• Open Access

Mutational complex genotype of the hepatitis B virus X /precore regions as a novel predictive marker for hepatocellular carcinoma


To whom correspondence should be addressed.
E-mail: sunphong@genematrix.net; ymp1@hotmail.com


This study explored the combined effect of number and pattern of mutations in the X/precore regions of the hepatitis B virus (HBV) genome, mutational complex genotype (MCG), on hepatocellular carcinoma (HCC) development. Sequence variations were determined by direct sequencing and multiplex restriction fragment mass polymorphism analysis in 150 age-, sex- and hepatitis B e antigen (HBeAg) status-matched patients with and without HCC. In addition, a longitudinal study and an external validation of MCG were conducted. All were HBV subgenotype C2. Eight high-frequency mutations (G1613A, C1653T, T1753V, A1762T, G1764A, A1846T, G1896A and G1899A) were significantly associated with HCC. Whereas C1653T, T1753V, G1764A and A1846T were independent mutational factors for HCC, the significance of these individual mutations was negligible when analyzed with all clinico-virological variables. The total number of mutations was the only independent viral factor for HCC, irrespective of HBeAg status. There was a significant dose–risk relationship between the number of mutations and HCC, in which high risks for HCC were associated with mutation numbers ≥6. Pattern analysis of the mutations revealed disparity in distribution among the top seven high-risk mutation combination patterns, which accounted for 40 and 2.7% of HCC and non-HCC cases, respectively. The predictive accuracy of the high-risk mutations for HCC was similar to that of α-fetoprotein. Longitudinal and external validation studies also supported the association of mutation number with HCC development. MCG in the HBV X/precore regions is a risk indicator for HCC, and might serve as a new guide to the HCC screening scheme for chronic HBV carriers. (Cancer Sci 2012; 103: 296–304)

Hepatocellular carcinoma (HCC) is a major malignancy worldwide, ranking as the fifth most common cancer. Hepatitis B virus (HBV) infection is a major global cause of HCC, with chronic HBV carriers having a 100-fold increased risk for developing HCC compared to non-carriers.(1) Identification of HCC indicators and risk stratification of patients with chronic HBV infection are essential for guiding appropriate surveillance.

Hepatocarcinogenesis as a process in individuals with chronic HBV infection is complex, and involves both host and viral factors. Various viral factors are associated with an increased risk of HCC, including HBV DNA levels, HBeAg status, HBV genotypes, pre-S deletion and mutations in the X/precore regions.

Hepatitis B virus is classified into eight genotypes (A–H), based on an intergroup sequence divergence of 8% or more. HBV genotypes B and C are the predominant types in eastern Asia.(2,3) HBV genotype C is the predominant type in Korea, accounting for more than 95% of Korean chronic HBV carriers.(4,5) Several published studies indicate that carriers infected with HBV genotype C exhibit a high risk of developing HCC,(6,7) but this has not been corroborated.(8,9) In addition to HBV genotype, mutations in the X/precore regions have been associated with advancement of liver disease leading to cirrhosis and HCC.(2,10–12) The A1762T/G1764A mutations, which are the most commonly studied molecular variants in the basal core promoter (BCP) region, and the C1653T and T1753V mutations, which were recently identified, have been the focus of attention as viral mutations associated with HCC.(2,13–16) The G1896A stop codon mutation in the precore region is often associated with a severe form of hepatitis and deteriorating liver disease in chronic HBV carriers.(17,18)

X/precore regions are a complex part of the HBV genome clustered across various functional sequences, such as: enhancer II, BCP, X-termination signal, two pregenomic RNA start points, poly A signal, epsilon (ε) and other important sequences. Mutations in the X/precore regions are supposedly associated with the development of HCC through the changes in presumed carcinogenetic capability. The association of mutations in the X/precore regions with the development of HCC has not been entirely confirmed in light of conflicting results.(19,20) Most studies have focused only on the evaluation of a specific mutation rather than on the combined effects of multiple mutations. To elucidate the carcinogenic role of mutations in the X/precore regions, it is better to take into consideration the combined effects of multiple mutations, which gradually accumulate during the long-term HBV infection.

We conducted a cross-sectional matched analysis of Korean chronic HBV carriers to evaluate the combined effects of the number and patterns of multiple mutations occurring in the X/precore regions on the risk of HCC development, and validated the effect of multiple mutations in an independent dataset and longitudinal cases in which samples were serially obtained before and after the occurrence of HCC. The analysis of mutational complex genotypes (MCG) consisting of the number of the X/precore mutations and their combination patterns permits molecular characterization of a high-risk MCG with the specific number and patterns of mutations associated with HCC.

Materials and Methods

Patients and serum samples.  This study included 150 patients with chronic HBV infection who visited Bundang Jesaeng General Hospital or the Catholic University of Korea Incheon St. Mary’s Hospital in Korea. All subjects were of Korean ethnicity. Serum samples were obtained from patients before antiviral treatment, and then stored at −20°C. Separately, 25 patients were included in the longitudinal analysis, because their serum samples were collected serially before and after the occurrence of HCC. None of the patients had co-infection with hepatitis C virus or human immunodeficiency virus. Patients with heavy alcohol consumption habits (≥80 g ethanol per day) were also excluded from the study. Each patient provided informed consent to participate in the study. This study was approved by the Ethics Committees of the participating institutions in accordance with the 1975 Declaration of Helsinki.

Assays for hepatitis B virus markers and hepatitis B virus genotyping.  Serum hepatitis B surface antigen was quantified using the ARCHITECT quantitative assay (Abbott Laboratories, Abbott Park, IL, USA). HBeAg and antibody to HBeAg were measured by commercial immunoassays (Abbott Laboratories). The serum HBV DNA titer was quantified using COBAS AmpliPrep-COBAS TaqMan (detection limit: 20 IU/mL; Roche Diagnostics GmbH, Mannheim, Germany). HBV genotypes were determined by the web-based National Center for Biotechnology Information (Bethesda, MD, USA) retrovirus genotyping analysis of the obtained sequences from each patient (http://www.ncbi.nlm.nih.gov/projects/genotyping).(21)

Sequence analysis of the X/core promoter and precore regions.  Sequence analysis was carried out in the X/precore regions, using the BigDye terminator v3.1 Cycle Sequencing Kit and ABI PRISM 3100 DNA analyzer (Applied Biosystems, Foster City, CA, USA). Sequencing primers (Seq-F, Seq-R) were designed based on the consensus sequences extracted from the multiple alignment of various HBV genotype sequences (genotypes A to H) retrieved from the Entrez Nucleotide database of the National Center for Biotechnology Information (queried by “hbv and promoter”) to maximize their usefulness against all genotype HBV sequences (Table S1). Sequence variants identified by sequencing were further verified by restriction fragment mass polymorphism (RFMP) assay. RFMP analysis was performed (Data S1) using specifically designed primers (Table S1), as described previously.(22) All mutations identified by direct sequencing were concordant with those obtained by RFMP analysis.

Case–control study.  According to the practice guidelines for diagnosis of HCC,(1,23) 75 patients were diagnosed with HCC on the basis of histologic evidence, typical radiologic findings (arterial enhancement with portal washed-out) or elevated serum α-fetoprotein (AFP) levels (>200 ng/mL). To obtain more reliable information about the association between HBV mutations and HCC, age-, sex-, HBV genotype- and HBeAg status-matched subjects of an additional 75 patients without HCC were compared. Among the non-HCC group, nine patients were HBeAg-positive asymptomatic carriers with persistently normal alanine aminotransferase (ALT) levels, 48 had chronic hepatitis, and 18 had liver cirrhosis diagnosed by pathology (n = 2) or liver imaging suggestive of cirrhosis, which was supplemented with clinical findings, such as ascites, thrombocytopenia, splenomegaly or varix (n = 16).

Validation study.  We performed a longitudinal analysis, using serially collected serum samples from 25 cases, in which HCC was detected during the follow-up period using the regular HCC screening program (male/female = 16/9; 55 ± 10 years; range, 33–74 years). In addition, using the GenBank database (accession numbers GQ475305GQ475357, which were registered by another research group),(24) external validation was performed on an independent set of 52 cases to evaluate the effects of MCG on HCC.

Statistical analysis.  Data were expressed as mean ± SD. Statistical analyses were carried out using the Mann–Whitney U-test for continuous variables, and the χ2-test and Fisher’s exact test for categorical variables, as appropriate. In addition to mutation patterns, the number of mutations, defined as the sum of individual mutations, was also analyzed. A linear trend in HCC risk with the increase in the summed mutation number was evaluated for statistical significance using a test to determine trends. The diagnostic value of the mutation number was assessed according to the area under the receiver operating characteristic (AUROC) curve. Multivariate analysis with logistic regression was performed to determine the independent factors associated with HCC. Two-tailed P-values <0.05 were considered significant. Data were analyzed using SPSS (version 15.0; SPSS, Chicago, IL, USA).


Baseline characteristics.  Clinical and virological characteristics of the 150 study subjects are shown in Table 1. Age, sex and HBeAg status were matched in patients with and without HCC. Serum platelet counts, ALT and HBV DNA levels were lower in patients with HCC than in those without HCC, whereas AFP levels were significantly higher in patients with HCC. HBV genotyping showed that all patients (100%) had subgenotype C2.

Table 1.   Clinical and virological characteristics of study population (n = 150)
 Non-HCC (n = 75)HCC (n = 75)P-value
  1. ALT, alanine aminotransferase; AFP, α-fetoprotein; HCC, hepatocellular carcinoma.

Sex (male/female) (%)59 (78.7)/16 (21.3)59 (78.7)/16 (21.3)>0.990
Age (years)52.9 ± 8.952.9 ± 8.9>0.990
ALT (U/L)45 (8–1633)36 (7–604)0.028
Total bilirubin (mg/dL)1.0 (0.1–23.9)1.1 (0.3–9.1)0.359
Platelet count (×103/mm3)164.8 ± 62.2110.9 ± 66.9<0.001
AFP (ng/mL)4.3 (1.2–152.1)81.5 (2.2–110 312)<0.001
HBeAg positivity (%)36 (48.0)36 (48.0)>0.990
HBV DNA (×103 copies/mL)457 (0.096–2 900 000)179 (0.110–1 380 000)0.070

Eight key X/precore mutations associated with hepatocellular carcinoma.  We analyzed all observed nucleotide substitutions to identify mutations associated with HCC. Sequence analysis of the X/precore regions revealed that 12 mutations (G1613A, C1631T, C1653T, T1674C, T1753V, T1754C, A1762T, G1764A, C1773T, A1846T, G1896A and G1899A) were more frequent in HCC patients than in non-HCC patients. There were eight specific mutations, including G1613A, C1653T, T1753V, A1762T, G1764A, A1846T, G1896A and G1899A, selected for analysis based on significance of association with HCC (odds ratio [OR] > 2, P < 0.05; Table 2).

Table 2.   Analyses of risk factors associated with HCC
AnalysesUnivariate analysisMultivariate analysis
  1. †Analysis done only for viral mutational factors, which were selected in the X/precore regions of HBV genome by correlation analysis with HCC. None of these eight mutations were significant when analyzed for all clinico-virological factors. ‡Of the clinico-virological factors, platelet count, AFP and the number of mutations were significant independent factors in multivariate analysis for all clinico-virological factors. These three factors were further analyzed by comparison of the degree of significance according to the HBeAg status. The other mutations including G1613A (P = 0.285), A1762T (P = 0.430), G1896A (P = 0.438) and G1899A (P = 0.690) were not significant in the multivariate analysis. AFP, α-fetoprotein; CI, confidential interval; OR, odds ratio; HBV, hepatitis B virus; HCC, hepatocellular carcinoma.

Eight viral mutational factors†
 G1613A (%)21 (28.3)41 (54.7)0.001  
 C1653T (%)19 (25.3)46 (61.3)<0.0014.240 (1.810–9.930)0.001
 T1753V (%)12 (16.0)38 (50.7)<0.0015.256 (2.124–13.007)<0.001
 A1762T (%)56 (74.7)71 (94.7)0.001  
 G1764A (%)59 (78.7)74 (98.7)<0.00111.705 (1.209–113.310)0.034
 A1846T (%)11 (14.7)38 (50.7)<0.0016.474 (2.496–16.788)<0.001
 G1896A (%)37 (49.3)52 (69.3)0.013  
 G1899A (%)6 (8.0)23 (30.7)<0.001  
Three clinico-virological factors‡
 HBeAg(+) patients (n = 78)
  Platelet count   1.019 (1.001–1.037)0.035
  Number of mutations   1.817 (1.087–3.037)0.023
 HBeAg(−) patients (n = 72)
  AFP   1.075 (1.002–1.154)0.044
  Number of mutations   2.709 (1.729–4.245)<0.001
 All patients (n = 150)
  AFP   1.008 (1.003–1.014)0.005
  Platelet count   1.013 (1.005–1.021)0.002
  C1653T   1.682 (0.554–5.100)0.359
  T1753V   1.829 (0.632–5.292)0.265
  G1764A   2.270 (0.405–12.736)0.352
  A1846T   1.394 (0.239–8.141)0.712
  Number of mutations   2.087 (1.502–2.899)<0.001

Significance of mutation occurrences at eight key positions for the prediction of hepatocellular carcinoma risk.  Multivariate analysis of the eight target nucleotide positions revealed that C1653T, T1753V, G1764A and A1846T mutations were associated with HCC risk (Table 2). However, when all significant clinico-virological factors, such as AFP, platelet counts, the four independent mutations (C1653T, T1753V, G1764A and A1846T) and the number of all eight mutations, were included in the analysis, the individual effects of specific mutations on HCC were not statistically significant. In contrast, the total number of mutations in the eight target positions was identified as a unified independent predictor of HCC (OR, 2.087; 95% CI: 1.502–2.899, P < 0.001), irrespective of HBeAg status (Table 2). The significance of the mutation number as an independent factor for HCC was higher in the HBeAg-negative group compared to the HBeAg-positive group (OR, 2.709, P < 0.001 vs OR, 1.817, P = 0.023). Notably, the number of mutations had a substantially higher odds ratio for HCC risk than AFP and platelet counts (OR, 2.087, P < 0.001 vs OR, 1.008, P = 0.005 and OR, 1.013, P = 0.002, respectively; Table 2). Therefore, the number of mutations at the eight target regions might be the most important viral factor for HCC risk prediction.

Hepatocellular carcinoma risk escalates according to the accumulated number of eight key mutations.  The prevalence of HCC increased in accordance with the accumulating number of mutations (P for trend <0.001) (Fig. S1). Of the eight key mutations, BCP (A1762T/G1764A) mutations tended to precede the occurrence of the other six mutations (Fig. S2). All mutations except A1762T/G1764A exhibited similar increasing patterns in their proportional percentage changes, in accordance with an increase in mutation number (Fig. S2). When the cumulative effects of MCG were evaluated through stratified analysis according to mutation number, MCG with four mutations were most prevalent (22.7%, 34/150) among study subjects. Double (24.0%) and triple (24.0%) mutations were most common in the non-HCC group, whereas multiple mutations at four or more sites were predominant (16.0–22.7%) in the HCC group (Table 3).

Table 3.   Comparison of the accumulated mutation numbers between patients with and without HCC
Number of mutationsCase number (n = 150)Non-HCCHCCOdds ratio (95% confidence interval)P-value†
Number (n = 75)%Number (n = 75)%
  1. †The risk of HCC as compared to that of 0–2 mutations. The significance of trends in HCC risk according to the number of mutations: P < 0.001. See Fig. S1, which shows a significant dose linear trend of the HCC proportion according to the number of accumulated mutations in total cases (n = 150). HCC, hepatocellular carcinoma.

3271824.0912.03.62 (0.97–13.52)0.055
4341722.71722.77.25 (2.09–25.12)0.002
521912.01216.09.66 (2.49–37.52)0.001
61711.31621.3116.00 (11.92–1128.12)<0.001
7–81811.31722.7123.25 (12.71–1194.91)<0.001

The wild-type genotype at the eight nucleotides in the X/precore regions was present in only six patients without HCC, whereas completely mutated forms (MCG with eight mutations) were detected in five patients, all of whom had HCC. In the analysis of MCG by mutation number, the risk of HCC was significantly lower in MCG with two or fewer mutations in the target region, with no case of HCC for individuals with genotypes having 0–1 mutations. Likewise, the risk of HCC in MCG with six or more mutations was extremely high, with the incidence of HCC in MCG having 7–8 mutations being 94.4% (17/18; Table 3).

Identification of mutation combination patterns related to a high risk of hepatocellular carcinoma. Figure 1 illustrates the prevalence of HCC according to the number of mutations for each individual mutation. Within the HCC group, the prevalence of each mutation, except for A1762T/G1764A, increased along with the number of mutations, and the overall risks of HCC were highest when the number of mutations in the X/precore regions was ≥6. In contrast, within the non-HCC group, the prevalence of mutation increased initially and decreased when mutation numbers were ≥5. Assuming that these findings suggest specific mutation combination patterns associated with HCC, we used a pattern analysis program to simplify the classification of 256 possible combination patterns formed across the eight highlighted positions. From a total of 55 MCG (Table S2), pattern analysis identified 37 and 33 patterns for HCC and non-HCC subjects, respectively, and their distribution patterns were substantially distinct between the two groups (Fig. S3A,B). Of the 37 mutation patterns for HCC, we could ultimately identify seven patterns that were found almost exclusively in the HCC group (Fig. 2A). These seven high-risk patterns accounted for 40.0% of all HCC cases, but only 2.7% of all non-HCC cases (Table 4). When mutation patterns were arrayed according to the number of mutations, HCC cases increased as the number of mutations increased, and were clearly differentiated from non-HCC cases when six or more mutations were present (Fig. 2B).

Figure 1.

 The prevalence of 8 key mutations in the X/precore regions according to the number of mutations. The horizontal axis indicates the number of accumulated mutations. The rates depict the percentage of patients with each specific mutation at each mutation number (n = 150). The most frequent mutations among all patients were G1764A (88.7%), A1762T (84.7%) and G1896A (59.3%). The positive rates of other mutations ranged from 19.3% to 43.3%. The distribution of each mutation according to mutation number appeared to be similar, showing low and high risk of HCC at the number of mutations of ≤2 and ≥6, respectively (a) G1613A, (b) C1653T, (c) T1753V, (d) A1762T, (e) G1764A, (f) A1846T, (g) G1896A and (h) G1899A.

Figure 2.

 Identification of high-risk mutation combination patterns for hepatocellular carcinoma (HCC) and their relationship with the number of multiple mutations. (a) Comparison of the frequency of cases between the HCC and non-HCC groups across the ranking order of mutation patterns associated with HCC. The solid line over the bar graph represents the total mutation number. All high-risk patterns for HCC had basal core promoter (BCP) double mutations and ≥2 additional mutations at specific sites. Detailed information on the seven high-risk patterns, including rank #1–7, is shown in Table 4. The combination patterns of multiple mutations are listed in Table S2. (b) Frequency of HCC and non-HCC cases in each mutation pattern arrayed by the rank of mutation number (solid line). HCC are almost exclusively found in settings of six or more mutations, a pattern well-differentiated from non-HCC.

Table 4.   Top seven mutation combination patterns and their amino acid changes in the HCC group: Comparison of frequency with the non-HCC group
RankMutation combination patternsMutation numbersHCC group (n = 75)Non-HCC group (n = 75)
  1. 0: wild, 1: mutant. †Amino acid change patterns are the same as those of nucleotide mutation combination patterns except for E80E and S11S, which are the silent mutations. ‡No amino acid changes.

Amino acid changes†E80E‡H94YI127
G29A Total40.0Total2.7
  HBx protein changes HBe protein changes     

Predictive performance of the number of mutations for hepatocellular carcinoma.  Based on the significant dose–risk associations between mutations and HCC, we evaluated the utility of mutation number as a potential biomarker for HCC and compared with that of AFP. Among the mutation number classes listed in Table 5, the presence of six or more mutations provided the best performance for predicting HCC, with 97.3% specificity and 94.3% positive predictive value, and was even comparable to that of AFP. The AUROC for mutation number and AFP, for prediction of HCC, were 0.824 (95% CI: 0.759–0.890) and 0.869 (95% CI: 0.812–0.925), respectively (Fig. S4). Interestingly, there was a linear increasing trend in the proportion of cases with AFP ≥ 20 ng/mL, with rising numbers of accumulated mutations among all patients (Table S3).

Table 5.   Predictive performance of the number of combined mutations in HBV genomes for HCC (in comparison with AFP)
 HCC (= 75)Non-HCC (= 75)Sensitivity (%) (95% CI)Specificity (%) (95% CI)PPV (%) (95% CI)NPV (%) (95% CI)
  1. AFP, α-fetoprotein; CI, confidence interval; NPV, negative predictive value; PPV, positive predictive value.

Number of mutations
 3 or more714694.7 (88.0–98.2)38.7 (32.0–42.2)60.7 (56.4–63.0)87.9 (72.6–96.0)
 4 or more622882.7 (74.6–89.2)62.7 (54.6–69.2)68.9 (62.2–74.3)78.3 (68.2–86.5)
 5 or more451160.0 (52.1–66.1)85.3 (77.4–91.4)80.4 (69.7–88.5)68.1 (61.8–73.0)
 6 or more33244.0 (37.8–46.2)97.3 (91.1–99.5)94.3 (81.0–99.0)63.5 (59.4–64.9)
AFP cut-off
 ≥20 ng/mL46661.3 (54.8–65.4)92 (85.5–96.1)88.5 (79.1–94.3)70.4 (65.4–73.5)

Validation analysis.  Notably, in a longitudinal cohort of 25 patients with serial serum samples spanning the years before and after HCC diagnosis (Fig. 3), most of the patients with HCC (24/25, 96.0%) already had an increasing number (≥4) of the X/precore mutations years before HCC development. All exhibited an equal or increasing number of mutations in the X/precore regions until HCC development. Most of the longitudinal HCC cases showed an accumulation of mutations at positions 1653, 1753, 1846 or 1896, in addition to BCP double mutations (data not shown). Otherwise, cross-validation on the independent set of 52 cases retrieved from the GenBank database (accession numbers GQ475305GQ475357)(24) confirmed a much higher number of mutations (≥4 or 5 mutations) in the X/precore regions in patients with HCC than those without (Table S4; correlation analysis, P = 0.0057).

Figure 3.

 Longitudinal observation of the number of mutations in the X/precore regions in hepatocellular carcinoma (HCC) patients. In most HCC cases (96.0%), a high number of mutations (≥4) are already present in the premalignant period. No., number of accumulated mutations; AFP, α-fetoprotein (ng/mL); 0, HCC development; *Liver transplantation.


This study examined the association between multiple mutations in the X/precore regions of the HBV genome and HCC in age-, sex- and HBeAg status-matched genotype C2 carriers. The key finding of this study is that MCG consisting of the number and pattern of multiple mutations in the X/precore regions showed the additive combined effects that relate to HCC occurrence and might play a significant role in the prediction of HCC risk. The utility of MCG as an indicator of HCC is supported by the following evidence from this study: (i) direct increasing trends for HCC with increasing mutation number; (ii) identification of the top seven high-risk mutation combination patterns for HCC; and (iii) statistical evidence for the high diagnostic power of MCG for HCC. We summarize that analysis of MCG in the X/precore regions might be helpful for early prediction of HCC risk and screening for HCC in chronic carriers with genotype C2.

Importantly, our study revealed that the number of the eight target mutations in the X/precore regions was the only independent viral factor for HCC, irrespective of HBeAg status. This yielded a much higher level of statistical significance in estimated OR for association with HCC than that observed with AFP and platelet counts. Although the well-known individual mutations (C1653T, T1753V, G1764A and A1846T) showed certain levels of correlation with HCC, they were not independently predictive of HCC when analyzed in the context of all clinico-virological factors, including HBV DNA levels, mutations at eight key positions, number of mutations, and the levels of ALT, AFP and platelets. Further stratification analyses indicated that MCG containing six or more mutations were specifically associated with HCC, with linearity in the accumulated dose–risk relationship between mutation number and HCC. This suggests that the total number of mutations at the eight target positions is more important for predicting the risk of HCC than any individual mutation.

It is noteworthy that accumulated mutations in the X/precore regions showed good diagnostic properties for HCC. Surprisingly, the number of mutations (≥6) in the X/precore regions was an excellent diagnostic indictor for HCC, comparable to AFP. The diagnostic utility of this index is encouraging, given that the reported sensitivity and specificity of AFP alone for detecting HCC are generally <50% and 90%, respectively.(1,25) With a scale of only 0 (wild-type) to 8 (completely mutated type), the mutation number offered a good AUROC of 0.824 for predicting HCC, which is comparable to the utility of AFP when a much greater range of values is involved. One advantage of mutation number as a diagnostic indicator is that there is a linear dose–risk relationship between mutation number and HCC, even when five or fewer mutations are present. These results suggest the feasibility of this new approach to cancer screening based on HBV mutations in chronic carriers, which has not been explicitly revealed thus far.

The pattern analysis of the X/precore mutations showed a distinct difference in the distribution of the seven high-risk mutation combination patterns between the HCC and non-HCC groups (40.0%vs 2.7%). All of the high-risk patterns contained BCP double mutations, and at least two additional mutations at nucleotides 1653, 1753, 1846, or 1896. Although the exact biological effects of MCG with these seven high-risk mutation patterns is not clear, we assume that the addition of H94Y (due to C1653T) and/or I127N/S/T (due to T1753V) substitutions in HBx protein to K130M and V131I (due to A1762T/G1764A mutations) amino acid changes synergistically acts to promote hepatocarcinogenesis through significant alteration of the function of X protein and interference with cell growth control and DNA repair.(26)

In addition to the BCP mutations,(27,28) the silent A1846T mutation has been recently associated with HCC, although its carcinogenic function is unclear.(14,29) Although each selected mutation alone might not be sufficiently associated with HCC, the synergistic effects of multiple mutations could generate the maximum risk of developing HCC. Indeed, the significant individual mutations, such as C1653T, T1753V and A1846T, only accounted for 30–50% of HCC patients in our study, which is similar to the findings in other reports.(14,30,31) Our study is the first to show clearly that the analyses of mutational number and patterns of the eight target nucleotides in the X/precore regions might be highly useful as a screening tool for predicting HCC risk in advance.

X/precore regions are very subtle and complicated parts in all HBV genotypes, because they contain various overlapping functional sequence arrangements. Our observation of a causal relationship between MCG and HCC risk is supported by experimental research, in which the combo mutant (A1762T/G1764A [TA], T1753A and T1768A), but not HBx with single or double core promoter mutations, accelerated p21(WAF1/CIP1) degradation by the upregulated expression of S-phase kinase-associated protein 2 and increased cyclin E expression in primary hepatocytes and HepG2 cells, resulting in that the combo mutant accelerated cell cycle progression.(32) These mechanisms might explain how the accumulation of X/precore mutations escalates the risk of HCC in chronic HBV carriers. The significant association of MCG formed from the eight target mutations in the X/precore regions with the risk of HCC might not be confined to subgenotype C2 analyzed in our study. It is hypothesized that our results will be generally applicable to other HBV genotypes.

Validation using a longitudinal cohort of independent cases and a group of HBV DNA sequences registered in the GenBank database revealed the significant correlation between four or more of the eight key mutations and the presence of HCC. This supports that the analysis of MCG could be a potential predictive biomarker for HCC.

In conclusion, our study strongly suggests that genotypes of mutation complexes in the X/precore regions are a novel risk indicator that might facilitate early prediction of HCC in chronic carriers. In particular, acquisition of MCG containing the seven high HCC-risk mutation combination patterns and/or six or more overall mutations might identify patients who are at very high risk of developing HCC. Risk stratification based on MCG analyses would be useful in determining patients who are at high risk of developing HCC, and improve tailored screening schemes for HCC.


The research was supported by the Converging Research Center Program through the Ministry of Education, Science and Technology (grant number: 2011K000886) and the Bilateral International Collaborative R&D Program through the Ministry of Knowledge and Economy, Korea.

Disclosure Statement

There are no conflicting interests in this work.