Hepatocellular carcinoma (HCC) is one of the leading cancers in the world. More than half of a million people suffer from this cancer annually.1 HCC usually developed after years of persistent chronic hepatitis, elicited either by hepatitis B virus (HBV) or hepatitis C virus (HCV) infections.2 An intriguing characteristic of HCC is the gender difference. Male patients dominate over female patients in most areas with the ratio around 2–7 to 1,3, 4 which is noted more evident in HBV-related HCCs.5, 6 Consistently, in the murine HCC models, males are also more susceptible to both chemically and viral protein (e.g. the HBx) induced liver cancers.7 Many lines of evidence have supported that the male hormones (androgens) play important roles in determining this gender preference of HCC. Use of androgenic steroids was associated with an increased risk of HCC;8, 9 castration or the anti-androgen treatment protects male rodents from tumor development.10, 11 Such testosterone-induced murine hepatocarcinogenesis was further demonstrated to be androgen receptor (AR) dependent by using the Tfm (testicular feminization) mice, which lack the functional AR protein.12 In human studies, the higher testosterone levels and several genetic polymorphisms attributing to the higher androgen level or higher AR transcriptional activity have been shown to be significantly associated with increased risk of HCC in male HBsAg carriers.13, 14 These results all suggested the critical role of androgen signaling pathway in liver carcinogenesis.
The biological actions of androgen are mediated through the receptor AR protein, which functions as a transcriptional factor to modulate the transcription of androgen-responsive genes in a ligand dependent manner.15 The AR gene contains 2 polymorphic trinucleotide repeats (TNR) in the first exon, (CAG)n and (GGC)n, which encode tracts of polyglutamine (polyQ) and polyglycine (polyG) with variable length among different individuals. It was well demonstrated that AR alleles with shorter (CAG)n possess higher transactivation potency.16, 17 The shorter (GGC)n was also reported to enhance the androgen action by increasing the yields of AR protein.18 In a prospective case–control cohort study in male HBsAg(+) patients, it has been previously demonstrated that male patients with shorter AR (CAG)n alleles carried increased risk of HCC,13, 14 which was in line supporting the role of higher activity of AR in promoting the hepatocarcinogenesis of HBsAg(+) males.
The correlation between enhanced activity of AR axis and higher risk of HCC reminds a similar one in prostate cancer. The shorter length of both AR TNRs was reported to be associated with the higher carcinogenic risk of prostate cancers.19, 20, 21, 22, 23, 24, 25 Moreover, a variety of aberrations resulting in the increase of AR activity have been identified in prostate cancers. These included AR amplification or up-regulation,26 the ligand-independent AR activation by growth factors or cytokines,27, 28, 29 and notably the somatic mutations that enhance AR sensitivity to androgens or to other steroids.30 In total, the somatic mutations of AR gene were identified in 2–30% of prostate cancers according to their clinical stages.31, 32, 33
In a significant proportion of HCC patients, the expression and activation of AR was increased in the tumorous tissues and even in the nontumorous liver tissues adjacent to HCC.34, 35 In the current study, we tried to further explore the role of AR axis in hepatocarcinogenesis by investigating the somatic mutations of AR in liver cancer and to evaluate their association with the disease progression. For this purpose, we conducted the sequence analysis of AR gene and also investigated the change of repeat number of the 2 AR TNRs in HCCs and some dysplatic nodules. The results discovered few mis-sense mutations and a more common somatic changes at the TNRs of AR gene in male HCCs. Although there was no significant correlation between the somatic mutations of AR and the disease progression, the TNR length brought certain clinical impact on HCC patients.
Material and methods
Study subjects, DNA extraction and determination of serum markers
From 1985 to 2003, primary HCC and the corresponding nontumorous tissues from 257 Taiwanese HCC patients were collected from National Taiwan University Hospital. The clinical and demographic characteristics of these patients are recorded. The resected surgical specimens were quickly frozen in liquid nitrogen until DNA extraction. The DNA was extracted from the frozen tissues by following the protocols described previously.36
To collect the tissues from cirrhotic nodules (CN) and the corresponding tumorous samples, another 11 male patients with liver cirrhosis showing distinctively fibrotic CN were chosen and processed for microdissection. We followed the detailed procedures for tissue section, staining, microdissection, and DNA extraction as described previously.36 For each patient, we have tried to collect liver tissues representing different carcinogenic stages, including (i) the nontumorous tissues (normal but maybe with hepatitis); (ii) the regenerating nodules (RN); (iii) the dysplastic nodules (DN), either the low-grade or the high-grade dysplastic nodules (LDN or HDN); or (iv) the HCC tissues. The histologic grade of dysplasia was determined by 2 pathologists independently, using the criteria standardized by the International Working Party.37
The Institutional Review Board of National Taiwan University Hospital approved the use of these archived tissues. The serum samples of these patients were collected for determination of the HBV and HCV infections by assaying both HBsAg and anti-HCV Ab (AxSYM HBsAg version 2 and AxSYM HCV version 3, Abbot Laboratories, Biesbaden, Germany). For the patients negative of HBsAg, we have further confirmed the HBV infection status by PCR of the tissue DNA with primers at HBV X and S genes.38
Direct sequencing analysis
AR gene contains 8 exons with the coding region starting from exon 1 and ending at exon 8.20 We designed 9 amplicons for these exons with the primer sequences and annealing temperatures summarized in Table I. PCR amplification was performed by Biometra T3 Thermocycler (Biometra, Whatman Corp., Gottingen, Germany) with 30 ng genomic DNA, 30 ng per primer, PCR buffer of 20 mM Tris, 50 mM KCl, 1.66 mM MgCl2, 200 μM dNTP, and 0.4 U Platinum Taq polymerase (Invitrogen, Carlsbad, CA) in a total of 15 μl mixture. The following PCR cycling conditions were used: 95°C 5 min; 35 cycles of denaturing temperature 95°C 40sec, annealing temperature (see Table IA) 40 sec, and extension temperature 72°C 60 sec, which is then followed by postextension 72°C 10 min.
Table I. Primer Sequences for the Amplicons used for (A) PCR Amplification and Sequencing Analysis of AR Gene; (B) Genotyping and Genescan Analysis
Forward primer (5′-3′)
Reverse primer (5′-3′)
The PCR products were processed to the direct sequencing analysis by using ABI PRISM Big-dye kits (Applied Biosystems, Foster City, CA) and analyzed via ABI 3100 Genetics Analyzer (Applied Biosystems, Foster City, CA).
Genotyping and genescan analysis for TNR determination
To analyze the CAG and GGC TNRs of AR gene in HCCs, we have designed 2 amplicons covering both repeats for the Genescan analysis (Table IB). Both forward primers were FAM-fluorescence dye-labeled (Applied Biosystems, Foster City, CA) and the PCR reactions were performed as described previously.38 In addition, to confirm the correct nontumorous/tumorus tissue pairing for certain patients, we conducted genotying by using 5 highly polymorphic microsatellite markers with procedures as described previously,39 including D4S1534, D4S414, D4S1532, D4S1578, and D4S1572 (Table IB).
The PCR fragments were then analyzed following the manufacturer's instructions (Applied Biosystems). We used the Genscan (version 2.5; Applied Biosystems) and the Genotyper software (version 3.1; Applied Biosystems) for analyzing the fluorescence labeled peaks.
Determination of genomic MSI
We have followed the criteria established by National Cancer Institute (NCI) for determining the microsatellite instability (MSI) status in cancer samples.40, 41 PCR amplification was performed by using a panel of 5 microsatellite markers recommended by the NCI, including BAT25, BAT26, D5S346, D2S123, and D17S250.40, 41 All the primers were fluorescence-labeled and their primer sequences and PCR amplification conditions are listed in Table IB. Based on the genotyping results, tumors can be classified as high-frequency MSI (MSI-H) if 2 or more of the 5 markers showed instability, and as low-frequency MSI (MSI-L) if only 1 of the 5 markers showed instability.40
Distribution of demographic and clinical characteristics of study subjects were represented in numbers and percentage. The association of these variables with TNR somatic mutations or the TNR lengths was evaluated by Fisher's exact test. p values less than 0.05 were considered statistically significant. Each subject's unique national identification number was used to link with the computerized national Death Record profiles in Taiwan to identify those who died during follow-up period. Person–years for each participant were calculated from the date of operation to the date of death, or December 31, 2004 for those who were still alive at the end of follow-up. Overall survival was estimated using the Kaplan–Meier methods and the survival curves between different categories of repeat numbers of CAG and GGC were compared by log-rank test. All above analyses were performed by Stata statistical software (version 8.0, Stata Corp., College Station, Texas).
Nucleotide mutations of AR gene were rare in primary HCCs
To examine the presence of any genetic mutations of AR gene in liver cancers, we sequenced the AR gene in 257 HCCs, including 179 male HCCs and 78 female HCCs. The 9 amplicons covering the entire coding region of AR gene were PCR-amplified individually for direct sequencing analysis (Table I). By comparing the sequence results with that the Androgen Receptor Gene Mutations Database (ARDB) used (NM_000044.2),20 we have first identified nucleotide substitution type of mutations in 7 primary HCCs. Six of them occurred at exon 1 and one at exon 8 (Table IIA). For comparison, we also sequenced the DNA extracted from the nontumorous liver tissues adjacent to these 7 HCCs. In 3 cases, the mutations were only present in the HCC but not in the surrounding nontumorous tissues (Table IIA and Fig. 1), supporting that they are true somatic mutations. In Patient 22, the T to A transversion occurred in nt 1,285 and changed the amino acid (codon 57) from Leu to Gln. Interestingly, the same somatic mutation was previously identified in a prostate cancer and in a laryngeal cancer tissue.20 In Patient 75, a G to T transversion occurred at nt 1,385 and changed the amino acid (codon 90) from Gln to His. In Patient 230, a C to T transition occurred at nt 2,558, however, which did not result in amino acid change (codon 481, Tyr to Tyr). The 2 somatic AR mutations resulting in amino acid changes all occurred in the male patients (Table IIA).
Table II. Genetic Mutations in the Exons of AR Gene in Primary HCCs. (A) Nucleotide Substitution Type of Mutations and (B) TNR Somatic Changes in 257 Primary HCCs. (C) The TNR Changes Identified in Different Stages of Liver Tissues from 11 Male Cirrhosis Patients
The other 4 nucleotide changes identified in HCC tissues can also be found in the corresponding surrounding nontumorous tissues (Table IIA). They were thus likely germ line mutations or the rare SNPs, although none of them were identified as known SNP so far in the SNP database of dbSNP (BUILD 121). Since the distant tissues (e.g. blood samples) for these patients were not available, we could not conclude whether these 4 mutations were inherited or somatic. Therefore, these 4 mutations will not be included for all the subsequent analysis in this study.
More frequent somatic changes at (CAG)n or (GGC)n of AR gene were identified in male HCCs
During analysis of the sequence data, we noted an unusual sequence pattern of AR exon 1 from 4 male HCCs, showing a homogeneous sequence pattern before the (CAG) or (GGC) repeat but a mixed pattern after the TNRs, in Patients 6 and 12 for (CAG)n and in Patients 9 and 44 for (GGC)n, respectively. The representative sequence results from the tumor DNA of Patient 6 and 9 were shown in Figure 2a (T, tumor part). Because AR locates on X chromosome, males have only 1 AR allele with the same TNR pattern in all somatic cells and should show a mono and homogenous sequence pattern. The unusual mixed sequence pattern of these 4 male HCCs were only found after the repeats and thus a change occurs at TNR in their AR gene.
To serve as a control, we sequenced the AR gene from corresponding adjacent nontumorous tissues of these 4 male patients. All of them showed homogeneous sequence patterns as predicted, with the representative data for Patients 6 and 9 shown in Figure 2a (NT, nontumor part). The complicated sequence pattern thus only occurred in their tumorous tissues, implicating the change of AR TNRs might happen during their malignant transformation.
To further explore what kind of changes occur at the TNRs, we first conducted the cloning and sequencing analysis with the PCR fragments amplified from these 4 patients. Two populations of clones were recovered from each HCC sample. One population of clones showed the same sequence with the clones from corresponding nontumor samples. However, there existed another population of clones showing different repeat number of the TNRs and no other sequence aberrations were identified. The results suggested a novel type of the AR somatic changes in male HCC with changes of the TNR repeat numbers. To further confirm this result, we have tried a more direct way to determine the number of TNRs in the paired tumor/nontumor DNA samples using the genotyping/genescan analysis. By PCR with specific primers flanking the (CAG)n or (GGC)n and the subsequent genescan analysis, the alleles containing different repeat number of TNRs can be clearly distinguished. Compared to the nontumorous tissues, in which only 1 allele was detected as predicted, the DNA from these 4 HCCs all showed another allele with different size, which are consistent with the results from cloning experiments. The typical results of genescan analysis were shown in Figure 2b with Patient 6 for (CAG)n changes and Patient 9 for (GGC)n changes. In the tumor part, Patient 6 showed an additional allele with a contraction of 6 CAG repeats and Patient 9 showed an additional allele with a contraction of 3 GGC repeats (Fig. 2b).
Because the initial mutation screening only included the tumor DNA for the sequencing analysis, detection of this novel type of TNR changes might be missed for the cases whose original allele (in the nontumor part) is significantly diminished and replaced by another allele with TNR changes during malignant transformation. Aiding by the simple and efficient genotyping analysis, we decided to re-examine this novel type of TNR changes in these 257 patients by including DNA samples both from the nontumorous and tumorous tissues for side-by-side comparison. Intriguingly, such approach indeed help to identify another 12 cases with this type of somatic TNR changes. Nine cases showed changes of the repeat number at (CAG)n and 3 cases showed changes at (GGC)n. We summarized these TNR changes in Table IIB. For the total 11 cases with (CAG)n changes, 8 cases showed contraction pattern and 3 cases showed expansion pattern. For the total 5 cases changing the repeat number at (GGC)n, 4 cases showed contraction pattern and 1 case showed expansion pattern. Notably, the somatic changes of AR TNRs all occurred in the male patients (Table IIB).
TNR changes of AR occur in the precancerous DN
As the AR triplet repeats changed in some HCCs, it will be interesting to ask if such mutations took place in precancerous tissues. As CN are well-known precancerous lesions for HCC and around 50% of them already undergo clonal expansion,36 we therefore tried to examine the AR TNR changes in the individual CN. In 11 male patients with liver cirrhosis, we performed microdissection to separate individual nodules ranging from RN, LDN or HDN, or HCC.
The results of genotying analysis identified (CAG) repeat changes in 3 out of 11 patients (Table IIC). Three specimens were collected from Patient no. 284, including nontumorous tissue, LDN and HDN. A contraction of 8 CAG triplets was identified in the HDN, but not in the LDN. For Patient no. 286, 3 specimens representing hepatitis tissue, RN, and HDN were collected. A contraction of 5 CAG triplets was identified only in the HDN. For these 2 patients, no HCC samples were available for analysis. For Patient no. 290, 3 specimens representing hepatitis tissue, LDN and HCC were collected. A contraction of 3 CAG triplets was identified both in LDN and HCC. In this patient, only 1 allele with 24 CAG repeat was identified in the hepatitis tissue; 2 alleles with 24 and 21 repeats were identified in LDN, and only 1 allele with 21 repeats was present in HCC. These results indicated that (CAG)n changes might occur as early as the stage of LDN or HDN during the process of malignant transformation.
Most TNR changes of AR in HCC are not associated with conventional MSI phenotype
We next tried to exclude the possibility that the somatic changes of TNRs identified in the current study were artifacts resulted from mispairing or mixing (or contaminating) with different DNA samples. For this purpose, 5 highly polymorphic microsatellite markers on chromosome 4q (with heterozygosity >70%), including D4S1534, D4S414, D4S1532, D4S1578, and D4S1572, were selected for clonality DNA fingerprinting analysis of these DNA samples.39 The results showed identical allele patterns for the nontumorous and tumorous DNA from each patient and thus excluded the possibility of sample mispairing or contaminating/mixing.
The change of dinucleotide or trinuceotide repeats have been reported in some other human cancers, as a consequence of MSI caused by the aberration of DNA mismatch repair systems.42 To investigate if the TNR changes of AR in these HCCs were associated with the MSI phenotype, we have followed the criteria established by NCI for examining the MSI phenotypes in these 19 HCC samples with TNR mutations.41 By genotyping with a panel of 5 microsatellite markers as suggested by NCI, only 3 cases (15.8%) showed the phenotype of low-frequency MSI (MSI-L) (marked as asterisks in Table II).40 The other 16 samples did not reveal any MSI instability patterns and thus their somatic TNR changes were unrelated with the MSI phenotype.
Correlation of TNR changes with clinical features
To evaluate if any clinical features might correlate with the TNR changes, as this might reflect a unique HCC subgroup, we compared several clinicopathological features in these patients (Table IIIA). To focus on the contraction type, which has been reported associated with higher AR transcriptional activity, the results first showed significant preference in male patients (p = 0.007, Table IIIA). We next focused on the male patients for all the subsequent association analysis. The results did not discover any association of TNR contraction with these clinical parameters. Although the percentage of TNR contractions seemed to be more prevalent in tumors at Stage III + IV compared with the percentage for tumors at Stage I + II (10% compared to 4%, Table IIIA, panel A), the sample size is too small for revealing the statistic significance. There is no difference when all the somatic mutations were included in the same analysis (data not shown).
Table III. Clinical and Demographic characteristics of HCC Patients and their Relations to (A) The Somatic TNR Contractions; (B) (CAG)n Stratified by Repeat Number of 20; and (C) (GGC)n Stratified by Repeat Number of 16
P't no. (%)
(A) TNR contraction
Only the patients with the specified clinical data available were included for the analysis. Except for the analysis for the gender factor in panel A (included all patients), only male patients were selected for all the other analysis.
We next examined the correlation of TNR numbers with these clinicopathological parameters; we have focused on the 190 male patients and stratified the patients with the CAG and GGC numbers with the cut-off values at 20 for (CAG)n and 16 for (GGC)n (Table IIIA, panels B and C). Based on the previous association studies in prostate cancers, the repeat number fewer than the cut off values set here were attributed to higher carcinogenic risk.21, 22, 23, 24, 25 The only significant association comes from the (CAG)n in relation with the HBV infection status. The percentage of CAG repeat number ≤20 was noted to be significantly higher in HBV(+) HCCs, compared with those of HBV(−) HCCs (Table IIIA, panel B, p = 0.011). When we further stratified the male patients into HBV(+) and HBV(−) groups and compared the average repeat number of (CAG)n and (GGC)n in these 2 groups, we found that the average CAG repeat number in HBV(+) HCC is significantly shorter than that of HBV(−) HCCs (HBV(+) HCCs, mean = 20.29, SD = 2.83; HBV(−) HCCs, mean = 22.84, SD = 3.94; p = 0.0032). In contrast, no significant difference of GGC repeat number was noted between these 2 groups (HBV(+) HCCs, mean = 16.66, SD = 1.35; HBV(−) HCCs, mean = 16.54, SD = 1.53; p = 0.581).
Finally, we examined the correlation of TNR repeat number with the overall survival after surgery. Figure 3 presents the Kaplan–Meier estimates of overall survival partitioned according to the (CAG)n and (GGC)n genotype groups, ≤20 repeats vs. >20 repeats for (CAG)n and ≤16 repeats vs. >16 repeats for (GGC)n. Log-rank tests indicated that the (GGC)n classification results in significant differences in the curve of overall survival (p = 0.047, Fig. 3b). The longer alleles (i.e., those with 17 or more repeats) were significantly associated with the worse overall survival. However, no significant association was found between (CAG)n and the overall survival (p = 0.767, Fig. 3a).
Androgen axis has long been proposed associated with hepatocarcinogenesis and attributed to the male dominance of HCCs. It was interesting to search for any AR mutation in HCC, as in the case of prostate cancers. In our analysis of the genetic aberrations of AR gene in total 268 HCC and cirrhotic patients, 22 somatic mutations of AR have been identified. Among them, 21 mutations will lead to amino acid change and all occur in male patients, 2 belong to the missense mutations, and 19 belong to the TNR somatic changes. The results indicated that somatic mutations taking place in 11.05% of male HCCs (21 out of 190 cases) mainly belong to the type of TNR changes.
Currently, more than 600 AR mutations have been documented and most of them belong to the nucleotide substitution type of mutations.20 In HCC, this type of somatic mutations did not occur often. Instead, the somatic changes of (CAG)n and (GGC)n seem more common. Somatic contraction of (CAG)n has been occasionally identified in a few colon cancers and prostate cancers and all the changes identified so far belonged to the contraction type.43, 44, 45, 46 Our current study found that (CAG) repeat changes also occur in male HCCs. The mutation could start in the precancerous lesions or HCC and the majority also belonged to the contraction type (in 11 out of 14 cases). For the (GGC) repeat, so far no somatic mutations have been documented in the literature. This type of (GGC)n changes thus belonged to a novel type of genetic aberrations first identified in HCCs. Again, most (GGC)n aberrations also belonged to the contraction type (in 4 out of 5 cases).
Regarding the functional significance of the somatic changes of AR TNR repeat numbers, previous studies already demonstrated an inverse correlation between the length of TNR, both for (CAG)n and (GGC)n, and the transactivation activity of AR.16, 17, 18 According to the finding that more potent AR activity was correlated with a higher risk for HBV-related HCC,13, 14 the somatic AR changes identified in the resulting HCCs might attribute to the higher AR activity, and thus are positively selected out during the carcinogenic process. Our results therefore might provide another line of genetic evidence supporting the involvement of AR axis in male hepatocarcinogenesis, although the extent to which these AR changes have enhanced activity contributing to the carcinogenic process warrants further investigation. The AR somatic mutations identified in the current study all locate at the N-terminal AF-1 domain with the function associated with transactivation, emphasizing the role of enhanced AR transcriptional activity in male hepatocarcinogenesis.
In several population-based case–control association studies from various ethnic populations, shorter alleles of (CAG)n (≤20 repeats) and (GGC)n (≤16 repeats) in AR gene (germ-line) were shown to be associated with an increased risk to develop the prostate cancers.21, 22, 23, 24, 25 Our prospective case–control study in the cohort of HBV(+) males also pointed out the shorter (CAG)n (≤20 repeats) to be associated with the higher HCC risk.13 The current study indeed identified the somatic AR changes with (CAG)n contraction to <20 repeats in 4 HCC and the (GGC)n contraction to <16 repeats in another 4 HCCs, which might render the hepatocytes to increased activity of AR axis during the carcinogeneic process.
To study the TNR length for their association with the clinical parameters, we have stratified all the male patients by the cutoff values of TNR numbers mentioned earlier. The results pointed out a significant correlation between shorter (CAG)n and the HBV(+) HCCs with the percentage of CAG repeat number ≤20 to be significantly higher in HBV(+) HCCs compared with those of HBV(−) HCCs (Table IIIB, p = 0.011). This correlation was also reflected by the results that the average CAG repeat number in HBV(+) HCC is significantly shorter than that of HBV(−) HCCs (p = 0.0032). Although our previous study already pointed out (CAG)n ≤ 20 to be associated with higher HBV related HCC risk, which was only conduced in the cohort of HBV(+) males.13, 14 The comparison of AR TNR repeat number stratified by different viral etiology has not been reported so far in any previous studies. The current results indicated that shorter (CAG)n might specifically associate the HBV viral infection in hepatocarcinogenesis. A possible synergistic carcinogenic effect specifically exists between higher AR activity, which were attributed by the shorter AR (CAG)n, and the HBV infection in male HCC was thus implicated and worthy to be investigated.
When we analyzed the repeat number of (CAG)n and (GGC)n with the survival probability, we found a significant association of (GGC)n ≤ 16 to be associated with better overall survival (Fig. 3b, p = 0.047). Intriguingly, similar results have also been reported in prostate cancers, in which (GGC)n ≤ 16 were associated with higher carcinogenic risk but with better overall survival.25, 47 Although the significance of (GGC)n in HCC has not been investigated so far, the short (GGC)n was noted to be associated with higher risk of prostate cancers but might identify a subpopulation of patients with clinically localized or less aggressive prostate cancers.25, 47 This intriguing implication in prostate cancers might also occur in male liver cancers, explaining the finding with dominantly somatic shortening in HCC but association of shorter (GGC)n with better overall survival.
The genetic mechanisms for TNR mutation of AR gene are not yet clear. The repetitive character of TNR allows stretches of such tracts to form slipped–stranded structures or more complex configurations, which allows the TNR changes to occur with defect of DNA repair system and it usually results the phenotype of MSI. By the criteria defined by NCI, the MSI-L phenotype was only detected in 3 out of the 19 cases showing TNR changes, suggesting most TNR changes not associated with the MSI phenotype. It was consistent with the findings that MSI phenotype seldom occurs in HCCs.48, 49 However, there still exists the possibility that such changes belonged to nonspecific genetic changes resulted by the genomic instability, which cannot be detected by the panel of markers suggested by the NCI. In that case, both contraction and expansion types of TNR changes will presumably occur with approximately equal opportunity in the resulting HCCs. Since current study found more contraction changes of TNRs in HCCs, the TNR somatic changes were thus more likely a result of positive selection for the tumorigenic advantage attributed by TNR contractions. The possibility that some other unidentified potential mechanisms, either associate with the viral factors or not, might contribute to the TNR contraction in HCC awaits to be studied.
The finding that TNR changes were only identified in male HCCs suggested its predominance attributing to male carcinogenesis. A plausible different carcinogenic mechanisms underlying male and female HCCs was thus implicated. In-line to support this, our previous studies in female HCCs, both by the cohort association analysis and by analyzing the tumorous tissues, showed longer CAG repeat to be more significantly associated with the higher HCC risk (also in HBV(+) HCCs), which was in contrast to the case of male HCCs.38, 50 Although the underlying mechanism is still unknown, the current study again emphasized the plausible different carcinogenic mechanisms underlying the gender specific hepatocarcinogenic process.
We sincerely thank Dr. Jung-Ta Chen for providing the microdissected materials for the experiments.