Potential conflict of interest: Nothing to report.
Hepatitis C virus (HCV) infection is a major risk factor for developing hepatocellular carcinoma (HCC). The host genetic factors that are involved in the development of HCC in patients with HCV infection remain to be investigated. To search for single nucleotide polymorphisms (SNPs) in HCC susceptibility genes, 393 SNPs in 171 candidate genes were examined in 188 Japanese patients with chronic HCV infection, including 77 patients with HCC. HCC-related SNPs were then examined in another 188 patients (including 93 patients with HCC) with chronic HCV infection. Haplotype analyses of HCC-related genes were performed in a total of 376 patients. Of the 393 SNPs, 31 SNPs in 29 genes were significantly associated with HCC based on an initial screening (P < .05). Of these 31 SNPs, 3 SNPs of 3 genes (SCYB14, GFRA1, and CRHR2) were significantly associated with HCC in a secondary screening. Haplotype analyses of these 3 genes identified 2 haplotype blocks associated with HCC. In conclusion, these SNPs and haplotypes located in the SCBY14, CRHR2, and GFRA1 genes will be used as markers to identify a subgroup of Japanese patients with chronic HCV infection who are at high risk of developing HCC. Supplementary material for this article can be found on the HEPATOLOGY website (http://www.interscience.wiley.com/jpages/0270-9139/suppmat/index.html). (HEPATOLOGY 2005;42:846–853.)
More than 170 million people worldwide are estimated to have chronic hepatitis C virus (HCV) infection (http://www.who.int/inf-fs/en/fact164.html). The most important sequelae of chronic HCV infection are progressive liver fibrosis leading to cirrhosis, and hepatocellular carcinoma (HCC), which is responsible for significant morbidity and mortality throughout the world.1–4 Many factors, such as alcohol intake, older age at time of infection, male sex, and co-infection with hepatitis B virus, are reported to accelerate disease progression in HCV infection.5–8 In addition, host genetic factors have been reported to affect the risk of developing HCC.9–15
Recently, we reported that genetic polymorphisms in interleukin-1β11 and uridine 5′-diphosphate-glucuronosyltransferase 1A712 are associated with the development of HCC in Japanese patients with chronic HCV infection. Genetic polymorphisms in CYP enzymes,13 the microsomal epoxide hydrolase gene,14 and the aldehyde dehydrogenase 2 gene15 also have been reported to be associated with HCC and the severity of HCV-related liver disease. The number of candidate genes that have been examined is, however, rather limited.
We performed a large-scale candidate-gene–based search of single-nucleotide polymorphisms (SNPs) to look for SNPs in genes associated with HCC susceptibility. A total of 393 SNPs in 171 candidate genes were examined in Japanese patients with chronic HCV infection.
HCV, hepatitis C virus; HCC, hepatocellular carcinoma; SNPs, single nucleotide polymorphisms; ALT, alanine aminotransferase; AFP, alpha fetoprotein; SCYB14, small inducible cytokine B14 precursor; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GFRA1, GDNF family receptor alpha 1; CRHR2, corticotropin-releasing hormone receptor 2.
Patients and Methods
We studied 376 consecutive Japanese patients with chronic HCV infection who consulted the outpatient clinic of the University of Tokyo Hospital between August 2001 and June 2003 (208 men and 168 women, 22–84 years old, a median age of 62.5 years, 170 with HCC and 206 without HCC). The genomic DNA of these patients was made available after obtaining written informed consent for genotyping. The first 188 patients were enrolled in the initial screening for SNPs in HCC susceptibility genes. The second 188 patients were enrolled in the secondary screening. Approval was obtained from the institutional ethics committee, and all the procedures followed institutional guidelines.16
Patients selected for the study were those who tested positive for HCV antibody by the second-generation enzyme immunoassay (Ortho Diagnostics, Tokyo, Japan); HCV RNA was measured using the Amplicor HCV assay version 1 (Roche, Tokyo, Japan). All patients were negative for the hepatitis B surface antigen (Abbott Laboratories, North Chicago, IL). HCV genotypes were determined by using a genotyping assay (SRL Laboratory Co., Tokyo, Japan). Any patient with an ethanol intake of 80 g/d or above for longer than 10 years was considered to have a positive history of alcohol abuse. The following clinical parameters were obtained for each patient at the time of whole-blood collection: age, sex, serum albumin level, serum total bilirubin level, serum alanine aminotransferase (ALT) level, serum alpha fetoprotein (AFP) level, prothrombin time, platelet count, and serum viral load measured using the Amplicor-HCV monitor assay. The diagnosis of HCC was made by several imaging methods (ultrasonography, computed tomography, arteriography, or magnetic resonance imaging) and confirmed histologically by sonography-guided fine-needle biopsy specimens in all 170 patients.17 All patients were shown not to have other cancers by an initial screening examination.
Genomic DNA was extracted from 100 μL whole blood using the SepaGene kit (Sanko Junyaku, Tokyo, Japan) according to the manufacturer's instructions. Extracted DNA was dissolved in 20 μL Tris-HCl buffer (10 mmol/L, pH 8.0) containing 1 mmol/L EDTA and was stored at −30°C until use.
The genes and SNPs examined are shown in Supplementary Table 1 . These genes were selected as possibly playing a role in hepatocarcinogenesis by modifying cell growth, hepatic inflammation, and hepatocyte apoptosis; they include growth factors, growth factor receptors, cytokines/chemokines, cytokine/chemokine receptors, genes related to apoptosis, genes related to interferon signals, and CD81 (a putative HCV receptor gene). SNPs of the selected genes were extracted from the JSNP database (http://snp.ims.u-tokyo.ac.jp),18 a database for SNPs found in the Japanese population, or were identified using TaqMan Assays-on-Demand SNP Genotyping Products (Applied Biosystems, Foster City, CA). When more than 1 SNP was listed in the database, several SNPs located 10 to 30 kb apart from each other were chosen. Altogether, 393 SNPs derived from 171 genes were tested for an association with HCC in patients with chronic HCV infection. The genotyping was conducted with a fluorogenic polymerase chain reaction.19, 20 The alleles and genotype frequencies of the SNPs were determined and combined with the clinical data to conduct statistical analyses.
Clinical parameters were evaluated using the two-tailed t test, the Mann-Whitney U test, and the chi-square test to determine their associations with the presence of HCC. The association between different genotypes and the presence of HCC was evaluated using the chi-squared test. For all tests, a P value of less than .05 was considered significant. Possible confounding effects among the variables were adjusted using a multivariate logistic regression model, and odds ratios and 95% confidence intervals were calculated. All data analyses were performed using SPSS v. 12.0 (SPSS Inc., Chicago, IL). The Hardy-Weinberg equilibrium of alleles at individual loci was evaluated using HWE (ftp://linkage.rockefeller.edu/software).
The haplotype-based analyses consisted of 3 steps: (1) haplotype block partitioning, (2) haplotype reconstruction, and (3) haplotype-based association tests. In the first step, a set of consecutive SNPs was defined as a haplotype block if the P value derived from the logarithmic odds score was less than .01 for every combination of the SNPs. In the second step, for each haplotype block, haplotype frequencies were estimated from the genotypes of unrelated individuals, and the posterior probability distribution of the diplotype configuration for each subject was determined using the LDSUPPORT program,21 which is based on the expectation-maximization (EM) algorithm. In the third step, an m × 2 (haplotypes × with/without HCC) contingency table was built for each haplotype block by counting haplotypes whose posterior probability was higher than 0.8. The empirical P value of the association between the haplotype distribution and HCC presence based on likelihood ratio test statistics was then obtained with 20,000 permutations using the FASTEHPLUS T5 test.22P values of less than .05 were considered significant. To identify risk haplotypes, Fisher's exact test was performed on the 2 × 2 (risk/non-risk haplotypes × with/without HCC) contingency table for each haplotype in the significant blocks.
Quantification of Small Inducible Cytokine B14 Precursor mRNA Expression.
Small inducible cytokine B14 precursor (SCYB14) mRNA expression was quantified in the cancerous and noncancerous liver tissues of 3 patients (53-year-old man, 59-year-old man, and 57-year-old woman) with chronic HCV infection who had been admitted to the University of Tokyo Hospital for surgical treatment of HCC. The diagnosis of HCC was made using several imaging methods and was confirmed histologically. Written informed consent according to the guidelines of the Helsinki Declaration was obtained from each patient. Approval for the study was obtained from the institutional ethics committee.
Total RNA samples were extracted from cancerous and noncancerous liver tissues using the Isogen RNA Extraction kit (Nippon-Gene, Tokyo, Japan) according to the manufacturer's instructions. The quality of the total RNA was judged from the ratio of the 28S to 18S ribosomal RNA after agarose gel electrophoresis. Total RNA was reverse transcribed to cDNA using Taqman Reverse Transcription reagents (Applied Biosystems). SCYB14 cDNA was then quantified in triplicate using the Assay-on-Demand Gene Expression product (Applied Biosystems) with the ABI PRISM 7000 sequence detection system (Applied Biosystems). As an endogenous control, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA was quantified in a similar manner.
The characteristics of the group of 188 subjects of the initial screening study for the 393 SNPs are shown in Table 1. No significant difference was found in the proportion of male subjects, alcohol abuse, HCV genotype, viral load, or serum ALT level between the groups of patients with and without HCC. In the group of patients with HCC, the patient age, proportion of patients with cirrhosis, serum total bilirubin level, and serum AFP level were higher; and serum albumin level, prothrombin time, and platelet count were lower, than in the group of patients without HCC.
NOTE. Age, albumin, TB, ALT, AFP, PT, platelet count, and HCV load are shown as median (range). Male sex, alcohol > 80 g/d, and HCV serotype 1 are shown as frequency (percentage). Abbreviations: HCC, hepatocellular carcinoma; TB, total bilirubin; ALT, alanine aminotransferase; AFP, alpha fetoprotein; PT, prothrombin time; HCV, hepatitis C virus; IU, international unit.
P values were calculated using the χ2 test.
P values were calculated using the Mann-Whitney U test.
Of the 393 target SNPs, 390 were successfully identified in patients. Of these 390 SNPs in 171 genes, 31 SNPs in 29 genes were found to be associated with the presence of HCC (Table 2): the genotype frequencies of 11 SNPs (11 genes) and allele frequencies of 15 SNPs (15 genes) were significantly different between the groups of patients with and without HCC. The at-risk alleles of 27 SNPs (27 genes) were associated with the presence of HCC in a dominant or recessive fashion.
Table 2. SNPs Associated With the Presence of HCC
NOTE. SNPs significantly associated with the presence of HCC in both first and second screening are indicated in boldface.
The characteristics of the group of 188 subjects of the secondary screening study for the 31 SNPs identified in the primary screening are shown in Table 1. There was no significant difference in HCV genotype, viral load, or serum ALT level between the groups of patients with or without HCC. In the group of patients with HCC, patient age, the proportion of male patients, the proportion of patients with cirrhosis, serum total bilirubin level, and serum AFP level were higher, whereas serum albumin level, prothrombin time, and platelet count were lower than in the group of patients without HCC.
Of the 31 SNPs screened, only 3 SNPs in 3 genes were associated with HCC in both the initial and secondary screenings. These were: an intron SNP [hCV1250702 of Celera SNP (Celera Genomics, Rockville, MD), C__1250702_10 of Applied Biosystems] of GDNF family receptor alpha 1 (GFRA1) associated by genotype frequency and at-risk allele analyses; an intron SNP (rs2267716 of dbDNP [http://www.ncbi.nlm.nih.gov/SNP/index.html], IMS-JST021253 of JSNP, C__2570970_1_ of Applied Biosystems) of corticotropin-releasing hormone receptor 2 (CRHR2) associated by allele frequency analysis; and an intron SNP (rs2237062 of dbDNP, IMS-JST017563 of JSNP) of SCYB14 associated by allele frequency and at-risk allele analyses (Table 2).
Association of Polymorphisms of GFRA1, CRHR2, and SCYB14 with HCC in Patients With HCV.
The characteristics of all 376 subjects are shown in Table 1. There was no significant difference in alcohol abuse, HCV genotype, viral load, or serum ALT level between the groups of patients with and without HCC. In the group of patients with HCC, patient age, the proportion of male patients, the proportion of patients with cirrhosis, serum total bilirubin level, and serum AFP level were higher, whereas serum albumin level, prothrombin time, and platelet count were lower than in the group of patients without HCC.
The distributions of genotypes, alleles, and at-risk alleles with regard to the presence of HCC are shown in Table 3. The genotype frequencies and allele frequencies of GFRA1, CRHR2, and SCYB14 were significantly different between the groups of patients with and without HCC. The at-risk alleles of GFRA1 and SCYB14 were also associated with the presence of HCC.
Table 3. Association of SCYB14, GFRA1, and CRHR2 Polymorphisms With HCC
Patients With HCV
Odds Ratio (95% CI) With vs. Without
Without HCC n = 206
With HCC n = 170
SCYB14 at-risk allele
G/G + G/C
GFRA1 at-risk allele
C/C + G/C
Interestingly, a combination of 2 gene genotypes increased the odds ratio up to 8.23 (SCYB14 + GFRA1), 12.6 (SCYB14 + CRHR2), and 21.5 (GFRA1 + CRHR2), respectively. Moreover, a combination of 3 gene genotypes increased the odds ratio up to 12.3.
Factors Associated With the Presence of HCC in Patients With HCV.
The following factors were significantly associated with the presence of HCC according to univariate analyses: GFRA1 genotype (P < .001), CRHR2 genotype (P = .003), SCYB14 genotype (P = .002), age older than 60 years (P < .001), male sex (P < .001), the presence of cirrhosis (P < .001), platelet count <12.5 × 104/μL (P < .001), albumin <3.9 g/dL (P < .001), total bilirubin >0.7 mg/dL (P < .001), prothrombin time <70% (P < .001), and serum AFP >20 μg/L (P < .001). To evaluate the effects of the polymorphisms in 3 SNPs on the presence of HCC, a stepwise multivariate logistic regression analysis was performed using these 11 variables. Nine variables (GFRA1 genotype, CRHR2 genotype, SCYB14 genotype, age >60 years, male sex, albumin <3.9 mg/dL, total bilirubin >0.7 mg/dL, prothrombin time <70%, and AFP >20 μg/L) were included in the final model with odds ratios of 2.54 (G/G vs. G/C), 9.81 (A/A vs. G/G), 3.13 (C/C vs. G/G), 2.65, 2.27, 3.08, 2.05, 3.04, and 3.62, respectively (Table 4).
Table 4. Factors Associated With the Presence of HCC in Multivariate Analysis
Factor and Category
Abbreviations: TB, total bilirubin; PT, prothrombin time; AFP, alpha fetoprotein.
Age > 60 years
Albumin < 3.9 g/dL
TB > 0.7 mg/dL
PT < 70%
AFP > 20 ng/mL
Haplotype analyses of the 3 genes shown to be significant factors by the secondary screening (SCYB14, GFRA1, and CRHR2) were performed in a total of 376 patients by assaying 3 SNPs in SCYB14, 12 SNPs in GFRA1, and 3 SNPs in CRHR2. Five haplotype blocks were built in 3 genes, and 2 haplotype blocks in 2 genes (SCYB14 and GFRA1) were found to be significantly associated with HCC.
The haplotype block in SCYB14 consisted of 3 SNPs (rs1148364, rs2237062, and rs1016666 of dbSNP) (Fig. 1, Table 5). For 3 subjects of 376, the diplotype configuration could not be determined, and 6 haplotypes were missing. The overall P value generated by the FASTEHPLUS T5 test was .001. Haplotype CCT was a significant protective factor for HCC (P = .013, OR = 0.68, 95% CI = 0.49–0.92), and haplotype GGC was a significant risk factor for HCC (P < .001, OR = 1.82, 95% CI = 1.32–2.53).
Table 5. Haplotypes in Haplotype Block of SCYB14 (rs1148364-rs2237062-rs1016666)
Fisher's P Value
NOTE. Overall P value (FASTEHPLUS T5; 20,000 permutations): .0014.
The haplotype block in GFRA1 consisted of 3 SNPs (rs953920 of dbDNP, hCV1250702 of Celera SNP, and rs2270181 of dbSNP) (Fig. 2, Table 6). For 17 subjects of 376, the diplotype configuration could not be determined, and 34 haplotypes were missing. The overall P value generated by the FASTEHPLUS T5 test was .003. Haplotype GCG was a significant risk factor for HCC (P = .007, OR = 1.74, 95% CI = 1.18–2.58).
Table 6. Haplotypes in Haplotype Block of GFRA1 (rs953920-hCV1250702-rs2270181)
Fisher's P Value
NOTE. Overall P value (FASTEHPLUS T5, 20000 permutations): .0029.
SCYB14 mRNA Expression in Cancerous and Noncancerous Live Tissues.
SCYB14 cDNA was quantified and normalized using GAPDH cDNA as an endogenous expression control. The expression of SCYB14 mRNA in noncancerous liver tissue was significantly higher than that in cancerous liver tissue in all 3 cases (Fig. 3).
In this study, the presumptive genetic markers for susceptibility to hepatocarcinogenesis were identified in patients with chronic HCV infection. Three SNPs in 3 different genes were identified as being associated with HCC.
A dual-step screening process for HCC susceptibility genes was adopted to reduce the possibility of a problem with multiple testing.23 Although 31 SNPs were identified in the initial screening for HCC susceptibility SNPs, only 3 SNPs were finally shown to be associated with HCC after the secondary screening. The other 28 SNPs were removed, probably because they were falsely associated with HCC in the initial screening. The results should be interpreted cautiously, however, because of the possibility that falsely associated SNPs could have been picked up and SNPs that are actually associated with HCC could have been missed. A larger population should be tested for the positive SNPs to confirm the results of this work.
In addition, SNPs associated with HCC development may not be the causative variation of the disease but may be merely genetic markers that are in linkage disequilibrium with other causative variations. All of the 3 SNPs associated with HCC in this study were intron SNPs with functions yet to be elucidated. MicroRNAs recently have been reported to be translated from introns and regulate the mRNA translation of genes associated with development.24 Thus these 3 SNPs may have a regulatory function. To further determine whether these 3 SNPs are linked with other causative variations, we performed haplotype analyses of these 3 genes and identified 2 haplotype blocks associated with HCC.
Despite the limitations of this study, the fact that 3 genes have been identified as being associated with HCC will allow the generation of new hypotheses for future studies. The association of these 3 genes with HCC is a novel observation. SCYB14, alternatively known as CXCL14 (C-X-C motif chemokine ligand 14) or BRAK (breast and kidney-expressed chemokine), belongs to the cytokine gene family and encodes secreted proteins involved in immunoregulatory and inflammatory processes.
Interestingly, the expression of SCYB14 mRNA was high in noncancerous liver tissue but was very low in cancerous tissue in hepatitis C patients with HCC. SCYB14 is moderately or highly expressed in normal liver tissue but is expressed at very low levels in human HCC cell lines, including HepG2, HLE, Huh6, Huh7, and PLC/PRF/5, according to the RefExA reference database for gene expression analysis from the Laboratory for System Biology and Medicine at RCAST, the University of Tokyo (Tokyo, Japan) (http://www.lsbm.org/site_e/database/index.html). In fact, SCYB14 has been reported to be ubiquitously expressed in normal tissue extracts but is absent from a variety of tumor cell lines.25 Moreover, a previous study with human prostate epithelial cells showed that the expression of SCYB14 is upregulated during senescence, which creates a barrier inhibiting the acquisition of an immortal phenotype, but is downregulated in immortalized cells.26 These results suggest that SCYB14 may play an important role in carcinogenesis.
CRHR2 was 1 of the 3 genes associated with HCC in our study. CRHR2 mRNA is expressed in liver, although, according to RefExA, its expression level is comparatively low and is similar between normal liver tissue and HCC cell lines. A microarray analysis in our laboratory using the Human 3.8I Glass Array (Clontech Laboratories, Palo Alto, CA) showed that 21 of 27 liver tissue samples obtained by liver biopsy from patients with chronic hepatitis had positive signals for CRHR2 (unpublished data). Although mice deficient for CRHR2 display anxiety-like behavior, are hypersensitive to stress, and have impaired cardiovascular function, the function of CRHR2 in liver is still unknown.27, 28 Urocortin, a corticotropin-related peptide, is thought be the endogenous ligand for CRHR2. Interestingly, the intracisternal injection of urocortin has been reported to exacerbate acute liver injury through the sympathetic nervous system.29 In addition, urocortin has been shown to be expressed and to exert an anti-inflammatory effect in human gastric mucosa.30 These data suggest that urocortin may contribute to inflammation, which ultimately increases the risk for developing HCC.31
GFRA1 encodes a receptor for glial cell line–derived neutrophic factor (GDNF) and neuturin. GFRA1 is a candidate gene for Hirschsprung disease.32 Recently, it has been reported that a −193C to G sequence variant in the 5′-untranslated region of GFRA1 was found in 15% of cases with sporadic medullary thyroid cancer, suggesting an association of this variant with carcinogenesis.33
Despite the limitations of a cross-sectional study, our analyses showed a prominent effect of 3 gene polymorphisms on the risk of developing HCC. Because most HCV-related HCCs arise from a background of cirrhosis, these 3 SNPs might have association with cirrhosis. In fact, among 3 SNPs, GFRA1 genotype was also associated with the presence of cirrhosis (P = .002), although its association was weaker than that with HCC (P < .001).
In conclusion, given that many patients are referred to our hospital for the treatment of HCC, our study population may be biased toward patients with HCC. Our multivariate model, however, included most of the previously reported risk factors for HCC plus the polymorphisms of the 3 genes. This implies that our results can be generalized to the Japanese population. The uncertainty of the odds ratios, owing to the study design, should be resolved in a subsequent controlled trial. The genotype of these SNPs may serve as a marker that can be used to identify a subgroup of Japanese patients with chronic HCV infection who have higher risk of developing HCC.
The authors thank Yasuhiko Sugawara and Masatoshi Makuuchi for providing liver tissues. We also thank Mitsuko Tsubouchi for technical assistance and Mina Nagata for secretarial assistance.