Dr Nirmitha Herath, Leukaemia Foundation Laboratory, Queensland Institute of Medical Research, H-Floor, CBCRC, 300, Herston Road, Brisbane, QLD 4029, Australia. Email: nirmithH@qimr.edu.au
Abstract Hepatocellular carcinoma (HCC) is associated with multiple risk factors and is believed to arise from pre-neoplastic lesions, usually in the background of cirrhosis. However, the genetic and epigenetic events of hepatocarcinogenesis are relatively poorly understood. HCC display gross genomic alterations, including chromosomal instability (CIN), CpG island methylation, DNA rearrangements associated with hepatitis B virus (HBV) DNA integration, DNA hypomethylation and, to a lesser degree, microsatellite instability. Various studies have reported CIN at chromosomal regions, 1p, 4q, 5q, 6q, 8p, 10q, 11p, 16p, 16q, 17p and 22q. Frequent promoter hypermethylation and subsequent loss of protein expression has also been demonstrated in HCC at tumor suppressor gene (TSG), p16, p14, p15, SOCS1, RIZ1, E-cadherin and 14–3-3 σ. An interesting observation emerging from these studies is the presence of a methylator phenotype in hepatocarcinogenesis, although it does not seem advantageous to have high levels of microsatellite instability. Methylation also appears to be an early event, suggesting that this may precede cirrhosis. However, these genes have been studied in isolation and global studies of methylator phenotype are required to assess the significance of epigenetic silencing in hepatocarcinogenesis. Based on previous data there are obvious fundamental differences in the mechanisms of hepatic carcinogenesis, with at least two distinct mechanisms of malignant transformation in the liver, related to CIN and CpG island methylation. The reason for these differences and the relative importance of these mechanisms are not clear but likely relate to the etiopathogenesis of HCC. Defining these broad mechanisms is a necessary prelude to determine the timing of events in malignant transformation of the liver and to investigate the role of known risk factors for HCC.
Hepatocellular carcinoma (HCC) is one of the most common malignancies worldwide, with an estimated annual incidence of 1 million cases.1 It has a poor prognosis with most patients dying within 1 year of diagnosis.1 Several clinical and histological patterns of HCC are recognized and the prognosis may differ according to associated conditions and the extent of tumor progression at diagnosis.
Most HCC arise on the background of chronic liver disease, usually in association with cirrhosis. Chronic hepatitis B virus (HBV), hepatitis C virus (HCV), aflatoxin, alcohol consumption and hemochromatosis are major factors associated with the development of HCC (Fig. 1). In most chronic liver diseases, patients with HCC are generally cirrhotic or have marked hepatic fibrosis. Thus cirrhosis is considered a premalignant liver lesion. However, cirrhosis is not always present in individuals with HCC and at times there is no clinical or laboratory evidence of underlying chronic liver disease, with the non-malignant liver being histologically unremarkable.2
REGIONAL VARIATION IN THE INCIDENCE OF HEPATOCELLULAR CARCINOMA
There are regional variations in the clinical presentation and the natural history of HCC. In black South African patients it is uncommon for HCC to be associated with advanced cirrhosis and these tumors are often poorly differentiated.3 In contrast, in other parts of the world HCC usually develop in livers with cirrhosis.3 Despite these differences the poor prognosis of HCC is similar around the world.
The estimated incidence of HCC worldwide is 3.1 cases per 100 000 people per year.1 Higher incidences are documented in East Asian countries such as China, Taiwan and Korea and in sub-Saharan Africa where the highest incidence is recorded in Mozambique.4 In Western countries such as the USA, Europe and Australia, the incidence is approximately 2–5 cases per 100 000 people per year. However, the rates of HCC incidence and mortality in Australia and around the world have increased over the past two decades and are continuing to rise, most likely due to the increasing prevalence of HCV.5
INTERACTIONS BETWEEN HOST AND ENVIRONMENTAL FACTORS AND HEPATOCARCINOGENESIS
The potential roles of environmental factors and their interaction with host factors can be illustrated by reviewing the role of aflatoxin B1 and the p53 tumor suppressor gene (TSG) in hepatocarcinogenesis. Aflatoxin B1 is produced by the fungus Aspergillus flavus which is a common food contaminant in African and Asian countries.4 Aflatoxin B1 is metabolized in the liver into a potent carcinogen, aflatoxin 8, 9-epoxide, which is usually detoxified by microsomal enzymes.4 A failure of detoxification processes can allow aflatoxin 8, 9-epoxide to bind to guanosine residues and ultimately lead to a G-T transversion (or mutation) at these sites.4
Codon 249 of p53 TSG is a particular target for this process in the liver.6 This mutation inactivates p53 and is virtually only seen in patients exposed to aflatoxin B1 in their diet.5 Whether aflatoxin acts alone or as a cocarcinogen is unknown. It has also been suggested that genetic variation can make individuals more susceptible to this toxin. It has recently been shown that patients with HCC from areas with high aflatoxin exposure are more likely to have inherited isoforms of microsomal epoxide hydrolase and glutathione S-transferase with either reduced or no activity.7 That is, patients from these regions with HCC were less efficient in the detoxification of carcinogenic aflatoxin metabolites. Furthermore, these patients were more likely to have HBV infection,7 although HBV infection alone has been shown to not be associated with the p53 codon 249 mutation. Thus there is an incompletely understood interplay between dietary aflatoxin exposure, chronic hepatitis B infection and genetic factors that contributes to the high incidence of HCC in Africa and China.
Interestingly, the majority of HCC from HBV virus endemic areas appear to contain integrated viral sequences.8 The X gene of HBV is highly conserved among different viral isolates. Although its function is not yet fully understood, this gene is known to play a regulatory role in HBV replication. It has been suggested that the binding of the HBx protein onto the carboxyl domain of the p53 protein may disrupt p53 induced apoptosis. It has been further suggested that HBV viral integration may facilitate chromosomal deletions.8
It seems likely that global variation in the incidence of HCC is due to regional differences in environmental factors and their interactions with host factors reflecting genetic background and other disease processes. However, apart from the role of aflatoxin B1 and chronic viral hepatitis infections, the various risk factors for HCC and their genetic interactions are relatively poorly understood.
GENETIC CHANGES ASSOCIATED WITH CARCINOGENESIS
It is widely accepted that the accumulation of genetic alterations underlie the evolution of most malignancies, including HCC. HCC exhibit numerous genetic abnormalities, including chromosomal deletions and rearrangements, aneuploidy, gene amplifications and mutations, as well as epigenetic alterations including modulation of DNA methylation.
MECHANISMS OF MALIGNANT TRANSFORMATION
Role of chromosomal instability in hepatocellular carcinoma
The most studied mechanism of genetic alterations in hepatocarcinogenesis involves the loss or gain of chromosomal segments during cell division. This is described as chromosomal instability (CIN).
Detailed studies of CIN in HCC have been performed by Nagai et al.9 and Boige et al.10 Nagai et al. studied 120 HCC from differing geographic areas with a series of 195 markers that spanned all chromosomal arms.9 The mean rate of CIN for any markers studied was 12.1 ± 8.4%, giving a background rate of regional CIN in HCC of up to 20%. A second study by Bogie et al. assessed 48 HCC from France using 275 chromosomal markers.10 These 48 cases were selected from a group of 100 HCC based on abnormal DNA content in the cells, with the underlying assumption that these cancers would be most likely to have CIN.
Both studies identified significant loss at similar sites including 1p, 4q, 6q, 8p, 13q, 16p, 16q, and 17p. These regions are likely to contain TSG for HCC. Due to the relatively large number of cases and markers used, these studies were able to show that putative TSG for HCC may be present on chromosomes 1p, 4q, 8p and 13q.
An important flaw in both these studies was the failure to identify the study population, because as discussed above, geographic location may significantly alter the underlying risk factors and potentially the pathogenic pathways for hepatocarcinogenesis.
In an effort to define major genetic targets during HCC development, a number of studies have investigated one or several chromosomes for evidence of CIN. Various studies of HCC have identified CIN affecting chromosomal arms, 1p, 4q, 5q, 6q, 8p, 10q, 11p, 16p, 16q, 17p and 22q in HCC.11–23 However, most of these studies have been in Japanese populations and the majority of chromosomal regions were examined in isolation. The genetic alterations identified in these studies have also been found in other malignancies including breast cancers, small-cell lung carcinomas, prostate cancers, renal cell carcinomas, malignant melanomas and meningiomas, suggesting that TSG implicated in a wide spectrum of tumors may also be involved in hepatocarcinogenesis.
CLINICOPATHOLOGICAL CORRELATES OF HEPATOCELLULAR CARCINOMA
A number of studies have demonstrated that chromosomal abnormalities observed in HCC have clinicopathological correlates. Aberrations at a number of chromosomal arms have been reported in different stages of HCC and an increased prevalence of these aberrations has been found in more poorly differentiated cancers.24 The overall prevalence of chromosomal aberrations was also significantly associated with larger tumor size.24 Further studies have shown a positive correlation between the degree of CIN and alpha-fetoprotein levels; advanced tumor grade and stage; and in patients with non-HCV-related HCC.25 A large study by Kuroki et al. identified CIN at 1p to be frequent in well-differentiated tumors, while CIN at 4q, 8p, 8q, 13q, 16q and 17p was correlated with more advanced stages of HCC and poor differentiation.22 Other studies have also identified CIN at 4q in association with larger tumors and more aggressive histological grade.10,26
These studies suggest that multiple genetic changes accumulate during hepatocarcinogenesis and are associated with worse histological grades, larger tumor sizes and poorer prognosis. However, the sequence in which these genetic alterations participate in the evolution of HCC has not been established as clearly as in other cancers such as colon cancer.
CHROMOSOMAL INSTABILITY IN PREMALIGNANT LIVER TISSUE
A number of studies have identified CIN occurring in premalignant liver tissue. Tsopanomichalou et al.,27 in a study of blood and needle biopsy samples from normal liver, chronic hepatitis and cirrhotic liver, found a significant increase in the prevalence of CIN at chromosome 1 as liver lesions progressed towards malignant transformation. Another study of cirrhotic nodules found a high prevalence of CIN on chromosomes 6q and 8p.28 Yeh et al.,29 in a study of microdissected cirrhotic nodules, identified chromosomal aberrations at 4q, 8p and Xq in most cases. Other reports have demonstrated allelic loss at 1p, 1q, 3p, 4q, 6q, 7q, 8p, 13q and 18q in cirrhotic tissue, although with a lower prevalence than in HCC.30 Taken together, these studies suggest that cirrhotic liver can harbor premalignant genetic alterations and that these alterations probably represent the early steps in hepatocarcinogenesis.
ROLE OF DNA HYPERMETHYLATION IN HEPATOCELLULAR CARCINOMA
In addition to HCC with obvious CIN, there are also tumors with normal DNA content.15,31 The mechanism of malignant transformation in this second group of HCC is not clear, however, it may include epigenetic alterations. DNA methylation of CpG islands has emerged as a likely mechanism of epigenetic inactivation of TSG in a number of malignancies.32
Methylation of cytosine is the only known naturally occurring modification of DNA in mammals.33 Methylation of the 5′ position of cytosine leads to the formation of 5-methylcytosine (5-mC).33 This reaction is catalyzed by the enzyme, cytosine-5-DNA methyl-transferase.33 The majority of 5-mC is present immediately upstream of guanosine residues, in CpG dinucleotides.33 Approximately 70% of the CpG dinucleotides in the genome are methylated and this epigenetic modification is passed on to daughter cells.33
CpG dinucleotides are not uniformly distributed throughout the human genome.34 In 98% of the genome, CpGs are present approximately once per 80 dinucleotides.34 However, there are regions referred to as CpG islands, which comprise 1–2% of the genome. These are approximately 200 base pairs (bp) to several kilo bases (kb) in length and have a frequency of CpGs approximately five times greater than the genome as a whole.35 Unlike the rest of the genome, the cytosine residues in CpG islands are generally maintained in the unmethylated state.34 There are an estimated 45 000 CpG islands through the genome, with the majority being associated with the promoter regions of genes.33 Approximately 50–60% of all genes have associated CpG islands.34 Methylation of the cytosine residing in the CpG islands prevents transcription of the gene. Hypermethylation of the CpG islands in the promoters of TSG with silencing of the gene has been recognized in a variety of malignancies as contributing to tumor progression.34
To assess global patterns of methylation, Toyota et al. identified a collection of CpG islands that were methylated in colorectal cancer cell lines.36 Toyota et al. described these as methylated in tumor (MINT) clones, and methylation of these was studied in malignant and non-malignant tissue from patients with colorectal carcinoma.36 Some of the clones were methylated in an age-dependent fashion in both malignant and adjacent non-malignant colorectal tissue and were described as type A MINT clones. Another group of MINT was only methylated in malignant colorectal tissue and these MINT were described as being type C. When these type C MINT clones were studied in more detail it was apparent that some cancers had simultaneous methylation at multiple MINT, while there was another group with a low level of methylation at these MINT.36 The subgroup of cancers with frequent type C MINT methylation was described as having a CpG island methylator phenotype (CIMP).36 It was also suggested that in this group of cancers, promoter hypermethylation would be a key mechanism involved in the inactivation of a number of TSG.
Currently the role of methylation in hepatocarcinogenesis is not clear. Frequent promoter methylation and subsequent loss of protein expression has been demonstrated in HCC at TSG p16, E-cadherin and 14-3-3 σ.13,37,38 More recently, studies have reported methylation at p1539, SOCS140 and RIZ141 in HCC, however, protein expression was not assessed in these cancers. We have also previously demonstrated methylation of p1442, p1542 and E-cadherin43 in HCC.
However, these genes have generally been studied in isolation and global studies of methylator phenotype are required to assess the significance of epigenetic silencing in hepatocarcinogenesis. A recent study by Kondo et al.44 investigated methylation of p16, MLH1, thrombospondin-1 and MINT clones 1, 2, 12, 25 and 31 in HCC, adjacent non-malignant liver tissue and in histologically normal liver tissue. CpG islands of p16, and MINT 1, 2, 12 and 31 were observed to be frequently methylated in HCC samples.44 Promoter methylation was also detected to a lesser degree in adjacent non-malignant liver tissue and was considered to represent premalignant epigenetic alteration in these tissues. These studies provide evidence for a role of CpG island methylation in malignant transformation in a subgroup of HCC.
ROLE OF DNA HYPOMETHYLATION IN HEPATOCELLULAR CARCINOMA
It is known that the transcriptional activity of a gene is inversely related to its methylation status.45 As demonstrated recently, the growth hormone gene and globin genes of certain human cancers are hypomethylated and are distinguished from the adjacent normal tissue.46 A loss of the methyl group has also been observed in rapidly replicating cells of the mouse embryo.46 These data suggest the possibility that demethylation may be involved in enhancing transcriptional activity of oncogenes. There is potentially an association between hypomethylation and CIN. Further studies are required, using large numbers of HCC and other tumors to determine the prevalence of this form of genomic alteration in neoplasia.
ROLE OF DNA MISMATCH REPAIR
Another important mechanism of malignant transformation includes the DNA mismatch repair system (MMR). Defective DNA MMR is implicated in hereditary non-polyposis colorectal cancer and a variety of sporadic malignancies.47 The DNA MMR system is a group of proteins that repair short DNA mismatches that occur during DNA synthesis, thereby maintaining the fidelity of genomic DNA.47
Cells with defective DNA MMR usually have an increased mutation rate as a consequence of failure to repair sporadic mutations. Human Mut S homolog 2 (hMSH2) and human Mut L homolog 1 (hMLH1) play important roles in MMR in humans.48MSH2 and MLH1 are most commonly found mutated in hereditary non-polyposis colo-rectal cancer (CRC) and MLH1 is inactivated in 10% of sporadic CRC. MLH1 and MSH2 are not implicated in HCC.
MICROSATELLITE INSTABILITY AS AN INDICATOR OF MISMATCH REPAIR SYSTEM DEFICIENCY
Microsatellites are composed of multiple copies of short DNA repeat sequences (1–5 bases in length) and are widely and evenly distributed throughout the entire genome.47 Kinzler et al.49 identified a small group of CRC with minimal evidence of CIN. Subsequent studies showed that these tumors were diploid and had a mutator phenotype with a high frequency of mutation in microsatellites. This phenomenon was termed microsatellite instability (MSI). Microsatellites are prone to changes in the number of repeats present during DNA synthesis. These mutations are usually repaired by the DNA MMR system. Cells with defective DNA MMR exhibit MSI because the spontaneous mutations are not repaired. In CRC with defective DNA MMR, MSI will affect one-third or more of microsatellites made up of one or two base-pair sequences. These cancers are designated MSI-high (MSI-H) cancers. Of the remaining CRC, approximately 10% appear to have lower levels of MSI (MSI-L), while 70–75% of cancers do not have evidence of MSI and are designated microsatellite stable cancers (MSS).50,51 In other malignancies such as breast, endometrial and gastric carcinomas, varying degrees of MSI are recognized.
The biological and clinicopathological significance of MSI in HCC still remains to be determined. The presence of low level MSI in HCC has been reported in a few studies.44,47,52–54 However, the degree of MSI in HCC, when identified, has been less than that typically seen in malignancies with known impairment of DNA MMR, such as in HNPCC. These studies suggest that HCC rarely, if ever, display a MSI-H phenotype and argue against a role for DNA MMR in hepatic carcinogenesis.
INTERACTION BETWEEN PATHWAYS
An interesting observation emerging from these studies is the potential for interaction between these pathways. Methylation upstream of the hMLH1 promoter appears to be an early event in sporadic colorectal cancers with high levels of MSI,55 suggesting there may be a link between methylation and MSI. Toyota et al.36 detected a strong correlation between CIMP and both the presence of hMLH1 methylation and MSI in colorectal carcinoma. Interestingly, they found that only half of all their cancers with a methylator phenotype and no non-malignant colonic mucosa showed hMLH1 methylation. These findings demonstrate that the CpG island methylation phenotype emerges first, leading to hypermethylation of hMLH1 in a proportion of cancers and the subsequent development of MSI.
Subsequent studies in CRC, gastric and endometrial cancers have confirmed the existence of a link between methylation and MSI.56–58 Hypermethylation of the hMLH1 gene appeared to be a common occurrence in gastric tumors with high levels of MSI, whereas methylation of this gene is a rare event in tumors with low MSI levels.57 Furthermore, methylation of the APC gene in endometrial cancer occurred in an increased frequency in cancers with high MSI.58
Although CIMP appears to be a common event in hepatocarcinogenesis, unlike CRC it does not seem to lead to methylation of hMLH-1 and the development of MSI.
RELATIONSHIP BETWEEN CHROMOSOMAL INSTABILITY AND CPG ISLAND METHYLATION IN HEPATOCELLULAR CARCINOMA
Data from previous studies suggest that there are at least two distinct mechanisms of malignant transformation in the liver, related to CIN and CpG island methylation (Fig. 2). We have previously shown that CIN is more prevalent in South African HCC than in Australian HCC.42,43,59 While this may be a feature related to the chromosomal regions selected for study, the consistency of this observation in these studies at multiple chromosomal arms suggest that it is a generalized phenomenon. The clinical scenario for the black South African patients with HCC is usually dominated by the aggressive nature of the HCC. These patients are significantly younger and often present with acute abdomen catastrophe, resulting from rupture of the tumor.4 These patients are also exposed to powerful environmental carcinogens such as aflatoxin in the setting of chronic HBV infection. These HCC appear to progress through CIN.
HCC from Australian patients have typically been less aggressive with a silent course in their early stages. Patients with these cancers have been comparatively much older with a range of etiologies and are likely to have accumulated chronic liver damage over a number of years. It is likely that these HCC progress through more subtle genetic alterations such as CpG island methylation. The increasing prevalence of HCC in Australia in recent years reflects migration of people from Asia and other countries where HCC is relatively common.60 It remains to be shown if the genetic and epigenetic changes in the HCC from these patients are more like those from Australia or their country of origin.
Based on cited studies it is apparent that methylation is an early event, with evidence that this can precede the development of cirrhosis. In contrast, tumors displaying CIN can undergo malignant transformation in the absence of cirrhosis. In this subset of tumors, DNA hypomethylation of critical chromosomal regions may play a central role in hepatocarcinogenesis.
There are several mechanisms that can contribute to malignant transformation in the liver. A subset of HCC will be aneuploid with CIN while another group will be diploid, perhaps in the setting of frequent CpG island methylation. However, these phenomena have generally been studied in isolation and the relationships between CIN and CpG island methylation have not been assessed. Additionally, the studies of CpG island methylation have mainly focused on the relationship of this to defective DNA MMR function and microsatellite instability. Defining these broad mechanisms is a necessary prelude to determine the timing of events in malignant transformation of the liver and to investigate the role of known risk factors for HCC, such as cirrhosis, viral hepatitis, and male gender.