for this approach is the nude mouse (nu -/-). These mice are athymic, hairless and have a deficiency of T lymphocytes as well as an impaired T and B cell function[23]. Next to nude mice, severe combined immunodeficient (SCID) mice are frequently used in xenograft models of HCC. These mice have a deficiency in number and function of both T and B lymphocytes[24]. Since the establishment of the first human HCC cell line in 1963[25], many human HCC cell lines have been described, of which the HepG2, the Hep3B, the SMMC-7721 and the HuH7 cell lines are the most commonly used.

In allograft models, murine HCC cell lines or tumor fragments are implanted in (not necessarily syngeneic) mice. The HCC grafts have been derived from spontaneously occurring HCCs in mice, from carcinogen-induced tumors or from genetically engineered mice (GEM), which is discussed in the last part of this review.

The above mentioned xenograft or allograft HCC cell suspension or tumor tissue fragments can be implanted into recipient mice, either at ectopic or at orthotopic sites. Ectopic implantation of tumor cells or fragments usually occurs subcutaneously. Orthotopic implantation can be accomplished by subserosal injection of HCC cells or by surgical orthotopical implantation (SOI) of tumor fragments. These fragments, approximately 1 mm3 in size, are derived from surgical specimens of human HCCs or from pieces of subcutaneously grown HCC cells, either from human or mouse origin.

Xenografts of human HCCs growing subcutaneously in mice are predominantly used in the preclinical evaluation of anticancer agents. The rapid formation of tumors, the minimal labor that is required, the relative inexpensiveness and the ability to measure tumor size non-invasively are the main advantages of the ectopic, subcutaneous, application of grafts from human HCCs[21,22]. However, many research groups have described the importance of the microenvironment on the biological behaviour of malignant cells. For example, many tumor cell lines do not spontaneously metastasize when they are subcutaneously implanted, while they do metastasize when they are orthotopically implanted. Hence, the interaction of organ-specific factors (such as fibroblasts, endothelial cells and inflammatory cells) with tumor cells, is important for HCC development[21,26]. For this reason, therapeutic results obtained by an ectopic implantation model of a HCC graft must always be verified in orthotopic models.

These orthotopic implantation models mimic human HCCs in a better way with respect to tumor morphology, microenvironment, metastatic potential and the response to anticancer agents[27,28]. Furthermore, processes involved in local invasion, like angiogenesis, can be examined in their normal microenvironment[20]. Nonetheless, disadvantages of orthotopic implants of HCC xenografts include a more difficult surgical implantation procedure and more expensive procedures. Furthermore, tumor growth and response cannot be determined as easily as in ectopic transplantation models.

Despite the fact that (ectopic and orthotopic) xenograft implantation models are among the most widely used models for preclinical evaluation of anticancer agents, it has been demonstrated excessively that these models have a poor predictive value for the anti-tumor effects in patients. This can possibly be explained by the fact that the injected tumor cells are often cultured for a long period. Due to selection pressures in culture, these tumor cells have no longer maintained the original molecular characteristics and the heterogeneity of the patients’ tumor[4,20,28,29]. By implanting a tumor fragment of a patient, the morphological and molecular characteristics of a patient’s tumor are preserved better. However, establishing a parallel in vitro cell line from a patient’s tumor is often very difficult.

The abovementioned features of implantation models are but a few arguments why implantation models are not ideal for the preclinical evaluation of anticancer agents. However, because of the suitability of these models, implantation models are still frequently used for this purpose.


The introduction of transgenic mouse models in the early 1980’s made it possible to study the molecular features of human malignancies in vivo[30,31]. Since then, much progress has been made in techniques of producing GEM[4].

In studying the molecular mechanisms involved in hepatocarcinogenesis, GEM are particularly used to explore the role of a specific gene and to explore the interaction of different genes (e.g. oncogenes and tumor-suppressor genes) in the development of HCC. GEM is also suitable to investigate the role of specific genes in combination with a liver-specific carcinogen.

As is the case in other types of cancer, genetic alterations in various cellular pathways (including pathways involved in growth, apoptosis, proliferation and angiogenesis) are needed for the development of HCC. Although the exact genetic events in hepatocarcinogenesis are not entirely clear, there is evidence that the p53, Rb and Wnt/β-catenin pathways are involved[32,33]. Several transgenic mouse lineages that are nowadays used to induce formation of HCCs are transgenic in one of these pathways (Table 1). The most commonly used models will be discussed here.

Since the late 1980’s, transgenic SV40 T-Ag (Simian Virus 40 T-antigen) mice have been studied extensively. The genome of the simian virus 40 (a DNA tumor virus) encodes two oncogenic proteins, the large and small T antigen (T-Ag and tAg, respectively, herein together referred to as T-Ag). After infection, large T-Ag can cause malignant transformation of the host cell primarily by inactivating the tumor-suppressor genes p53 and Rb[34,35].

Research groups have reported the production of transgenic mice expressing SV40 T-Ag directed to the liver by the promoter/enhancer antithrombin-III (ATIII)[36], albumin (Alb)[37] and α-1-antitrypsin (AAT)[38]. For example, Dubois et al[36] produced transgenic mice by putting the SV40 T-Ag under the control of the human ATIII promoter. In mouse lineages that expressed the highest level of the transgene, by the age of 8 mo, 100% of mice had developed HCCs and 10% had developed lung metastases.

Another commonly used transgenic mouse model was described by Murakami et al[39]. They generated double transgenic mice overexpressing c-myc and TGF-α in the liver (Alb-c-myc/MT-TGF-α mice) by crossing Alb/c-myc mice (transgenic mice overexpressing c-myc, directed by the albumin promoter) with MT/TGF-α mice (transgenic mice overexpressing TGF-α, directed by the metallothionein 1 promoter). Santoni-Rugiu et al[40] demonstrated that these mice developed HCCs substantially earlier and at a higher rate than single transgenic mice, overexpressing either c-myc or TGF-α. Within 8 mo after birth, 100% of male and 30% of female Alb-c-myc/MT-TGF-α mice had developed HCCs.

Although these conventional transgenic mouse models have been very useful to study the role of particular genes in hepatocarcinogenesis and to study the multistep nature of HCC development, one limitation of these models is the fact that the transgene is expressed in all hepatocytes, including the tumor microenvironment. Furthermore, the mutations are already present during embryogenesis and thus, might activate compensatory (molecular) pathways[26]. To overcome these limitations, mouse models have recently been developed in which the genetic alterations are induced in a tissue-specific and time-controlled fashion (conditional mouse models).

For instance, Lewis et al[41] used a retroviral transduction strategy to deliver oncogenes to hepatocytes in situ. They made use of the fact that mice do not express the TVA receptor, which is the receptor for the avian leukosis sarcoma virus subgroup A (ALSV-A). Lewis et al[41] generated TVA transgenic mice, in which TVA was specifically expressed within the liver. Delivery of ALSV-A viruses encoding PyMT (mouse polyoma virus middle T antigen, an oncogene) to these mice at the age of 2-3 d, subsequently led to tumor formation by the age of 4-6 mo (in 17 of 26 mice). They also exposed TVA transgenic mice that were deficient for p53 to PyMT-bearing ALSV-A viruses. Interestingly, the tumor incidence in these mice was not increased, but 6 of 16 p53 null mice that had developed HCCs, showed lung metastases (in contrast with 1 of 17 p53 wild-type mice). Consequently, this mouse model might be of value as a metastatic HCC model. Moreover, this model can be easily used to study the effect of other oncogenes in hepatocarcinogenesis, through the delivery of other oncogene-bearing ALSV-A viruses to TVA transgenic mice.

In addition, Lou et al[42] created mice with a regulated expression of liver-specific SV40 T-Ag. The SV40 T-Ag in these mice is preceded by a stop signal flanked by loxP sites. Hence, the SV40 T-Ag is expressed upon Cre-mediated excision, either by adenoviral expression of Cre recombinase or by administration of tamoxifen to mice that are transgenic for a liver-specific tamoxifen-inducible Cre. HCCs were observed in mice 5 mo after administration of adenoviral Cre recombinase or tamoxifen.

Several research groups employed alternative recombinase-mediated conditional gene-mutation strategies[32,33,43]. Colnot et al generated a mouse strain in which exon 14 of both Apc (adenomatous polyposis coli) alleles were flanked by loxP sites. The Apc alleles become invalidated (leading to β-catenin signaling) upon liver-targeted expression of Cre recombinase. Of these mice, 67% develop HCCs 8-9 mo after Cre recombinase administration[32,33].

Promising results have been published with these and other conditional mouse models to induce HCC-formation. Nonetheless, to date, these models are mainly used to study the effect of genetic alterations (mutation, deletion, or overexpression of a certain gene) on hepatocarcinogenesis and not to induce HCCs.


More than 80% of HCCs in humans are attributable to infection with either hepatitis B virus (HBV) or hepatitis C virus (HCV) or both[44]. HBV- and HCV-related HCC are characteristically preceded by liver cirrhosis, though this is not always the case[45]. It may take more than 20 years for HCC to develop in HBV or HCV infected persons. For this reason, hepatocarcinogenesis due to viral hepatitis probably requires multiple steps of genetic alterations.

Finding the molecular mechanisms that drive these multiple steps by using cell-culture and non-genetic animal models is difficult. Therefore, in the past decades various animal models for investigation of viral hepatitis were developed. One problem in establishing such a model is that HBV and HCV require human hepatocytes to induce hepatitis, due to the stringent human tropism of these viruses[46,47].

In HBV research the finding of HBV-related viruses, e.g. the woodchuck hepatitis virus (WHV) and the ground squirrel hepatitis virus (GSHV), has provided opportunities for in vivo studies[46,48]. Another approach for studying hepatitis B and C infection is the use of immunocompromised mice or rats. Recently, several animal models have been developed in which human hepatocytes or human liver tissue are transplanted into these animals. The transplanted hepatocytes in these animals can be infected with HBV or HCV in vivo or ex vivo. Alternatively, an already intrinsically infected specimen is transplanted[46,47]. These models are promising for the evaluation of therapeutics and prophylactics against hepatitis due to HBV or HCV, but are not useful to study HBV- or HCV-associated HCC. For that purpose, transgenic mice expressing HBV or HCV proteins represent a better model. In this section, the most frequently used transgenic mouse models for studying HBV- and HCV-associated HCC will be discussed.

By means of these models, two pathways have been proposed that might participate in the hepatocarcinogenic effect of chronic viral hepatitis. First, it is considered that chronic inflammation of the liver, continuous cell death and subsequent chronic hepatocyte regeneration due to viral hepatitis might increase the incidence of genetic alterations[48-53]. The second pathway encompasses a direct oncogenic effect of HBV or HCV on the infected hepatocyte. In the case of HBV (a DNA virus), this carcinogenic effect is believed to be accomplished through cis-activation or trans-activation of cellular genes. In cis-activation, genomic instability is a result of integration of HBV DNA into the host genome. In trans-activation, HBV proteins activate transcription of the HBV genome and host genes by binding to cellular sequences[48,50,51]. For HCV, a direct cytopathic effect has also been reported. As HCV is a RNA virus, it cannot integrate into the host genome. Therefore, other pathways must be of importance[52,53].

As a consequence of the chronic inflammatory state of the infected liver and the direct oncogenic effects of the hepatic viruses (as mentioned before), genetic alterations occur in various cellular pathways, which might eventually lead to the development of HCC. HBV proteins have been shown to manipulate the p53-, Rb-, cyclinD1- and p21-genes[51]. HCV is frequently associated with mutations of p53 and β-catenin[52,53].


Approximately 380 million people are chronically infected with HBV. These chronic HBV infected people have a 100-fold greater lifetime risk of developing HCC in comparison with non-carriers[3]. For this reason, HBV infection is the leading risk factor for the development of HCC. Worldwide, over 50% of HCC cases are associated with chronic HBV infection and the highest incidence of HCC is in South East Asia and sub-Saharan Africa, regions with a high prevalence of HBV infection[44,54].

As early as 1985, the first transgenic mouse models for investigating HBV infection were developed[55,56]. HBV transgenic mice have been created with the full HBV genome and with every single HBV gene, namely those encoding for the surface envelope proteins (large, middle and small), X protein (HBx), core and precore proteins.

It did not take long before the first transgenic mouse models for evaluation of HBV-associated hepatocarcinogenesis appeared. Until now, merely the large envelope protein and the HBx protein have displayed a carcinogenic role[48,57].

Chisari et al[58] described a mouse model in which transgenic mice were generated that carried an integrated HBV DNA fragment coding for the HBV large envelope polypeptides on a C57BL/6 genetic background. As a result, non-secretable hepatitis B surface antigen (HBsAg) particles formed that accumulated in the endoplasmic reticulum of the hepatocyte. In mice with 100% of the hepatocytes expressing HBsAg (lineage 50-4), liver injury begins at 2-3 mo of age; at 6 mo regenerative nodules appear and from the age of 15 mo HCCs develop.

Another HBV gene that has been extensively studied is the HBx gene. Though several research groups could not find evidence for a hepatocarcinogenic role of HBx in HBx transgenic mice[54,59], Kim et al[60] did report such a role in 1991. They produced HBx transgenic mice by injection of HBV DNA containing the HBx gene into single-cell embryos from CD1-mice. In these mice, liver tumors began to emerge after 8-10 mo. Both male and female transgenic mice died early in comparison with control CD1 mice, at the age of 11-15 mo vs 17-21 mo, respectively. On autopsy, 80%-91% of male transgenic mice and 60%-67% of female transgenic mice showed one or multiple HCCs. Yu et al[61] generated transgenic HBx mice using the same technique as Kim et al[60], but in a C57BL/6 genetic background and with a much weaker HBx expression in the liver. They reported an incidence of grossly identified HCCs and small neoplastic nodules, without signs of cirrhosis or inflammation, in 86% of 11-18 mo old HBx transgenic mice.

Possible explanations for the different outcome in transgenic mouse models for the hepatocarcinogenic role of HBx, may include a difference in mouse strains that were used. Male mice of the CD-1 strain develop spontaneous HCC in 5.7%[62], an incidence that is somewhat higher than the rate in for instance C57BL/6J mice (< 4.0%)[61]. In addition, the expression level of HBx-mRNA in the livers of transgenic mice and the type of HBx used may be different in the various studies. Finally, the integration site of HBx in the genome of the mice might influence the hepatocarcinogenic effect of HBx[57,61,63].

Efforts have been made to accomplish a model in which complete HBV genome transgenic mice demonstrate HCCs. Thus far, this has not been successful[63,64].

Nowadays, models based on the HBsAg transgenic mouse model of Chisari et al and the HBx transgenic mouse model of Kim et al[60] are commonly used to study mechanisms involved in hepatocarcinogenesis. These models are also applied to study possible synergistic relations between chemical carcinogens (such as aflatoxin B1 or diethyl nitrosamine) and HBV-infection[65-67]. Another application is the use of bitransgenic mouse models, in which mice are produced that are transgenic for a gene of interest (such as c-myc or TGF-α) in conjunction with HBsAg or HBx[48,68,69]. For instance, Jakubczak et al[68] produced bitransgenic mice by pairing HBsAg transgenic mice described by Chisari et al to TGF-α transgenic mice. At 8 mo of age, 76% (13 of 17) of male bitransgenic mice developed HCCs, while TGF-α transgenic control mice showed HCC in only 6% (1 of 17) and HBsAg transgenic control mice in 0%. These bitransgenic mouse models can be used to investigate the effect of a particular gene on HBV-induced hepatocarcinogenesis.


Worldwide, approximately 30% of HCC cases are related to chronic HCV infection, making HCV the second most frequent cause of HCC[44]. In some areas, like Southern Europe and Japan, HCV infection is the strongest predisposing factor for HCC[70]. Patients infected with HCV have a risk of up to 35% for developing liver cirrhosis[47,70]. Thereafter, the cumulative risk of developing HCC in these cirrhotic patients is 1%-7% per year. HCC is the most frequent cause of death in HCV infected persons[47,70,71].

Various HCV proteins have been expressed in transgenic mice to study the pathogenesis of HCV-associated HCC, particularly the HCV polyprotein, the core protein and the core protein in combination with E1 and E2 envelope proteins. Interestingly, the expression of the core protein of HCV seems to be the major factor contributing to the hepatocarcinogenic effect of HCV infection, as transgenic mice that do not express this protein, no HCCs arise[47,52].

Moriya et al[72] were the first to describe such a transgenic mouse model. They generated transgenic mice that carried the HCV core gene. These mice showed histological features of steatosis in the liver, without inflammation, from the age of 3 mo and showed HCCs with close histological resemblance of HCCs in human chronic HCV infection, by the time they were 16 mo old. The incidence of HCC in 16-19 mo old male transgenic mice was 26% to 31%, in contrast to a low incidence in the female transgenic mice, which is in accordance with the human situation[73]. By means of such transgenic mouse models numerous molecular and pathogenetic pathways have been investigated that have led to a better understanding of HCV-associated hepatocarcinogenesis.

To study the role of HCV proteins other than the HCV core protein in hepatocarcinogenesis, Lerat et al[74] developed full-length HCV polyprotein transgenic mice and compared them with transgenic mice encoding merely the structural HCV proteins (including the core and the E1 and E2 envelope proteins). HCCs occurred (exclusively in males) in 5 of 38 transgenic mice expressing the full HCV polyprotein and in 1 of 43 transgenic mice expressing the structural HCV proteins. These findings suggest that HCV proteins, other than the HCV core protein, may endorse development of HCC as well, because in these mice the HCV protein levels are much lower in the first group[74].

The HCCs that develop in the mouse models described by Moriya et al and Lerat et al show proper (histological) resemblance to the corresponding lesions in patients with HCV-associated HCC. In the model described by Lerat et al[74], tumors develop regardless of the absence of detectable levels of the expression of HCV proteins, mimicking the situation in HCV infected patients. Furthermore, constitutive HCV gene expression results in immunological tolerance to the HCV genes, allowing the study of the direct hepatocarcinogenic effect of HCV proteins in the absence of an immune response to the viral proteins[52,74]. Disadvantages of these models are the possible significance of the genetic background of the mice and the relative unpredictability of HCC formation.

As described above for HBV, the HCV transgenic mouse models of Moriya et al, Lerat et al and comparable models, are presently used to study carcinogenic mechanisms in HCV-related HCC. In addition, these models are applied to study relations between carcinogens (like DEN) and HCV-infection in inducing HCC[75,76]. Furthermore, bitransgenic mouse models have been developed to investigate the effect of a particular gene on HCV-induced hepatocarcinogenesis.


Mouse models in cancer research are developed to imitate human carcinogenesis. Although the ideal animal model does not (yet) exist, mouse models can imitate parts of human carcinogenesis. To date, different types of mouse models are available to induce HCC, varying in complexity. The most appropriate model for a particular research question should be chosen to answer that specific question. Each approach has its own advantages and disadvantages, which are discussed in this review.

First, carcinogen-induced models are used to identify chemicals that might be carcinogenic to humans. Furthermore, these models are used for establishing a relationship between carcinogen exposure and specific genetic changes. In HCC, DEN is especially used to induce HCC. Major disadvantages remain the influence of sex, age and genetic background of the mice on the predictability of HCC-development. Moreover, there is a species difference in the response to hepatocarcinogens between humans and mice.

Next, because of their suitability, implantation models are still frequently used for the screening of different types of anticancer drugs. Nonetheless, these models have a poor predictive value for the anti-tumor effects in patients. This is probably the consequence of culturing the tumor cells for a long period, which alters the molecular characteristics and the heterogeneity of the original tumor.

Agents tested in mice with subcutaneously implanted tumors, should always be tested in orthotopic models as well, because of the importance of the microenvironment on the biological behaviour of malignant cells.

HCC is a result of different genetic mutations. Conventional transgenic mouse models have been developed to study the role of different genes in HCC formation and to study molecular features involved in hepatocarcinogenesis. Limitations of these models include the expression of the transgene in all hepatocytes (hence, the tumor microenvironment as well) and the presence of the genetic alterations during embryogenesis (which might activate compensatory pathways). To overcome these problems, conditional mouse models have been developed recently. To date, these models are mainly used to study the effect of genetic alterations and, unfortunately, not to induce HCCs. Genetically engineered mouse models are also used in studying the role of viral hepatitis in HCC formation, as HBV and HCV require human hepatocytes to induce hepatitis and, consequently, hepatitis-induced hepatocarcinogenesis.

Highly sophisticated genetically engineered mouse models will become increasingly available and will help to answer a variety of research questions. Nevertheless, significant differences between mice and humans have to be taken into account when interpreting the (molecular) mechanisms of hepatocarcinogenesis. The most familiar of these (interspecies) differences are the much longer telomeres in mice, due to persistent telomerase expression in mice (as opposed to limited or absent telomerase expression in humans). Humans also differ from mice in respect to, for instance, their metabolism and immune system. To extrapolate the results from cancer studies in mice to humans, humanized mice should be generated and used in (genetically engineered) mouse models in the future.

This will bring us one step closer to the ideal animal model for cancer research.

As mentioned above, the adequate mouse model will be used depending on the research question. However, whether the impact of a possible carcinogen is investigated, the development of anticancer drugs or the genetic background of HCC formation is studied, all experiments have the goal to reduce the prevalence of HCC.


Peer reviewer: Ezio Laconi, MD, PhD, Professor of General Pathology, Department of Sciences and Biomedical Technologies, Unit of Experimental Pathology, University of Cagliari, Via Porcell, 4-IV Piano, 09125-Cagliari, Italy

S- Editor Li DL L- Editor Rippe RA E- Editor Ma WH

1.  American Cancer Society. Global Cancer Facts & Figures 2007.  Available from: URL: http://www cancer org/docroot/STT/stt_0_2007 asp?sitearea=STT&level=1.  [PubMed]  [DOI]
2.  Farazi PA, DePinho RA. Hepatocellular carcinoma pathogenesis: from genes to environment. Nat Rev Cancer. 2006;6:674-687.  [PubMed]  [DOI]
3.  Llovet JM, Burroughs A, Bruix J. Hepatocellular carcinoma. Lancet. 2003;362:1907-1917.  [PubMed]  [DOI]
4.  Frese KK, Tuveson DA. Maximizing mouse cancer models. Nat Rev Cancer. 2007;7:645-658.  [PubMed]  [DOI]
5.  Naugler WE, Sakurai T, Kim S, Maeda S, Kim K, Elsharkawy AM, Karin M. Gender disparity in liver cancer due to sex differences in MyD88-dependent IL-6 production. Science. 2007;317:121-124.  [PubMed]  [DOI]
6.  Chen CJ, Yu MW, Liaw YF. Epidemiological characteristics and risk factors of hepatocellular carcinoma. J Gastroenterol Hepatol. 1997;12:S294-S308.  [PubMed]  [DOI]
7.  Buchmann A, Bauer-Hofmann R, Mahr J, Drinkwater NR, Luz A, Schwarz M. Mutational activation of the c-Ha-ras gene in liver tumors of different rodent strains: correlation with susceptibility to hepatocarcinogenesis. Proc Natl Acad Sci USA. 1991;88:911-915.  [PubMed]  [DOI]
8.  Lee GH. Paradoxical effects of phenobarbital on mouse hepatocarcinogenesis. Toxicol Pathol. 2000;28:215-225.  [PubMed]  [DOI]
9.  Hann B, Balmain A. Building ‘validated’ mouse models of human cancer. Curr Opin Cell Biol. 2001;13:778-784.  [PubMed]  [DOI]
10.  Wogan GN. Impacts of chemicals on liver cancer risk. Semin Cancer Biol. 2000;10:201-210.  [PubMed]  [DOI]
11.  Williams GM. Chemicals with carcinogenic activity in the rodent liver; mechanistic evaluation of human risk. Cancer Lett. 1997;117:175-188.  [PubMed]  [DOI]
12.  Seitz HK, Stickel F. Molecular mechanisms of alcohol-mediated carcinogenesis. Nat Rev Cancer. 2007;7:599-612.  [PubMed]  [DOI]
13.  Gonzalez FJ. The peroxisome proliferator-activated receptor alpha (PPARalpha): role in hepatocarcinogenesis. Mol Cell Endocrinol. 2002;193:71-79.  [PubMed]  [DOI]
14.  Huff J, Cirvello J, Haseman J, Bucher J. Chemicals associated with site-specific neoplasia in 1394 long-term carcinogenesis experiments in laboratory rodents. Environ Health Perspect. 1991;93:247-270.  [PubMed]  [DOI]
15.  Vesselinovitch SD, Mihailovich N. Kinetics of diethylnitro-samine hepatocarcinogenesis in the infant mouse. Cancer Res. 1983;43:4253-4259.  [PubMed]  [DOI]
16.  Vesselinovitch SD, Koka M, Mihailovich N, Rao KV. Carcinogenicity of diethylnitrosamine in newborn, infant, and adult mice. J Cancer Res Clin Oncol. 1984;108:60-65.  [PubMed]  [DOI]
17.  Vesselinovitch SD. Infant mouse as a sensitive bioassay system for carcinogenicity of N-nitroso compounds. IARC Sci Publ. 1980;108:645-655.  [PubMed]  [DOI]
18.  Anisimov VN, Ukraintseva SV, Yashin AI. Cancer in rodents: does it tell us about cancer in humans? Nat Rev Cancer. 2005;5:807-819.  [PubMed]  [DOI]
19.  Hirst GL, Balmain A. Forty years of cancer modelling in the mouse. Eur J Cancer. 2004;40:1974-1980.  [PubMed]  [DOI]
20.  Troiani T, Schettino C, Martinelli E, Morgillo F, Tortora G, Ciardiello F. The use of xenograft models for the selection of cancer treatments with the EGFR as an example. Crit Rev Oncol Hematol. 2008;65:200-211.  [PubMed]  [DOI]
21.  Killion JJ, Radinsky R, Fidler IJ. Orthotopic models are necessary to predict therapy of transplantable tumors in mice. Cancer Metastasis Rev. 1998;17:279-284.  [PubMed]  [DOI]
22.  Heijstek MW, Kranenburg O, Borel Rinkes IH. Mouse models of colorectal cancer and liver metastases. Dig Surg. 2005;22:16-25.  [PubMed]  [DOI]
23.  Clarke R. Human breast cancer cell line xenografts as models of breast cancer. The immunobiologies of recipient mice and the characteristics of several tumorigenic cell lines. Breast Cancer Res Treat. 1996;39:69-86.  [PubMed]  [DOI]
24.  Bankert RB, Egilmez NK, Hess SD. Human-SCID mouse chimeric models for the evaluation of anti-cancer therapies. Trends Immunol. 2001;22:386-393.  [PubMed]  [DOI]
25.  Tang ZY, Sun FX, Tian J, Ye SL, Liu YK, Liu KD, Xue Q, Chen J, Xia JL, Qin LX. Metastatic human hepatocellular carcinoma models in nude mice and cell line with metastatic potential. World J Gastroenterol. 2001;7:597-601.  [PubMed]  [DOI]
26.  Jonkers J, Berns A. Conditional mouse models of sporadic cancer. Nat Rev Cancer. 2002;2:251-265.  [PubMed]  [DOI]
27.  Khanna C, Hunter K. Modeling metastasis in vivo. Carcinogenesis. 2005;26:513-523.  [PubMed]  [DOI]
28.  Bibby MC. Orthotopic models of cancer for preclinical drug evaluation: advantages and disadvantages. Eur J Cancer. 2004;40:852-857.  [PubMed]  [DOI]
29.  Becher OJ, Holland EC. Genetically engineered models have advantages over xenografts for preclinical studies. Cancer Res. 2006;66:3355-8335, discussion 3358-3359.  [PubMed]  [DOI]
30.  Shuldiner AR. Transgenic animals. N Engl J Med. 1996;334:653-655.  [PubMed]  [DOI]
31.  Palmiter RD, Brinster RL, Hammer RE, Trumbauer ME, Rosenfeld MG, Birnberg NC, Evans RM. Dramatic growth of mice that develop from eggs microinjected with metallothionein-growth hormone fusion genes. Nature. 1982;300:611-615.  [PubMed]  [DOI]
32.  Buendia MA. Genetics of hepatocellular carcinoma. Semin Cancer Biol. 2000;10:185-200.  [PubMed]  [DOI]
33.  Colnot S, Decaens T, Niwa-Kawakita M, Godard C, Hamard G, Kahn A, Giovannini M, Perret C. Liver-targeted disruption of Apc in mice activates beta-catenin signaling and leads to hepatocellular carcinomas. Proc Natl Acad Sci USA. 2004;101:17216-17221.  [PubMed]  [DOI]
34.  Ali SH, DeCaprio JA. Cellular transformation by SV40 large T antigen: interaction with host proteins. Semin Cancer Biol. 2001;11:15-23.  [PubMed]  [DOI]
35.  Ahuja D, Saenz-Robles MT, Pipas JM. SV40 large T antigen targets multiple cellular pathways to elicit cellular transformation. Oncogene. 2005;24:7729-7745.  [PubMed]  [DOI]
36.  Dubois N, Bennoun M, Allemand I, Molina T, Grimber G, Daudet-Monsac M, Abelanet R, Briand P. Time-course development of differentiated hepatocarcinoma and lung metastasis in transgenic mice. J Hepatol. 1991;13:227-239.  [PubMed]  [DOI]
37.  Kitagawa T, Hino O, Lee GH, Li H, Liu J, Nomura K, Ohtake K, Furuta Y, Aizawa S. Multistep hepatocarcinogenesis in transgenic mice harboring SV40 T-antigen gene. Princess Takamatsu Symp. 1991;22:349-360.  [PubMed]  [DOI]
38.  Sepulveda AR, Finegold MJ, Smith B, Slagle BL, DeMayo JL, Shen RF, Woo SL, Butel JS. Development of a transgenic mouse system for the analysis of stages in liver carcinogenesis using tissue-specific expression of SV40 large T-antigen controlled by regulatory elements of the human alpha-1-antitrypsin gene. Cancer Res. 1989;49:6108-6117.  [PubMed]  [DOI]
39.  Murakami H, Sanderson ND, Nagy P, Marino PA, Merlino G, Thorgeirsson SS. Transgenic mouse model for synergistic effects of nuclear oncogenes and growth factors in tumorigenesis: interaction of c-myc and transforming growth factor alpha in hepatic oncogenesis. Cancer Res. 1993;53:1719-1723.  [PubMed]  [DOI]
40.  Santoni-Rugiu E, Nagy P, Jensen MR, Factor VM, Thorgeirsson SS. Evolution of neoplastic development in the liver of transgenic mice co-expressing c-myc and transforming growth factor-alpha. Am J Pathol. 1996;149:407-428.  [PubMed]  [DOI]
41.  Lewis BC, Klimstra DS, Socci ND, Xu S, Koutcher JA, Varmus HE. The absence of p53 promotes metastasis in a novel somatic mouse model for hepatocellular carcinoma. Mol Cell Biol. 2005;25:1228-1237.  [PubMed]  [DOI]
42.  Lou DQ, Molina T, Bennoun M, Porteu A, Briand P, Joulin V, Vasseur-Cognet M, Cavard C. Conditional hepatocarcinogenesis in mice expressing SV 40 early sequences. Cancer Lett. 2005;229:107-114.  [PubMed]  [DOI]
43.  Harada N, Oshima H, Katoh M, Tamai Y, Oshima M, Taketo MM. Hepatocarcinogenesis in mice with beta-catenin and Ha-ras gene mutations. Cancer Res. 2004;64:48-54.  [PubMed]  [DOI]
44.  Parkin DM. The global health burden of infection-associated cancers in the year 2002. Int J Cancer. 2006;118:3030-3044.  [PubMed]  [DOI]
45.  Fattovich G, Stroffolini T, Zagni I, Donato F. Hepatocellular carcinoma in cirrhosis: incidence and risk factors. Gastroenterology. 2004;127:S35-S50.  [PubMed]  [DOI]
46.  Dandri M, Volz TK, Lutgehetmann M, Petersen J. Animal models for the study of HBV replication and its variants. J Clin Virol. 2005;34 Suppl 1:S54-S62.  [PubMed]  [DOI]
47.  Kremsdorf D, Brezillon N. New animal models for hepatitis C viral infection and pathogenesis studies. World J Gastroenterol. 2007;13:2427-2435.  [PubMed]  [DOI]
48.  Singh M, Kumar V. Transgenic mouse models of hepatitis B virus-associated hepatocellular carcinoma. Rev Med Virol. 2003;13:243-253.  [PubMed]  [DOI]
49.  Cougot D, Neuveut C, Buendia MA. HBV induced carcinogenesis. J Clin Virol. 2005;34 Suppl 1:S75-S78.  [PubMed]  [DOI]
50.  Brechot C. Pathogenesis of hepatitis B virus-related hepato-cellular carcinoma: old and new paradigms. Gastroenterology. 2004;127:S56-S61.  [PubMed]  [DOI]
51.  Park NH, Song IH, Chung YH. Chronic hepatitis B in hepatocarcinogenesis. Postgrad Med J. 2006;82:507-515.  [PubMed]  [DOI]
52.  Liang TJ, Heller T. Pathogenesis of hepatitis C-associated hepatocellular carcinoma. Gastroenterology. 2004;127:S62-S71.  [PubMed]  [DOI]
53.  Guidotti LG, Chisari FV. Immunobiology and pathogenesis of viral hepatitis. Annu Rev Pathol. 2006;1:23-61.  [PubMed]  [DOI]
54.  Lee TH, Finegold MJ, Shen RF, DeMayo JL, Woo SL, Butel JS. Hepatitis B virus transactivator X protein is not tumorigenic in transgenic mice. J Virol. 1990;64:5939-5947.  [PubMed]  [DOI]
55.  Chisari FV, Pinkert CA, Milich DR, Filippi P, McLachlan A, Palmiter RD, Brinster RL. A transgenic mouse model of the chronic hepatitis B surface antigen carrier state. Science. 1985;230:1157-1160.  [PubMed]  [DOI]
56.  Babinet C, Farza H, Morello D, Hadchouel M, Pourcel C. Specific expression of hepatitis B surface antigen (HBsAg) in transgenic mice. Science. 1985;230:1160-1163.  [PubMed]  [DOI]
57.  Akbar SK, Onji M. Hepatitis B virus (HBV)-transgenic mice as an investigative tool to study immunopathology during HBV infection. Int J Exp Pathol. 1998;79:279-291.  [PubMed]  [DOI]
58.  Chisari FV, Klopchin K, Moriyama T, Pasquinelli C, Dunsford HA, Sell S, Pinkert CA, Brinster RL, Palmiter RD. Molecular pathogenesis of hepatocellular carcinoma in hepatitis B virus transgenic mice. Cell. 1989;59:1145-1156.  [PubMed]  [DOI]
59.  Reifenberg K, Lohler J, Pudollek HP, Schmitteckert E, Spindler G, Kock J, Schlicht HJ. Long-term expression of the hepatitis B virus core-e- and X-proteins does not cause pathologic changes in transgenic mice. J Hepatol. 1997;26:119-130.  [PubMed]  [DOI]
60.  Kim CM, Koike K, Saito I, Miyamura T, Jay G. HBx gene of hepatitis B virus induces liver cancer in transgenic mice. Nature. 1991;351:317-320.  [PubMed]  [DOI]
61.  Yu DY, Moon HB, Son JK, Jeong S, Yu SL, Yoon H, Han YM, Lee CS, Park JS, Lee CH. Incidence of hepatocellular carcinoma in transgenic mice expressing the hepatitis B virus X-protein. J Hepatol. 1999;31:123-132.  [PubMed]  [DOI]
62.  Chandra M, Frith CH. Spontaneous neoplasms in aged CD-1 mice. Toxicol Lett. 1992;61:67-74.  [PubMed]  [DOI]
63.  Bagis H, Arat S, Mercan HO, Aktoprakligil D, Caner M, Turanli ET, Baysal K, Turgut G, Sekmen S, Cirakoglu B. Stable transmission and expression of the hepatitis B virus total genome in hybrid transgenic mice until F10 generation. J Exp Zool A Comp Exp Biol. 2006;305:420-427.  [PubMed]  [DOI]
64.  Guidotti LG, Matzke B, Schaller H, Chisari FV. High-level hepatitis B virus replication in transgenic mice. J Virol. 1995;69:6158-6169.  [PubMed]  [DOI]
65.  Ghebranious N, Sell S. Hepatitis B injury, male gender, aflatoxin, and p53 expression each contribute to hepatocarci-nogenesis in transgenic mice. Hepatology. 1998;27:383-391.  [PubMed]  [DOI]
66.  Sell S, Hunt JM, Dunsford HA, Chisari FV. Synergy between hepatitis B virus expression and chemical hepatocarcinogens in transgenic mice. Cancer Res. 1991;51:1278-1285.  [PubMed]  [DOI]
67.  Kaplanski C, Chisari FV, Wild CP. Minisatellite rearrangements are increased in liver tumours induced by transplacental aflatoxin B1 treatment of hepatitis B virus transgenic mice, but not in spontaneously arising tumours. Carcinogenesis. 1997;18:633-639.  [PubMed]  [DOI]
68.  Jakubczak JL, Chisari FV, Merlino G. Synergy between transforming growth factor alpha and hepatitis B virus surface antigen in hepatocellular proliferation and carcinogenesis. Cancer Res. 1997;57:3606-3611.  [PubMed]  [DOI]
69.  Terradillos O, Billet O, Renard CA, Levy R, Molina T, Briand P, Buendia MA. The hepatitis B virus X gene potentiates c-myc-induced liver oncogenesis in transgenic mice. Oncogene. 1997;14:395-404.  [PubMed]  [DOI]
70.  Di Bisceglie AM. Hepatitis C and hepatocellular carcinoma. Hepatology. 1997;26:34S-38S.  [PubMed]  [DOI]
71.  Koike K, Tsutsumi T, Fujie H, Shintani Y, Kyoji M. Molecular mechanism of viral hepatocarcinogenesis. Oncology. 2002;62 Suppl 1:29-37.  [PubMed]  [DOI]
72.  Moriya K, Yotsuyanagi H, Shintani Y, Fujie H, Ishibashi K, Matsuura Y, Miyamura T, Koike K. Hepatitis C virus core protein induces hepatic steatosis in transgenic mice. J Gen Virol. 1997;78:1527-1531.  [PubMed]  [DOI]
73.  Moriya K, Fujie H, Shintani Y, Yotsuyanagi H, Tsutsumi T, Ishibashi K, Matsuura Y, Kimura S, Miyamura T, Koike K. The core protein of hepatitis C virus induces hepatocellular carcinoma in transgenic mice. Nat Med. 1998;4:1065-1067.  [PubMed]  [DOI]
74.  Lerat H, Honda M, Beard MR, Loesch K, Sun J, Yang Y, Okuda M, Gosert R, Xiao SY, Weinman SA. Steatosis and liver cancer in transgenic mice expressing the structural and nonstructural proteins of hepatitis C virus. Gastroenterology. 2002;122:352-365.  [PubMed]  [DOI]
75.  Kamegaya Y, Hiasa Y, Zukerberg L, Fowler N, Blackard JT, Lin W, Choe WH, Schmidt EV, Chung RT. Hepatitis C virus acts as a tumor accelerator by blocking apoptosis in a mouse model of hepatocarcinogenesis. Hepatology. 2005;41:660-667.  [PubMed]  [DOI]
76.  Kato T, Miyamoto M, Date T, Yasui K, Taya C, Yonekawa H, Ohue C, Yagi S, Seki E, Hirano T. Repeated hepatocyte injury promotes hepatic tumorigenesis in hepatitis C virus transgenic mice. Cancer Sci. 2003;94:679-685.  [PubMed]  [DOI]
77.  Jhappan C, Stahle C, Harkins RN, Fausto N, Smith GH, Merlino GT. TGF alpha overexpression in transgenic mice induces liver neoplasia and abnormal development of the mammary gland and pancreas. Cell. 1990;61:1137-1146.  [PubMed]  [DOI]
78.  Lee GH, Merlino G, Fausto N. Development of liver tumors in transforming growth factor alpha transgenic mice. Cancer Res. 1992;52:5162-5170.  [PubMed]  [DOI]
79.  Conner EA, Lemmer ER, Omori M, Wirth PJ, Factor VM, Thorgeirsson SS. Dual functions of E2F-1 in a transgenic mouse model of liver carcinogenesis. Oncogene. 2000;19:5054-5062.  [PubMed]  [DOI]
80.  Calvisi DF, Conner EA, Ladu S, Lemmer ER, Factor VM, Thorgeirsson SS. Activation of the canonical Wnt/beta-catenin pathway confers growth advantages in c-Myc/E2F1 transgenic mouse model of liver cancer. J Hepatol. 2005;42:842-849.  [PubMed]  [DOI]