Hepatocellular carcinoma is the most common type of primary tumour of the liver. Worldwide, it is the fifth most common tumour type and the third most common cause of death from cancer. There are two basic incidence patterns: it is relatively rare in western countries but extremely common in sub-Saharan Africa and south Asia, with rates as high as 120 cases per 100,000 reported in some countries. In 30% to 70% of all cases, the tumour arises after chronic infection with the hepatitis C virus (HCV). Two independent studies from Japanese groups that shed light on the genetics of this condition and its complex relationship with HCV have now been published in the same issue of Nature Genetics.

The first study, led by Tatsuhiro Shibata from the National Cancer Center Research Institute in Tokyo, involved the high-resolution sequencing of the genome of a primary hepatocellular carcinoma and matched lymphocyte cells from the same individual [1]. The patient selected was a Japanese male positive for chronic hepatitis C infection. Using an Illumina GAIIx sequencer, the researchers obtained 36x coverage of the tumour genome and 28x coverage of the matched lymphocyte genome, covering more than 99.5% of the human reference genome sequence in each case. Approximately three million differences between the germline of this individual and the human reference genome were observed, along with 11,751 differences between the tumour and normal genomes, presumed to have been somatically acquired during tumour development.

Comparisons between the tumour and lymphocyte genomes showed that somatic substitutions were much more common in the intergenic regions than in the genic regions (including introns and non-coding exons), a phenomenon that may arise through either negative selection or DNA repair. C>T/G>A and T>C/A>G transitions were very significantly more common than other somatic substitutions, with the C>T substitution occurring most frequently at CpG sites. There were fewer T>C substitutions on the transcribed strand than on the untranscribed strand, suggesting that DNA repair occurs preferentially on the transcribed strand.

A total of ninety somatic substitutions were observed in protein-coding regions of the genome, including 63 that were validated to be non-synonymous. Seven of the 670 observed small insertions and deletions were also validated as located in protein-coding regions. Mutations in protein-coding regions were found to affect two tumour suppressor genes already known to be involved in hepatocellular carcinoma, TP53 and AXIN1; other genes that have been found to be mutated in other cancer types; and genes encoding phosphoproteins and proteins with bipartite nuclear localization signals. Twenty-two somatic rearrangements were validated using Sanger sequencing. Most of these were intra-chromosomal, with nine clustered in one small region of chromosome 11 (11q12.2-11q13.4). These generated four fusion transcripts with changes in transcription regulation or promoter activity.

The complete exome (the expressed part of the genome) of both the tumour and the normal cells was then sequenced at higher depth than the whole genome sequences (with approximately 75x coverage for each cell type). This sequencing detected 47 non-synonynmous substitutions, forty of which could be validated. One of these was a nonsense substitution in the gene TSC1 on chromosome 9. This gene encodes a protein that regulates the mammalian target of rapamycin signaling, an important target in oncogenesis. Although the substitution was found to occur in only a fairly small subpopulation of cancer cells, Shibata and his colleagues believe that TSC1 may be a useful therapeutic target for this tumour. They further conclude that the characteristic base substitution pattern observed in this case, which is different from that recorded for some smoking- and ultraviolet light-related cancers, is likely to be either organ specific or characteristic of HCV infection.

Recently, genome-wide association studies have led to significant progress in identifying genes associated with the risk of acute infection with HCV progressing to the chronic infection that greatly increases the risk of hepatocellular carcinoma. Much less, however, is known about genes associated with the progression of chronic hepatitis C to more serious liver disease and to cancer. The second of these two studies is a genome-wide association study, led by Koichi Matsuda from the University of Tokyo, into genetic susceptibility factors for HCV-induced hepatocellular carcinoma [2].

Matsuda and his colleagues obtained DNA from 721 Japanese individuals diagnosed with hepatocellular carcinoma and 2,890 HCV-negative controls from BioBank Japan and genotyped the samples at just under half a million locations. This first-stage analysis identified eight SNPs where association with progression from HCV infection to cancer could be suspected but not confirmed. Genotyping a further independent cohort of 673 cases and 2,596 controls at these eight positions revealed one SNP, labeled rs2596542, where the association was significant. Furthermore, genotyping another 1,730 individuals with chronic hepatitis C who had not gone on to develop serious liver disease at this position showed this SNP to be associated with progression to cancer but not with susceptibility to chronic HCV infection; the association also remained significant after adjusting for levels of alcohol consumption. No significant association was observed at any of the other seven positions.

The SNP rs2596542 is located in the highly complex and polymorphic class I major histocompatibility complex (MHC) region of chromosome 6; several positions in this region are already known to be associated with response to HCV infection. Matsuda and co-workers searched the whole of the MHC region for further SNPs that might be linked to hepatocarcinoma development, and identified a second, rs9275572, that had a moderate association. Other potentially interesting SNPs were associated with hepatitis therapy, which is not relevant to this study, or were known to be almost fixed in the Japanese population.

The most strongly associated SNP, rs2596542, is located very close to and upstream (5') of the MHC class I poplypeptide-related sequence A gene, MICA. This gene has two forms; one is membrane-bound and the other, generated by proteolysis at its transmembrane domain, is soluble in serum. The membrane-bound form is known to be necessary for activating the immune system to destroy virus-infected cells, and both forms have been found to be elevated in some cancers. Individuals with chronic hepatitis C who are homozygous for the cancer risk allele A at rs2596542 were shown to have lower levels of soluble MICA (sMICA). Furthermore, sMICA levels were raised in HCV-infected compared to non-infected individuals, but there was no significant rise on disease progression, indicating that MICA expression is induced by the stress of viral infection. As sMICA levels are known to correlate with those of the membrane-bound form, at least in wild-type MICA, Matsuda concluded that MICA may be under-expressed in both forms in individuals carrying A alleles at rs2596542, leading to an inadequate immune response to the virus and thus an increased risk of cancer development.

References

[1] Totoki, Y., Tatsuno, K., Yamamoto, S. and 18 others (2011). High-resolution characterization of a hepatocellular carcinoma genome. Nature Genetics, published online ahead of print 17 April 2011. doi:10.1038/ng.804

[2] Kumar, V., Kato, N., Urabe, Y. and 15 others (2011). Genome-wide association study identifies a susceptibility locus for HCV-induced hepatocellular carcinoma. Nature Genetics, published online ahead of print 17 April 2011. doi:10.1038/ng.809