Hepatocellular carcinoma (HCC) is one of the most common cancers in the world, with rates that show considerable geographic variation. The major etiological factors associated with the development of HCC are chronic hepatitis B virus (HBV) and hepatitis C virus (HCV) infections. Several environmental factors, including aflatoxin B1 (AFB1), a dietary mold contaminant, and polycyclic aromatic hydrocarbons (PAHs), ubiquitous pollutants, are also associated with HCC incidence. Data have come from studies monitoring exposure by the measurement of urinary levels of aflatoxin metabolites and excised AFB1-guanine adducts,1, 2 AFB1-Alb adducts in plasma1, 3 and AFB1-, 4ABP- and PAH-DNA adducts in liver tissues.4, 5 The measurement of AFB1-Alb adducts has been used as a surrogate for DNA adducts when liver tissue is not available. While a correlation between AFB1 adducts in liver DNA and blood albumin was demonstrated in rats,6 no data in humans are available. To determine whether there is a correlation for both AFB1 and PAH between the liver DNA adducts and blood albumin adducts, paired samples were obtained from HCC cases in Taiwan and analyzed by immunohistochemistry and enzyme-linked immunosorbent assay (ELISA), respectively.
As with other cancers, the development of HCC is a complex, multistep process.7 The molecular pathogenesis of HCC appears to involve multiple genetic aberrations in the molecular control of hepatocyte proliferation, differentiation and death, and the maintenance of genomic integrity. This process is influenced by the cumulative activation and inactivation of oncogenes, tumor suppressor genes and other genes. p53, a tumor suppressor gene, is the most frequently mutated gene in human cancers. There is also a striking sequence-specific binding and induction of mutations by AFB1 at codon 249 of p53 in HCC.8, 9, 10, 11
Epigenetic alterations are also involved in cancer development and progression.12, 13, 14 Methylation of promoter CpG islands is known to inhibit transcriptional initiation and cause permanent silencing of downstream genes. It is now known that the most important tumor suppressor genes are inactivated, not only by mutations and deletions but also by promoter methylation. Several studies indicated that p16,15, 16, 17p15,18RASSF1A,19MGMT20 and GSTP121, 22 promoter hypermethylation are prevalent in HCC. In addition, geographic variation in the methylation status of tumor DNA indicates that environmental factors may influence the frequent and concordant degree of hypermethylation in multiple genes in HCC and that epigenetic–environmental interactions may be involved in hepatocarcinogenesis.19, 23
DNA, isolated from serum or plasma of cancer patients, frequently contains the same genetic and epigenetic aberrations as DNA isolated from an individual's tumor.24, 25, 26, 27 The process by which tumor DNA is released into the circulating blood is unclear, but may result from accelerated necrosis, apoptosis or other processes.28 In previous studies, p53 mutation and p16 promoter hypermethylation have been detected in paired tumor and plasma of HCC26, 27 and were also measured in this study. In addition, relationships between the markers in tissue and plasma from the same HCC patient were investigated. Our goal is to develop biomarkers, using plasma, that will assist in the early detection of HCC.
Patient population and data on clinical parameters
The study population consisted of 40 frozen dissected tumor tissues and 39 paired plasma samples of HCC patients, collected at the Department of Surgery, National Taiwan University Hospital. Informed consent was obtained from patients, and the study was approved by the appropriate institutional review committees. Demographic data, clinicopathologic characteristics' obtained from hospital charts, and HBV and HCV status, determined by immunoassay, are presented in Table I.
Table I. Demographics, HCV Status, HBsAg Status and Tumor Characteristics of HCC Cases
Immunohistochemical detection of AFB1- and PAH-DNA adducts in paraffin-embedded sections
Detection of AFB1-DNA adducts in 5 μm paraffin-embedded sections was carried out, basically, as described previously,4 using monoclonal antibody 6A10, which recognizes imidazole ring-opened guanine adducts. Slides were incubated at room temperature for 2 hr in 15 mM Na2CO3, 30 mM NaHCO3 (pH 9.6) to ring-open the guanine adducts. Peroxidase staining was performed using Vectastain and diaminobenzidine kits (ABC Elite and DAB Kits, Vector Laboratories, Burlingame, CA). Sensitivity was increased by further staining with 0.125% silver nitrate (Eastman Kodak, Rochester, NY) in 1.5% hexamethylene–tetramine (Sigma Diagnostics, St. Louis, MO) for 2 min. The following categories were used for scoring: intensity of staining, none (0), mild (1), moderate (2), strong (3) and percentage of positive staining, <5% (0), 5–25% (1), 25–50% (2), >50% (3). Combining intensity and percentage resulted in the following scores: 0–2 = low, 3, 4 = moderate, 5, 6 = high. Categories reflect levels of staining only observed in tumor hepatocytes. A normal human liver tissue section from the U.S.A. and a slide of smeared human lymphocytes treated in vitro with AFB1 were used as negative and positive controls, respectively, and were stained with each batch of slides.
Detection of PAH-DNA adducts was carried out, basically, as described previously,5 using monoclonal antibody 5D11 generated against benzo[a]pyrene diol epoxide (BPDE)-modified DNA.29 This antibody significantly crossreacts with DNA, modified by the diol epoxides of structurally related PAH. ABC staining was performed as earlier, but without silver nitrate, and also scored as low, moderate and high as mentioned earlier. Human lymphocytes treated with or without BPDE were used as positive and negative controls and stained with each batch of slides.
Quantitation of AFB1- and PAH-Alb adducts in plasma
Albumin was isolated from plasma, essentially, as described previously.1 A competitive ELISA, using polyclonal antiserum No. 7, was used to determine the level of AFB1-Alb adducts after enzymatic digestion of albumin.1 Two control samples were analyzed with each batch of sera, a pooled sample of plasma from nonsmoking U.S.A. subjects and a positive control of plasma from a rat treated with 1.5 mg AFB1.
PAH-Alb adducts were determined after acid release of BPDE tetrols and other PAH metabolites by competitive ELISA, using monoclonal antibody 8E11, essentially, as previously described.30 The pooled U.S.A. plasma sample was used for quality control. In both assays, samples were assayed by duplicate analysis in duplicate wells, and samples with <20% inhibition were considered nondetectable and assigned a value of 1 fmol/μg for AFB1-Alb and 150 fmol/μg for PAH-Alb.
DNA was isolated from frozen tissue samples, as previously described.31 Briefly, tissue was placed in liquid nitrogen and pulverized with a blender. The tissue powder was lysed with a DNA lysing buffer (10 mM Tris, 10 mM NaCl, 0.1% sodium dodecyl sulfate at pH 7.9, and 200 μg/ml, proteinase K). DNA was isolated by RNase treatment, phenol/chloroform extraction and ethanol precipitation. DNA was isolated from 200 μl of plasma by using QIAamp DNA Blood Mini Kits (Qiagen, Stanford Valencia, CA).
p53 mutation detection
Mutations in exons 5–8 of p53 were detected by DNA sequencing, using an ABI 3100 capillary sequencer. DNA (25 ng) from both tumor tissue and plasma was amplified using the following primers and PCR conditions: exon 5, forward primer ATC TGT TCA CTT GTG CCC TG, reverse primer AAC CAG CCC TGT CGT CTC TC, amplification at 95°C for 15 min, 35 cycles of 94°C for 30 sec, 58°C for 30 sec and 72°C for 30 sec, respectively. Exon 6, forward primer AGG GTC CCC AGG CCT CTG AT, reverse primer CAC CCT TAA CCC CTC CTC CC, amplification at 95°C for 15 min, 35 cycles of 94°C for 30 sec, 61°C for 30 sec and 72°C for 30 sec, respectively. Exon 7, forward primer CCA AGG CGC ACT GGC CTC ATC, reverse primer CAG AGG CTG GGG CAC AGC AGG, amplification at 95°C for 15 min, 35 cycles of 94°C for 30 sec, 66°C for 30 sec and 72°C for 30 sec, respectively. Exon 8, forward primer TTC CTT ACT GCC TCT TGC TT, reverse primer TGT CCT GCT TGC TTA CCT CG, amplification at 95°C for 15 min, 35 cycles of 94°C for 30 sec, 55°C for 30 sec and 72°C for 30 sec, respectively. Before sequencing, PCR products were purified using a MinElute PCR Purification kit (Qiagen). Samples with mutations were resequenced, to rule out the possibility of artifacts. Standard sequencing methods are reported to have a sensitivity for detection of an altered base, when it is present at a level of 10% (Applied Biosystems).
Immunohistochemistry for mutant p53 protein
Five micrometers of paraffin-embedded sections of tumor tissue were stained for p53 protein, using monoclonal antibody NCL-p53-DO7 (Novocastra Laboratories, Burlingame, CA). After deparaffinization and rehydration in graded ethanol, the slides were immersed in 10 mM citric acid (pH 6.0) and microwaved for 10 min at 400 W. Staining was carried out according to the manufacturer's instructions: primary antibody (1:100 dilution) was added, and sections were incubated overnight at 4°C, followed by adding a secondary antibody and ABC reagent and DAB (Vector Laboratories). Slides were counterstained with Harris Hematoxylin (Sigma). Samples were considered positive if >5% of cells were stained. Lung cancer tissues with or without mutant p53 protein were used as positive and negative controls.
Analysis of p16 hypermethylation–methylation-specific polymerase chain reaction
Methylation-specific polymerase chain reaction (MSP) was carried out, essentially, as described previously,32 and was based on the principle that treating DNA with sodium bisulfite results in the conversion of unmethylated cytosine residues into uracil. Thus, the sequence of the treated DNA will differ if the DNA is originally methylated versus unmethylated, and is then distinguishable by sequence-specific PCR primers. Bisulfite modification of both tissue and plasma DNA was conducted with the CpGnome DNA Modification Kit (Chemicon International, Temecula, CA). Bisulfite-treated DNA was amplified with primers for the methylated p16 sequence. As a quality control for the bisulfite modification process, all bisulfite-treated DNA were also amplified with primers specific for the unmethylated p16 sequence. PCR was conducted with the CpGWIZ p16 Amplification Kit (Chemicon International) and AmpliTaq Gold Polymerase (Perkin-Elmer, Norwalk, CT) and a total of 40 cycles. The thermal profile consisted of an initial denaturation step of 95°C for 10 min, followed by repetitions of 95°C for 45 sec, 60°C for 45 sec, and 72°C for 60 sec, with a final extension step of 72°C for 10 min. PCR products were analyzed by agarose gel electrophoresis and ethidium bromide staining. The methylated DNA control, from CpGnome Amplification Kit, was used as a positive control and distilled water as a negative control.
We initially conducted F-tests, using ANOVA, to evaluate the differences in mean albumin adducts across different levels of DNA adducts in the tumor tissues. Multiple linear regression models were then used to examine the relationship between albumin adducts and DNA adducts by testing the difference in mean of AFB1- and PAH-Alb adducts, comparing (1) medium level to low level and (2) high level to low level of AFB1- and PAH-DNA adducts in tumor tissues, respectively. Trend tests were performed by fitting the model with a single ordinal variable, indicating levels of AFB1- and PAH-DNA adducts in tumor tissues. Pairing of tumor tissue and plasma for the same subject was controlled in the analysis. χ2 tests were performed to evaluate the agreement of p53 mutation and p16 promoter hypermethylation status between the tissue and plasma DNA samples. In addition, we evaluated the associations of AFB1- and PAH-Alb adducts with HBV status, p53 mutation and p16 promoter hypermethylation. All the analyses were performed using statistical analysis software (SAS 8.0).
AFB1- and PAH-DNA and albumin adducts in paired samples
AFB1 and PAH exposure was assessed by measuring AFB1- and PAH-DNA adducts in HCC tumor tissues by immunoperoxidase staining and albumin adducts in plasma, by ELISA. Representative stainings of high (a) and low (b) tumors for AFB1-DNA and PAH-DNA are given in Figures 1 and 2, respectively. Staining intensity, categorized as low, moderate or high, is given in Table II, as are the data on AFB1- and PAH-Alb adducts. The DNA-Alb adduct correlations for both AFB1 and PAH are given in Table III. Subjects with high levels of AFB1-DNA adducts in tumor tissues have higher values of AFB1-Alb adducts compared with subjects with low levels of AFB1-DNA adducts in tumor tissues (p = 0.06). AFB1-Alb adducts in subjects with low, moderate and high levels of AFB1-DNA adducts in tumor tissues were 51.0 ± 36.5, 70.5 ± 48.1 and 84.9 ± 48.2 fmol/mg, respectively (ptrend = 0.05). For PAH-Alb, the corresponding values were 170 ± 56, 156 ± 20 and 235 ± 180 fmol/mg, respectively (ptrend = 0.36).
Table II. AFB1- and PAH-DNA Adducts in Liver Tissue, AFB1- and PAH-Alb Adducts in Plasma, Mutant p53 Protein and p53 Gene Mutations, and p16 Methylation Status in Liver Tissues and Plasma DNA of HCC Cases
Pairing of tumor tissue and plasma for the same subject was controlled in the analysis.
Level of adducts in plasma was treated as a dependent variable, and levels of DNA adducts in tumor tissues was an independent variable (0, 1, 2).
Levels of AFB1-DNA adducts in tumor tissues
51.0 ± 36.5
70.5 ± 48.1
24.0 ± 16.8
84.9 ± 48.2
38.3 ± 20.2
Levels of PAH-DNA adducts in tumor tissues
170 ± 56
156 ± 20
−24 ± 32
235 ± 180
53 ± 39
p53 mutations in HCC tissue and plasma DNA samples
Data on mutant p53 protein by immunohistochemistry and p53 gene mutations, by sequencing the tumors, are given in Table II. Fourteen of 40 (35%) tissues were positive for mutant p53 protein, as ascertained by immunohistochemistry and 11 of 40 (28%) for DNA mutation, as assessed by direct sequencing. A representative DNA sequencing analysis for a tumor with a mutation in p53 is shown in Figure 3. Five samples were positive for mutations by immunohistochemistry but not by sequencing, and 2 samples containing gene mutations were negative by immunohistochemistry. The agreement between sequencing and immunohistochemistry was 83% (p < 0.01). The majority of mutations were located in exon 7 (5 of 11); 3 of 5 mutations were G:C→A:T transitions at codon 245. A codon 249 mutation was found. The distribution of mutations at exons 5, 6 and 8 were 1, 3 and 2, respectively. When DNA isolated from plasma was sequenced, no p53 mutations were detected.
p16 promoter hypermethylation in HCC tissue and plasma DNA samples
Methylation of the promoter region of p16, determined by MSP, was frequent in the HCC tumors with 25 of 40 (63%) samples positive (Table II). A representative example of the MSP analysis is shown in Figure 4. DNA isolated from the plasma of 39 subjects was also tested for p16 promoter methylation of which 12 (31%) samples were positive (Table II). For 13 subjects whose tumor DNA was positive for p16 promoter hypermethylation, plasma DNA was negative. One subject with a p16 negative tumor was positive for plasma hypermethylation. The agreement between tissue and plasma DNA was 66% (p = 0.01).
AFB1- and PAH-Alb adducts, HBV status, p53 mutation and p16 promoter hypermethylation
AFB1-Alb adduct levels did not vary by HBV status, but were higher in subjects with p53 mutations (86.9 ± 45.6 fmol/mg) than in those without mutations (54.6 ± 40.2, p = 0.03) (Table IV). Plasma DNA p16 hypermethylation was also associated with higher levels of AFB1-Alb adducts. AFB1-Alb adduct levels were 87.7 ± 52.6 and 53.0 ± 35.0 fmol/mg in those with or without p16 plasma DNA methylation, respectively (p = 0.02). There was no association of AFB1-Alb adducts with p16 methylation in the tumors. PAH-Alb adducts were not associated with HBV status, p53 alterations or p16 methylation.
Table IV. Associations of AFB1- and PAH-Alb Adducts in Plasma with HBV, p53 Mutation, p16 Methylation Status
Many chemical carcinogens can bind to DNA, RNA and protein to form adducts. The measurement of DNA and protein adducts has been used to monitor carcinogen exposure and provides relatively precise dosimetry. Protein adduct formation is considered to be a valuable surrogate for DNA adduct formation, since in animal models, many chemical carcinogens bind to both DNA and protein in blood with similar dose–response kinetics.33, 34 AFB1-Alb adducts have been detected in peripheral blood of both animals and humans exposed to aflatoxin. As with DNA adducts, there is a linear dose–response relationship, following aflatoxin exposure in rats.35, 36 Furthermore, with multiple exposure, adducts accumulate and reach a plateau. However, alterations in metabolism and repair of DNA damage may occur in tumors. For example, interference of the HBxAg of HBV with the host's repair system could cause slower repair9. Thus, the level of adducts may not always directly correlate with the level of exposure.
Although blood protein adducts have been used extensively as biomarkers of exposure, little information is available on their correlation with DNA adducts in the target tissue in humans.37 Our results indicate that there is a good correlation between AFB1-Alb in blood and AFB1-DNA adducts in liver tumors. While there was a significant association between AFB1-DNA adducts level in tissues and albumin-adducts in plasma, discordant results were found in some cases (Table II). There are several possibilities for this discordance. The half-life for albumin may be different from the life span of DNA adducts in liver tissue. Thus, the single time point for collecting the samples may result in adducts being influenced by recent exposures that may vary among subjects. Many of the tissue samples were small and may not adequately represent adduct formation throughout the tissue, limiting the data on detection of DNA adducts. For PAH, although the highest albumin adducts were in those with the highest liver DNA adducts, the relationship was not statistically significant probably because of the small sample size and the low percentage of subjects with high adduct levels.
Our previous studies have supported a role for aflatoxin in HCC in Taiwan; 80% of HCC cases had detectable AFB1-DNA adducts in liver tissue compared to 43% of controls.4 Detectable AFB1-Alb adducts were present in 59.6% of cases and 34.4% of controls in a case-control study, nested in a cancer screening study.1 In the present study, moderate and high levels of AFB1-DNA adducts were detected in 20 (56%) HCC tumor tissues, a level within the range of our previous studies (53–80%).4, 38, 39
Epidemiological study suggested that cigarette smoking is a risk factor for HCC.40 Our prior study demonstrated an association between PAH-DNA adducts and HCC.5 In the present study, moderate and high levels of PAH-DNA adducts were detected in 16 (43%) HCC tumor tissues, lower than our prior study (75%). This result may reflect differences in exposure to PAH between the 2 sets of patients.
Mutations in the p53 tumor suppressor gene are found in more than 50% of human cancers, and distinct mutational spectra have been observed for different cancer types.41 A selective mutation in codon 249 has been identified as a “hotspot” mutation in HCC, occurring in populations exposed to AFB1.42, 43 In recent years, the p53 codon 249 mutation has also been detected in plasma DNA of HCC patients.27, 44, 45, 46 For example, among 20 paired tumors and plasma from HCC cases, 11 tumors contained the codon 249 mutation, and this mutation was also detected in 6 of the paired plasma samples.27 In the present study, 14/40 (35%) tissues were positive for mutant p53 protein, as ascertained by immunohistochemistry and 11/40 (28%) for p53 gene mutations. These results are consistent with our previous study in Taiwan.4 However, in contrast with our prior study in which codon 249 mutations were 40% of the total mutations found, only one codon 249 mutation was found in the present study. Perhaps, in these recently collected samples, exposure to AFB1 is lower than in the past. In addition, this more diverse spectrum of p53 mutations may be related to other environmental factors, such as HBV infection and carcinogen (e.g., PAH) exposure. Unfortunately, using direct sequencing of plasma DNA, we were unable to identify any samples containing p53 mutations. Previous studies of HCC specifically looked for codon 249 mutations, using mass spectroscopy,27 by digestion with a restriction enzyme to enrich the sample for mutant DNA. The latter method is not possible for all the noncodon 249 mutations found in this study. Studies in breast cancer have been able to detect mutant p53 in plasma DNA, but only after isolation of the PCR product from the gel.47 We attempted allele-specific PCR for detection of mutations as described,48 but found that the method was not specific. Since the goal of our work was to determine a method that could be used for screening relatively large numbers of samples, we did not pursue alternate methodologies.
In this study, 25 of 40 tumor samples (62%) were positive for p16 methylation by MSP (Table II), consistent with previous frequencies of 42–73%.15, 16, 17 Analysis of DNA from the available 39 plasma samples found 12/39 (30%) positive, lower than a previous study (72%),26 in which only 22 HCC cases were analyzed. These results suggest that further testing of p16 methylation as a biomarker of disease is appropriate. We are in the process of using blood collected in a nested case-control study, to determine the specificity and sensitivity of p16 plasma DNA methylation, for early detection of HCC.
A statistically significant correlation was found between AFB1-Alb adducts and p16 plasma DNA hypermethylation (p = 0.02) and p53 mutations (p = 0.05). These observations are consistent with our previous studies, which demonstrated significant correlations between AFB1-DNA adducts and methylation in RASSF1A,19MGMT20 and GSTP122 and p53 mutations,4 and provide further evidence that gene-environment and epigenetic–environment interactions may play important roles in the development of HCC.
In summary, this is the first study to detect adducts in paired liver tissue and plasma samples in humans. We observed a good correlation between AFB1-DNA adducts in liver tissue and albumin adducts in blood, providing support for the use of albumin adducts as a marker of exposure to AFB1 in studies investigating the role of environmental exposures on liver cancer risk. In addition, the use of serum or plasma DNA methylation analysis is a promising tool for studying the pathological basis of HCC. Blood samples can be used to detect genetic and epigenetic alterations in tumors as well as the adducts of chemical carcinogens, and may permit high-throughput screening for HCC at an early and potentially resectable stage. However, the sensitivity and specificity of HCC detection, using serum or plasma DNA samples, and thus the clinical utility of a test based on plasma DNA, remain uncertain at the present time.