Abstract
Estrogens have been proposed to act as tumor promoters and induce hepatocarcinogenesis. Recently, we observed a significant association between the risk for hepatocellular carcinoma and the polymorphisms of the estrogen receptor (ESR) α (ESR1) gene, supporting the hypothesis of involvement for the estrogen-ESR axis in the estrogen-induced hepatocarcinogenesis. In this study, based on another hypothesis in which estrogen metabolites can directly cause DNA damage and affect tumor initiation, we examined whether the polymorphisms of the estrogen-metabolizing enzymes (EME), which are involved in biogenesis (CYP17, CYP19), bioavailability (CYP1A1, CYP1B1), and degradation (catechol-O-methyltransferase) of the estrogens, have any bearing on the risk for hepatocellular carcinoma. Seven functional polymorphisms in five EMEs (CYP17 MspAI site, CYP19 Trp39Arg, Ile462Val and MspI site in CYP1A1, CYP1B1 Val432Leu, and Ala72Ser and Val158Met in catechol-O-methyltransferase) were genotyped in 434 patients with hepatocellular carcinoma and 480 controls by PCR-RFLP analysis. The associations between the polymorphisms and hepatocellular carcinoma risk were evaluated while controlling for confounding factors. No significant association with the risk for hepatocellular carcinoma was observed with the seven polymorphisms in hepatitis B virus carriers and non–hepatitis B virus carriers after correction for multiple comparisons. After stratification by common confounding factors of hepatocellular carcinoma, the EME polymorphism remained no significant association with the hepatocellular carcinoma risk. Furthermore, no signs of gene-gene interactions were observed for each combination of the seven polymorphisms. Our findings suggest that the polymorphisms of EMEs may not contribute significantly to the risk for hepatocellular carcinoma. (Cancer Epidemiol Biomarkers Prev 2008;17(12):3621–7)
Introduction
Animal models and human epidemiologic studies have suggested that estrogens act as tumor promoters and might induce hepatocarcinogenesis (1-4). The estrogens exert the effects by binding to their receptors [estrogen receptors (ESR)]. In a recent study, we have hypothesized that the genetic polymorphisms within ESRs could influence the effects of estrogens, which in turn results in genotype-dependent differences in risk for hepatocellular carcinoma. Indeed, the polymorphisms in the 5′ end of the ESR α (ESR1) gene have been shown to be associated with an increased hepatocellular carcinoma risk, supporting the hypothesis of involvement for the estrogen-ESR axis in the estrogen-induced hepatocarcinogenesis (5).
In the present study, we focused on another hypothesis. Several studies have shown that estrogen metabolites can bind to DNA and trigger damage, suggesting that the estrogens might be endogenous genotoxic agents that can directly cause genetic alteration and affect tumor initiation (6, 7). This possibility is supported by the finding that women with reduced amounts of the enzymes responsible for removing reactive estrogen metabolites are at higher risk of developing breast cancer (8). Based on the obvious relevance of estrogens in human hepatocarcinogenesis (3), it is reasonable to anticipate that the same events involved in hepatocellular carcinoma. We therefore hypothesize that the estrogen-metabolizing enzymes (EME), which are involved in the biogenesis, bioavailability, and degradation of estrogens, may also be the excellent biological candidate susceptibility genes for hepatocellular carcinoma. It is expected that the polymorphisms within the EMEs may contribute to interindividual differences of levels of estrogen metabolites and then potentially affect the risk for hepatocellular carcinoma.
Several functional polymorphisms of the EMEs, which are thought to affect the respective EME activity in an allele-specific manner, have been well characterized. Cytochrome P450c17α (CYP17) is a key enzyme in estrogen biosynthesis. In the 5′-untranslated region of CYP17, a T→C transition 34 bp upstream of the translation initiation site generates an MspAI restriction site (known as A2 allele). This polymorphism creates a new Sp1-type (CCACC box) binding site, providing an additional promoter activity and thus increasing the estrogen biosynthesis (9, 10). CYP19 is another key enzyme in estrogen biosynthesis. A nonsynonymous polymorphism, Trp39Arg, which is caused by a substitution of arginine for tryptophan at codon 39 of CYP19, results in the lack of estrogen biosynthesis in vitro (11).
Two other CYP enzymes, CYP1A1 and CYP1B1, are involved in the hydroxylation of estrogen. To date, two functional polymorphisms have been identified. One is the nonsynonymous polymorphism Ile462Val at codon 462 in the heme-binding region; the other is a T→C transition at 3′-noncoding region, which creates an MspI restriction site (known as m1 allele). Both of the polymorphisms are found to contribute to elevated enzyme activity (12, 13). In the CYP1B1 gene, a nonsynonymous polymorphism, Val432Leu, which is caused by a substitution of valine to leucine at codon 432 in exon 3, is linked to a higher catalytic activity (14).
Catechol-O-methyltransferase (COMT) is among the major enzymes responsible for inactivation of catechol estrogens, which are major metabolites of estrogens and can cause oxidative DNA damage. It has been shown that two nonsynonymous polymorphisms, Ala72Ser at codon 72 in exon 3 and Val158Met at codon 158 in exon 4, cause a dramatic reduction in enzyme activity (15, 16).
Several studies have shown these functional polymorphisms to be associated with a wide spectrum of cancers, including breast (17, 18), endometrial (19), ovarian (20), and prostate (21). In the present study, we examined whether the polymorphisms of EMEs have any bearing on the risk for hepatocellular carcinoma in hepatitis B virus (HBV) carriers and non–HBV carriers.
Materials and Methods
Patients and Controls
This case-control study consists of 434 incident patients with hepatocellular carcinoma and 480 control subjects. The diagnosis of cases, the inclusion and exclusion criteria for cases and controls, and the definition of HBV carriers, smokers, and drinkers were described in detail in our previous studies (5, 22). Briefly, the diagnosis of hepatocellular carcinoma was made by either positive histologic findings or an elevated α-fetoprotein level (≥400 ng/mL) combined with at least one positive image on the angiography, sonography, and/or high-resolution contrast computerized tomography. The controls had no evidence of hepatocellular carcinoma based on ultrasonography and serum α-fetoprotein level, and no individual history of other cancers. All participants were negative for antibodies to hepatitis C virus, hepatitis D virus, or HIV and had no any other type of liver disease such as autoimmune hepatitis, toxic hepatitis, and primary biliary cirrhosis or Budd-Chiari syndrome. At recruitment, informed consent was obtained from each subject, and personal information on demographic factors, medical history, history of cigarette smoking and alcohol drinking, and family history of cancer were collected via structured questionnaire. This study was done with the approval of the Medical Ethical Committee of Chinese National Human Genome Center.
Genotyping of Polymorphisms
Seven polymorphisms, including CYP17 MspAI site (rs743572), CYP19 Trp39Arg (rs2236722), CYP1A1 Ile462Val (rs1048943), MspI site (rs4646903), CYP1B1 Val432Leu (rs1056836), COMT Ala72Ser (rs6267), and Val158Met (rs4680), were genotyped by PCR-RFLP analysis. Briefly, the DNA sequence containing the relevant polymorphic site was amplified by PCR, and then the amplicon was digested with an appropriate restriction enzyme that cleaves only one of the two alleles. The digests were then subjected to gel electrophoresis and visualized by ethidium bromide staining. The PCR primers used in the aforementioned PCR-RFLP assays and the appropriate restriction enzymes are described in Table 1.
Primers and restriction enzymes used in EME polymorphisms genotyping by PCR-RFLP analysis
Polymorphisms . | Primers sequences* . | Amplicon size (bp) . | Restriction enzymes . | Digest size (bp) . |
---|---|---|---|---|
CYP17 MspAI site | 5′-TGGCTGGGTGCCG GCAGGCAAGATAGACcGC-3′ (forward) | 261 | ACC|| | A1/A1: 261 |
5′-TCCCTTACCTAGCTCCTCCTC-3′ (reverse) | A2/A2: 231, 30 | |||
A1/A2: 261, 231, 30 | ||||
CYP19 Trp39Arg | 5′-ATGACTGACTTACCTGGTATTGAGGATGTGCCCTCATAATTgC-3′ (forward) | 271 | Hin6I | Trp/Trp: 271 |
5′-CCTCTGAAGCAACAGGAGCTA-3′ (reverse) | Arg/Arg: 228, 43 | |||
Trp/Arg: 271, 228, 43 | ||||
CYP1A1 Ile462Val | 5′-GATAGCCAGGAAGAGAAAGACCTCCCAGCGGttAA-3′ (forward) | 183 | KspAI | Ile/Ile: 183 |
5′-TCCCTCTGGTTACAGGAAGCTA-3′ (reverse) | Val/Val: 150,33 | |||
Ile/Val: 183, 150,33 | ||||
CYP1A1 MspI site | 5′-CAGTGAAGAGGTGTAGCCGCT-3′ (forward) | 342 | MspI | wt/wt: 342 |
5′-TAGGAGTCTTGTCTCATGCCT-3′ (reverse) | m1/m1: 207, 135 | |||
wt/m1: 342, 207, 135 | ||||
CYP1B1 Val432Leu | 5′-TTGGCCCTGAAATCGCACcGGT-3′ (forward) | 240 | BseNI | Leu/Leu: 240 |
5′-CCAAGGACACTGTGGTTTTTGTCAAgCAG-3′ (reverse) | Val/Val: 194, 46 | |||
Leu/Val: 240, 194, 46 | ||||
COMT Ala72Ser | 5′-TGCTGTTGGCAGCTGTGT-3′ (forward) | 239 | ACC|| | Ser/Ser: 239 |
5′-GTAGGTGTCAATGGCCTCCAGCACGCTCTGcG-3′ (reverse) | Ala/Ala: 208, 31 | |||
Ser/Ala: 239, 208, 31 | ||||
COMT Val158Met | 5′-CTCTCCTCCGTCCCCAAC-3′ (forward) | 241 | SphI | Val/Val: 241 |
5′-AACGGGTCAGGaATGCACACCTTGTCCTgCA-3′ (reverse) | Met/Met: 212, 29 | |||
Val/Met: 241, 212, 29 |
Polymorphisms . | Primers sequences* . | Amplicon size (bp) . | Restriction enzymes . | Digest size (bp) . |
---|---|---|---|---|
CYP17 MspAI site | 5′-TGGCTGGGTGCCG GCAGGCAAGATAGACcGC-3′ (forward) | 261 | ACC|| | A1/A1: 261 |
5′-TCCCTTACCTAGCTCCTCCTC-3′ (reverse) | A2/A2: 231, 30 | |||
A1/A2: 261, 231, 30 | ||||
CYP19 Trp39Arg | 5′-ATGACTGACTTACCTGGTATTGAGGATGTGCCCTCATAATTgC-3′ (forward) | 271 | Hin6I | Trp/Trp: 271 |
5′-CCTCTGAAGCAACAGGAGCTA-3′ (reverse) | Arg/Arg: 228, 43 | |||
Trp/Arg: 271, 228, 43 | ||||
CYP1A1 Ile462Val | 5′-GATAGCCAGGAAGAGAAAGACCTCCCAGCGGttAA-3′ (forward) | 183 | KspAI | Ile/Ile: 183 |
5′-TCCCTCTGGTTACAGGAAGCTA-3′ (reverse) | Val/Val: 150,33 | |||
Ile/Val: 183, 150,33 | ||||
CYP1A1 MspI site | 5′-CAGTGAAGAGGTGTAGCCGCT-3′ (forward) | 342 | MspI | wt/wt: 342 |
5′-TAGGAGTCTTGTCTCATGCCT-3′ (reverse) | m1/m1: 207, 135 | |||
wt/m1: 342, 207, 135 | ||||
CYP1B1 Val432Leu | 5′-TTGGCCCTGAAATCGCACcGGT-3′ (forward) | 240 | BseNI | Leu/Leu: 240 |
5′-CCAAGGACACTGTGGTTTTTGTCAAgCAG-3′ (reverse) | Val/Val: 194, 46 | |||
Leu/Val: 240, 194, 46 | ||||
COMT Ala72Ser | 5′-TGCTGTTGGCAGCTGTGT-3′ (forward) | 239 | ACC|| | Ser/Ser: 239 |
5′-GTAGGTGTCAATGGCCTCCAGCACGCTCTGcG-3′ (reverse) | Ala/Ala: 208, 31 | |||
Ser/Ala: 239, 208, 31 | ||||
COMT Val158Met | 5′-CTCTCCTCCGTCCCCAAC-3′ (forward) | 241 | SphI | Val/Val: 241 |
5′-AACGGGTCAGGaATGCACACCTTGTCCTgCA-3′ (reverse) | Met/Met: 212, 29 | |||
Val/Met: 241, 212, 29 |
Italicized lowercase letters are the base mismatches.
Genotyping was done by staff blinded to the subjects' case or control status. The accuracy of genotyping data for polymorphisms obtained from PCR-RFLP analyses was tested by direct DNA sequencing of a 15% masked, random sample of cases and controls, and all results were in 100% concordance.
Statistical Analyses
Genotype and allele frequencies for each polymorphism were determined by direct gene counting. The fitness to Hardy-Weinberg equilibrium was tested using the random-permutation procedure implemented in the Arlequin package (available at http://lgb.unige.ch/arlequin/). The associations between the polymorphisms and risk for hepatocellular carcinoma were evaluated by multiple logistic regression analyses while controlling for confounding factors (including age, sex, status of smoking and drinking, pack-years of smoking, and family history), and the Ps, odds ratios, and 95% confidence intervals (95% CI) were calculated. Potential modification of the effect of the polymorphisms on hepatocellular carcinoma risk was assessed for the above confounding factors by the addition of interaction terms in the logistic model and by stratification analyses of subgroups of subjects determined by these factors. In view of the multiple testing, the correction factor n × (m - 1), where n means loci with m alleles each, was applied to correct the significance level. P < 0.0071 (= 0.05 / 7) was considered to be statistically significant, and all statistical tests were two sided. These analyses were done using SPSS software (version 9.0, SPSS, Inc.).
Haplotypes based on the polymorphisms Ile462Val and MspI in CYP1A1 and Ala72Ser and Val158Met in COMT were inferred using program PHASE 2.1 (available at http://www.stat.washington.edu/stephens/). The pairwise linkage disequilibrium index (Lewontin's D' and r2) calculation was done using the Arlequin package. Haplotype frequencies between cases and controls were compared using χ2 test; haplo.glm program (available at http://rss.acs.unt.edu/Rdoc/library/haplo.stats/html/haplo.glm.html) was then done to calculate adjusted odds ratios and Ps while adjusting for age, gender, status of smoking and drinking, pack-years of smoking, and family history of hepatocellular carcinoma for each haplotype, and the number of simulations for empirical Ps was set as 1,000. For high-order gene-gene interactions, multifactor dimensionality reduction analysis was done using multifactor dimensionality reduction software (version 1.0.0, available at http:/www.epistasis.org/mdr.html). Multifactor dimensionality reduction is a nonparametric and genetic model–free approach and can be used in case-control and discordant sib-pair study designs (23). Cross-validation consistency and balanced accuracy estimates were calculated for each combination of a pool of polymorphisms. The model with the highest accuracy and maximal cross-validation was considered to be the best. The statistical significance was determined by comparing the accuracy of the observed data with the distribution of accuracy under the null hypothesis of no associations derived empirically from 1,000 replicates of permutations.
Results
The genotyping results of seven polymorphisms are presented in Table 2. The genotype distribution of the COMT Ala72Ser polymorphism conformed to the Hardy-Weinberg equilibrium (P > 0.05) in patients but not in controls (P = 0.009). However, all the other six polymorphisms conformed to the Hardy-Weinberg equilibrium in patients and controls (P > 0.05). On the basis of logistic regression analysis with adjustment for age, sex, status of smoking and drinking, and family history, a significant association with the risk for hepatocellular carcinoma was observed for the COMT Ala72Ser in HBV carriers (Table 2). An increased risk for hepatocellular carcinoma was found to be associated with the Ala/Ser genotype, with the odds ratio being 4.06 (95% CI = 1.37-12.05; P = 0.012) compared with the Ala/Ala genotype. However, after correction for multiple comparisons, the association was never again significant. For the other six polymorphisms, that is, CYP17 MspAI site, CYP19 Trp39Arg, CYP1A1 Ile462Val and MspI site, CYP1B1 Val432Leu, and COMT Val158Met, we found no association with the risk for hepatocellular carcinoma in HBV carriers and non–HBV carriers (Table 2). The associations between these polymorphisms and the risk for hepatocellular carcinoma were further examined with stratification by age, sex, family history, status of smoking and drinking, and pack-years of smoking. Again, no significant association was found in HBV carriers and non–HBV carriers while correcting for multiple testing (data not shown).
The genotype and allele frequencies of 7 EME polymorphisms in patients with hepatocellular carcinoma and controls
Polymorphisms . | HBV carriers . | . | . | Non–HBV carriers . | . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Cases/controls . | OR (95% CI)* . | P* . | Cases/controls . | OR (95% CI)* . | P* . | ||||||
CYP17 MspAI site | ||||||||||||
A2/A2 | 113/70 | Reference | 37/124 | Reference | ||||||||
A2/A1 | 148/84 | 1.07 (0.71-1.61) | 0.75 | 54/132 | 1.38 (0.84-2.28) | 0.20 | ||||||
A1/A1 | 61/30 | 1.24 (0.72-2.12) | 0.44 | 14/38 | 1.34 (0.64-2.84) | 0.44 | ||||||
A1 allele | 0.42/0.39 | 0.39/0.35 | ||||||||||
CYP19 Trp39Arg | ||||||||||||
Trp/Trp | 297/173 | Reference | 98/277 | Reference | ||||||||
Trp/Arg | 16/13 | 1.32 (0.61-2.85) | 0.48 | 4/17 | 1.55 (0.48-4.94) | 0.46 | ||||||
Arg/Arg | 0/0 | NA | NA | 1/0 | NA | NA | ||||||
Arg allele | 0.026/0.035 | 0.029/0.029 | ||||||||||
CYP1A1 Ile462Val | ||||||||||||
Ile/Ile | 157/97 | Reference | 56/126 | Reference | ||||||||
Ile/Val | 118/70 | 1.03 (0.69-1.53) | 0.86 | 37/137 | 0.62 (0.37-1.01) | 0.057 | ||||||
Val/Val | 28/17 | 1.04 (0.53-2.05) | 0.90 | 5/30 | 0.40 (0.14-1.14) | 0.086 | ||||||
Val allele | 0.29/0.28 | 0.24/0.34 | ||||||||||
CYP1A1 MspI site | ||||||||||||
m1/m1 | 75/50 | Reference | 21/88 | Reference | ||||||||
m1/wt | 145/86 | 1.13 (0.71-1.79) | 0.61 | 50/134 | 1.46 (0.80-2.66) | 0.22 | ||||||
wt/wt | 76/49 | 1.14 (0.68-1.93) | 0.62 | 25/72 | 1.45 (0.73-2.86) | 0.28 | ||||||
wt Allele | 0.50/0.50 | 0.52/0.47 | ||||||||||
CYP1B1 Val432Leu | ||||||||||||
Leu/Leu | 290/168 | Reference | 97/273 | Reference | ||||||||
Leu/Val | 32/18 | 1.06 (0.57-1.97) | 0.85 | 8/21 | 1.12 (0.46-2.70) | 0.81 | ||||||
Val/Val | 2/0 | NA | NA | 0/0 | NA | NA | ||||||
Val allele | 0.056/0.048 | 0.038/0.036 | ||||||||||
COMT Ala72Ser | ||||||||||||
Ala/Ala | 298/180 | Reference | 100/283 | Reference | ||||||||
Ala/Ser | 24/4 | 4.06 (1.37-12.05) | 0.012 | 5/9 | 1.63 (0.52-5.16) | 0.41 | ||||||
Ser/Ser | 0/0 | NA | NA | 1/1 | 3.15 (0.19-52.07) | 0.42 | ||||||
Ser allele | 0.037/0.011 | 0.033/0.019 | ||||||||||
COMT Val158Met | ||||||||||||
Val/Val | 204/113 | Reference | 54/173 | Reference | ||||||||
Val/Met | 101/60 | 0.91 (0.61-1.36) | 0.65 | 43/97 | 1.46 (0.89-2.39) | 0.14 | ||||||
Met/Met | 12/10 | 0.73 (0.30-1.75) | 0.48 | 6/22 | 0.81 (0.30-2.23) | 0.69 | ||||||
Met allele | 0.20/0.22 | 0.27/0.24 |
Polymorphisms . | HBV carriers . | . | . | Non–HBV carriers . | . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Cases/controls . | OR (95% CI)* . | P* . | Cases/controls . | OR (95% CI)* . | P* . | ||||||
CYP17 MspAI site | ||||||||||||
A2/A2 | 113/70 | Reference | 37/124 | Reference | ||||||||
A2/A1 | 148/84 | 1.07 (0.71-1.61) | 0.75 | 54/132 | 1.38 (0.84-2.28) | 0.20 | ||||||
A1/A1 | 61/30 | 1.24 (0.72-2.12) | 0.44 | 14/38 | 1.34 (0.64-2.84) | 0.44 | ||||||
A1 allele | 0.42/0.39 | 0.39/0.35 | ||||||||||
CYP19 Trp39Arg | ||||||||||||
Trp/Trp | 297/173 | Reference | 98/277 | Reference | ||||||||
Trp/Arg | 16/13 | 1.32 (0.61-2.85) | 0.48 | 4/17 | 1.55 (0.48-4.94) | 0.46 | ||||||
Arg/Arg | 0/0 | NA | NA | 1/0 | NA | NA | ||||||
Arg allele | 0.026/0.035 | 0.029/0.029 | ||||||||||
CYP1A1 Ile462Val | ||||||||||||
Ile/Ile | 157/97 | Reference | 56/126 | Reference | ||||||||
Ile/Val | 118/70 | 1.03 (0.69-1.53) | 0.86 | 37/137 | 0.62 (0.37-1.01) | 0.057 | ||||||
Val/Val | 28/17 | 1.04 (0.53-2.05) | 0.90 | 5/30 | 0.40 (0.14-1.14) | 0.086 | ||||||
Val allele | 0.29/0.28 | 0.24/0.34 | ||||||||||
CYP1A1 MspI site | ||||||||||||
m1/m1 | 75/50 | Reference | 21/88 | Reference | ||||||||
m1/wt | 145/86 | 1.13 (0.71-1.79) | 0.61 | 50/134 | 1.46 (0.80-2.66) | 0.22 | ||||||
wt/wt | 76/49 | 1.14 (0.68-1.93) | 0.62 | 25/72 | 1.45 (0.73-2.86) | 0.28 | ||||||
wt Allele | 0.50/0.50 | 0.52/0.47 | ||||||||||
CYP1B1 Val432Leu | ||||||||||||
Leu/Leu | 290/168 | Reference | 97/273 | Reference | ||||||||
Leu/Val | 32/18 | 1.06 (0.57-1.97) | 0.85 | 8/21 | 1.12 (0.46-2.70) | 0.81 | ||||||
Val/Val | 2/0 | NA | NA | 0/0 | NA | NA | ||||||
Val allele | 0.056/0.048 | 0.038/0.036 | ||||||||||
COMT Ala72Ser | ||||||||||||
Ala/Ala | 298/180 | Reference | 100/283 | Reference | ||||||||
Ala/Ser | 24/4 | 4.06 (1.37-12.05) | 0.012 | 5/9 | 1.63 (0.52-5.16) | 0.41 | ||||||
Ser/Ser | 0/0 | NA | NA | 1/1 | 3.15 (0.19-52.07) | 0.42 | ||||||
Ser allele | 0.037/0.011 | 0.033/0.019 | ||||||||||
COMT Val158Met | ||||||||||||
Val/Val | 204/113 | Reference | 54/173 | Reference | ||||||||
Val/Met | 101/60 | 0.91 (0.61-1.36) | 0.65 | 43/97 | 1.46 (0.89-2.39) | 0.14 | ||||||
Met/Met | 12/10 | 0.73 (0.30-1.75) | 0.48 | 6/22 | 0.81 (0.30-2.23) | 0.69 | ||||||
Met allele | 0.20/0.22 | 0.27/0.24 |
NOTE: The frequencies of genotypes are indicated in absolute values. The number of samples genotyped varies because of genotyping failure for some individuals. The HBV carriers are subjects positive for hepatitis B surface antigen and anti–hepatitis B core antigen for at least 6 mo. All odds ratios and Ps are adjusted for age, gender, status of smoking and drinking, pack-years of smoking, and family history of hepatocellular carcinoma.
Abbreviations: OR, odds ratio; NA, not applicable.
No correction was made for testing multiple polymorphisms.
Furthermore, we did the haplotype analysis for evaluating the haplotype frequencies of polymorphisms located nearby at the same gene regions, trying to derive haplotypes specifically correlated with hepatocellular carcinoma. The linkage disequilibrium analyses showed that the Ile462Val and MspI in CYP1A1 and the Ala72Ser and Val158Met in COMT are in strong linkage disequilibrium (for CYP1A1: |D′| = 0.85, r2 = 0.30, P < 0.001; for COMT: |D′| = 1.00, r2 = 0.30, P < 0.001). Haplotypes based on the polymorphisms Ile462Val and MspI in CYP1A1 and Ala72Ser and Val158Met in COMT were then constructed, respectively. Four haplotypes of CYP1A1 were observed, and only 3 haplotypes had allele frequency of >5%. For COMT, three haplotypes were observed, and only two haplotypes had allele frequency of >5%. The estimated haplotype distribution in these two genes was not significantly different between the patients with hepatocellular carcinoma and controls in HBV carriers and non–HBV carriers after correction for multiple comparisons (Table 3).
Haplotype distribution of CYP1A1 and COMT in patients with hepatocellular carcinoma and controls
Haplotypes . | HBV carriers . | . | . | Non–HBV carriers . | . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Cases/controls . | OR (95% CI)* . | P* . | Cases/controls . | OR (95% CI)* . | P* . | ||||||
CYP1A1 | ||||||||||||
Ile-wt | 272/179 | Reference | 99/268 | Reference | ||||||||
Val-m1 | 146/101 | 0.96 (0.70-1.30) | 0.39 | 44/187 | 0.63 (0.42-0.94) | 0.019 | ||||||
Ile-m1 | 146/85 | 1.20 (0.86-1.70) | 0.56 | 46/121 | 1.03 (0.68-1.54) | 0.31 | ||||||
Val-wt | 24/3 | 4.41 (1.34-14.60) | 0.0096 | 1/10 | 0.27 (0.034-2.23) | 0.21 | ||||||
COMT | ||||||||||||
Ala-Val | 501/286 | Reference | 150/436 | Reference | ||||||||
Ala-Met | 125/80 | 0.92 (0.67-1.30) | 0.39 | 55/141 | 1.20 (0.81-1.70) | 0.58 | ||||||
Ser-Val | 24/4 | 3.62 (1.23-10.60) | 0.013 | 7/11 | 1.70 (0.71-4.20) | 0.27 |
Haplotypes . | HBV carriers . | . | . | Non–HBV carriers . | . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Cases/controls . | OR (95% CI)* . | P* . | Cases/controls . | OR (95% CI)* . | P* . | ||||||
CYP1A1 | ||||||||||||
Ile-wt | 272/179 | Reference | 99/268 | Reference | ||||||||
Val-m1 | 146/101 | 0.96 (0.70-1.30) | 0.39 | 44/187 | 0.63 (0.42-0.94) | 0.019 | ||||||
Ile-m1 | 146/85 | 1.20 (0.86-1.70) | 0.56 | 46/121 | 1.03 (0.68-1.54) | 0.31 | ||||||
Val-wt | 24/3 | 4.41 (1.34-14.60) | 0.0096 | 1/10 | 0.27 (0.034-2.23) | 0.21 | ||||||
COMT | ||||||||||||
Ala-Val | 501/286 | Reference | 150/436 | Reference | ||||||||
Ala-Met | 125/80 | 0.92 (0.67-1.30) | 0.39 | 55/141 | 1.20 (0.81-1.70) | 0.58 | ||||||
Ser-Val | 24/4 | 3.62 (1.23-10.60) | 0.013 | 7/11 | 1.70 (0.71-4.20) | 0.27 |
NOTE: The sums of haplotypes vary because of genotyping failure for some individuals. All odds ratios and Ps are adjusted for age, gender, status of smoking and drinking, pack-years of smoking, and family history of hepatocellular carcinoma.
No correction was made for testing multiple polymorphisms.
Epistasis or gene-gene interaction is increasingly assumed to play a crucial role in the genotype-to-phenotype relationship of common diseases. Thus, a nonparametric and genetic model–free approach, multifactor dimensionality reduction analysis, was done to explore the potential gene-gene interactions. However, we did not detect any statistically significant interactive effect for each combination of seven polymorphisms in our case-control data set (data not shown).
Discussion
In the present study, we assessed whether there was an association between the EME functional polymorphisms and the risk for HBV-related and non-HBV-related hepatocellular carcinoma. However, no significant association was observed in our case-control population. In addition, no evidence for gene-gene interactions was observed. It would be expected that the effects of EME genotypes may be masked and only become detectable in the presence of certain conditions. Indeed, many factors such as chronic infection with HBV, male gender, family history, smoking, and alcoholic consumption have been shown as independent risk factors for hepatocellular carcinoma (24-26). However, we did not find a statistically significant interaction between the EME polymorphisms and these risk factors, suggesting that these factors may not have modification effect on the susceptibility to hepatocellular carcinoma related to EME genotypes. These results thus do not support the hypothesis that the EME polymorphisms might modify susceptibility to hepatocellular carcinoma.
Previous study suggested that the endogenous hormonal environment in humans might modify the association between the high-risk EME genotypes and increased breast cancer risk; the cancer risk related to the EME genotypes was stronger in women with prolonged estrogen exposure and higher estrogen levels compared with that with a shorter duration of estrogen exposure and lower estrogen levels (27). On the basis of the obvious relevance of endogenous estrogens in human hepatocarcinogenesis, it is reasonable to anticipate the same events involved in hepatocellular carcinoma. Unfortunately, the information on personal estrogen exposure was not obtained in the present study. Additional studies investigating the interaction between the estrogen exposure and EME polymorphisms in hepatocellular carcinoma should be required in the future.
The genotype distribution of COMT Ala72Ser deviated from Hardy-Weinberg proportions in controls (P = 0.009); however, the concordance rate for the quality control samples (n = 152), which were randomly selected from cases and controls, was 100% for all the seven polymorphisms, including the COMT Ala72Ser. Therefore, we do not believe that the deviation from Hardy-Weinberg equilibrium for this polymorphism is due to genotyping error.
Evidences supporting the roles for polymorphisms in interindividual variation of EME activity ex vivo have been extensively analyzed in several studies (9-16). Moreover, several lines of evidence indicate that the EME polymorphisms can influence individual susceptibility to the development of hormone-related cancers (17-21). In view of the biological role for EMEs in metabolism of estrogens and the obvious biological plausibility of estrogen metabolites in hepatocarcinogenesis (3), the following factors may have contributed to lack of an association between the EME polymorphisms and hepatocellular carcinoma. First, additional polymorphisms that alter EME activity are likely. Our polymorphism selection strategy focused on variants with functional significance; we did not fully characterize risk in relation to all polymorphisms in these genes. Thus, variation in EME activity in human hepatocellular carcinoma may be only partly explained by these seven functional polymorphisms. Second, inadequate power may be an explanation for our negative results. This study had >85% power at a significance of 0.05 to detect a recessive allele with a minor allele frequency of 0.20 that confers a risk of 1.4. Thus, additional larger population-based case-control studies are warranted. Lastly, our negative results may be due to inherent selection bias. As a hospital-based study, our hepatocellular carcinoma cases were enrolled from the hospitals and the control subjects were selected from the community population, inherent selection bias cannot be completely excluded. However, by matching age and residential area, relying on covariate adjustments in the final analysis, and using analyses stratified by potential confounders, the potential selection bias might have been minimized.
The EME polymorphisms and hepatocellular carcinoma risk have been investigated previously, but the results are conflicting. Yin et al. (28) have reported an association of the high-activity m1 allele of CYP1A1 MspI site and the increased risk for hepatocellular carcinoma in Taiwan females. Furthermore, the women harboring two or three high-risk genotypes (including the high-activity CYP17 A2, high-activity CYP1A1 m1, and low-activity COMT 158Met alleles) had significantly increased risk for hepatocellular carcinoma compared with those harboring no or one variant. In contrast, in agreement with our present results, some reports also showed no association between certain EME polymorphism and risk for hepatocellular carcinoma. For instance, no association of the two CYP1A1 polymorphisms, MspI and Ile462Val, was observed with the hepatocellular carcinoma risk in either chronic hepatitis B carriers or hepatitis C virus–infected patients (29, 30). It was also reported that the CYP17 MspAI and COMT Val158Met are not associated with the risk for hepatitis C virus–related hepatocellular carcinoma (31). The conflicting results could be attributable to the differences in demography, ethnicity, lifestyles, type of viral infections, and clinical settings. In addition, other methodologic factors in the studies, such as small sample size, inadequate adjustment for confounding factors, or lack of correction for multiple testing, could also cause the inconsistent results.
In summary, our findings do not support associations between the seven functional polymorphisms of EME genes and the risk for hepatocellular carcinoma, suggesting that these EME polymorphisms might not involved in the predisposition to develop hepatocellular carcinoma. However, additional studies are warranted before the importance of EME polymorphisms in the etiology of hepatocellular carcinoma can be fully ascertained. First, data from larger population-based case-control studies among Chinese and from ethnically diverse populations are required to confirm our observation. Second, the other potentially functional polymorphisms in these EME genes and their associations with hepatocellular carcinoma risk should be systematically investigated. Lastly, investigation of additional polymorphic genes participating in the estrogen-metabolizing pathways, for example, steroid sulfatase, estrogen sulfotransferase, and 17β-hydroxysteroid dehydrogenase, should also be of interest.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: Chinese National Science Fund for Creative Research Groups Program grant 30621063 (F. He) and Chinese High-tech Program grant 2006AA02A412, Chinese National Basic Research Program grant 2006CB910803, Beijing Science & Technology NOVA Program grant 2006A54, and Chinese Key Project for the Infectious Diseases 2008ZX10002-016 (G. Zhou).
Acknowledgments
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank all the tested individuals, their families, and collaborating clinicians for their participation.