Prevention of hepatitis B virus–related hepatocellular carcinoma with antiviral therapy


  • Ching-Lung Lai,

    Corresponding author
    1. Department of Medicine, The University of Hong Kong, Queen Mary Hospital, Hong Kong
    2. State Key Laboratory for Liver Research, University of Hong Kong, Queen Mary Hospital, Hong Kong
    • Department of Medicine, The University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong Kong
    Search for more papers by this author
    • fax: 852 28162863

  • Man-Fung Yuen

    1. Department of Medicine, The University of Hong Kong, Queen Mary Hospital, Hong Kong
    2. State Key Laboratory for Liver Research, University of Hong Kong, Queen Mary Hospital, Hong Kong
    Search for more papers by this author

  • Potential conflict of interest: Dr. Lai received speakers' fee from Bristol-Myers Squibb, Gilead, and GlaxoSmithKline. Dr. Man-Fung received grants from Roche and Novartis. He also received grants from and speakers' fee from Bristol-Myers Squibb and Gilead.


Chronic hepatitis B (CHB) infection is the major cause of hepatocellular carcinoma (HCC). Primary prevention of hepatitis B infection by vaccination is effective in reducing the incidence of HCC. In persons with CHB infection, the two accepted treatment modalities are interferon alpha (IFN-α) given subcutaneously for a limited period and nucleoside/nucleotide analogs given orally on a long-term basis. These treatments are effective in suppressing viral activity and improving disease markers in short-term studies. The long-term effect on the development of liver cancers with these two forms of treatment appears to be different. However, there are no studies directly comparing IFN-α and nucleoside/nucleotide analogs. Comparisons across studies are inevitably limited by differences in the baseline characteristics of the study cohorts. Long-term follow-up studies of IFN-α therapy show inconsistent results. The beneficial effect in reducing the development of liver cancer is observed mainly in treatment responders who have preexisting cirrhosis of the liver. The long-term studies of lamivudine (and adefovir) show a consistent reduction in the development of liver cancers in patients with, and without, cirrhosis. This beneficial effect is blunted by the development of resistance. The effects of the newer nucleoside/nucleotide analogs, with higher potency and minimal risk of resistance development, are, as yet, unknown. (HEPATOLOGY 2013)

Approximately 350 million people are carriers of the hepatitis B virus (HBV).1 Twenty-five to forty percent of these carriers will develop complications of cirrhosis and hepatocellular carcinoma (HCC).2 According to the World Health Organization (2011), HCC is the third-most common cause of cancer death, following carcinomas of lung and stomach.3 The estimated annual death rate is 700,000. Seventy-eight percent of HCCs are related to chronic hepatitis B (CHB) and chronic hepatitis C (CHC) infections, occurring approximately in the ratio of 7 to 3, respectively. Seventy-five percent of HBV carriers are from Asian countries. In these countries, the incidence of HCC has remained high, ranging from 14 to 36 per 100,000 men in the past two decades.4

There are two ways to reduce the global incidence of HCC: (1) universal HBV vaccination to reduce the rate of new infection and (2) minimizing the risk of HCC in existing CHB carriers.

The effect of HBV vaccination in preventing the development of HCC is well documented. After implementing nationwide HBV vaccination in 1984 in Taiwan, a study in 1997 showed that the incidence of HCC in children 6-9 years of age declined from 0.52 per 100,000 between 1974 and 1984 to 0.13 per 100,000 between 1984 and 1986 (P < 0.001).5 A second 20-year follow-up report in 2009 showed that HCC incidence among subjects 6-19 years of age was lower in the vaccinated cohort (64 HCC in 37,709,304 person-years), compared to the nonvaccinated cohort (444 HCC in 76,496,406 person-years), with the adjusted relative risk (RR) being 0.31 (P < 0.001).6

For persons who are already CHB carriers, maximum control of viral replication may reduce the risk of development of HCC. The seven approved agents for the treatment of CHB, which include two immunomodulators (conventional interferon alpha [IFN-α] and pegylated IFN-α [Peg-IFN-α]) and five nucleoside/nucleotide analogs (lamivudine, adefovir, entecavir, telbivudine, and tenofovir), have proven short-term efficacy.

The current review is a comprehensive critical review of the data on the long-term efficacy of antiviral treatment in the prevention of HBV-related HCC. Data were identified from Medline, Current Contents, and references from articles from January 1991 to December 2011; numerous articles were also identified from the extensive files of the investigators. Search terms were “HBV,” “hepatitis B virus,” “treatment,” “liver cancer,” “hepatocellular carcinoma,” “long-term outcome,” “prevention,” “interferon-alpha,” “antiviral,” “lamivudine,” “adefovir,” “telbivudine,” “entecavir,” and “tenofovir”. The search was restricted to peer-reviewed English-language articles of studies with control arms receiving no treatment for comparison. Studies without control arms examining the effect of treatment on the risk of HCC are also included.


Ab, antibody; ALT, alanine transaminase; anti-HBc, Ab against hepatitis B core antigen; anti-HCV, Ab against HCV; CHB, chronic hepatitis B; CHC, chronic hepatitis C; CI, confidence interval; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HLA, human leukocyte antigen; HR, hazard ratio; IFN-α, interferon alpha; PCR, polymerase chain reaction; Peg-IFN-α, pegylated IFN-α; RR, relative risk; YMDD, tyrosine methionine aspartate aspartate.

Prevention of HBV-Related HCC by IFN-α Treatment

All the studies of the effect of IFN-α on the subsequent development of HBV-related HCC were with traditional IFN-α, because Peg-IFN was only licensed relatively recently and meaningful long-term follow-up studies have yet to be published. The usefulness of IFN-α in preventing HBV-related HCC, unlike its effect on CHC-related HCC, is widely divergent in various studies.

Table 1 summarizes the details of 10 studies of the effect of IFN-α treatment on the development of HCC. Only one study11 calculated the RR for HCC development of IFN-α-treated patients, compared to controls. With the exception of the studies of Mazzella et al.12 and Yuen et al.,16 the assays for the detection of HBV DNA were insensitive by today's standards. The majority of the studies used hepatitis B e antigen (HBeAg) seroconversion as the endpoint of treatment. As shown in the studies of Truong et al.,15 Papatheodoridis et al.,15 and Yuen et al.,16 most patients continued to have detectable HBV DNA after HBeAg seroconversion. This is the most likely reason for the inconsistent results concerning the reduction of HCC development after IFN-α treatment. Further critical comments on each individual study are given below.

Table 1. Summary of 10 Studies of IFN-αTreatment on Development of HBV-Related HCC
AuthorsNature of StudyNumber of Patients (Treated Versus Controls)% With Cirrhosis (Treated Versus Controls)Treatment DetailsHBV DNA Assay and Detection LimitDuration of Follow-up in Years (Treated Versus Controls)HCC Development (% Treated Versus % Controls)P ValueComments
  1. pred, prednisolone; NS, not significant; NA, not available.

Lin et al.7Prospective, randomized, controlled121; all HBeAg+ (67 versus 34)10.3 versus14.7Lymphoblastoid IFN × 12 weeks ± pred × 4 weeksSlot-blot hybridization; 1.5 × 105 copies/mL7.4 versus 6.51.5 versus 11.80.043• Prospective controlled trial
• Number of patients relatively small
Mixed treatment regimes
Lin et al.8Matched controls466; all HBeAg+ (233 versus 233)8.1 versus 10.7Lymphoblastoid/ recombinant IFN 11-28 weeks ± predSlot-blot hybridization; 200 pg/mL, ∼6 × 107 copies/mL6.8 versus 6.12.1 versus 6.9<0.025• Highly mixed treatment regimes
• IFN only decreased HCC in patients with preexisting cirrhosis
Conclusions in conflict with above study (see text)
Ikeda et al.9Unmatched controls313; 36% HBeAg+ (94 versus 219)100 versus 100Lymphoblastoid/ recombinant IFN × 10 monthsBranched DNA 0.55 × 106 copies/mL6.8 versus 7.017.0 versus 30.80.012• Highly variable treatment duration
• All patients had cirrhosis
Krogsgaard10Multicenter controlled trials with questionnaire follow-up308; all HBeAg+ (207 versus 98) (discrepancy with patient numbers)5.6 (no separate data for treated patients and controls)Recombinant IFN, 12-24 weeksSpot hybridization, (qualitative assay)4.7 (no separate data for treated patients and controls)0.96 versus 1.0NS• Highly mixed treatment regimes
• Follow-up of only 66% of original population
Equal proportion of treated patients and controls developed cirrhosis (10%)
International IFN-alpha HCC Study Group11Retrospective multicenter with matched controls146; HBeAg status NA (49 versus 97)88 versus 94 (including patients positive for anti-HCV)Lymphoblastoid/ recombinant/ leucocyte IFN × 3-30 monthsAbbott assay (no other details mentioned)5.7 versus 6.916 versus 19NS• Mixed HBV and HCV patients
IFN useful in reduction of HCC in HCV patients without HBV markers
Mazzella et al.12Prospective, randomized, controlled64; all HBeAg (33 versus 31)0 versus 0Lymphoblastoid IFN × 24 weeksPCR assay; 100 copies/mL7.2 versus 6.63.0 versus 6.4NS• Prospective controlled trial
• Number of patients relatively small
No patients with cirrhosis
Tangkijvanich et al.13Retrospective, unmatched controls139; all HBeAg+ (67 versus 72)17.9 versus 22.2Lymphoblastoid/ recombinant IFN × 20-24 weeksNA4.9 versus 5.04.7 versus 12.5NS• Unmatched controls
• HBV DNA probably not performed
New cirrhosis occurred less in sustained IFN responders than in controls (P = 0.04)
Papatheodoridis et al.14Retrospective, unmatched controls404; all HBeAg (209 versus 195)27.3 versus 34.9Recombinant IFN × 14-16 weeksLiquid hybridization assay; 1.5 pg/mL, 4.2 × 105 copies/mL6.0 versus 6.18.1 versus 7.7NS• Unmatched controls
Sustained responders better survival than nonsustained responders (P = 0.027) and controls (P = 0.048)
Truong et al.15Retrospective, matched controls67; 59.6% HBeAg+ (27 versus 35)NARecombinant IFN × 4-24 weeksPCR assay; 2.6 log copies/mL2-14 (no separate data for treated patients and controls)3.7 versus 0.0NS• Number of patients relatively small
• Highly variable treatment duration
• No clear duration of follow-up
Yuen et al.16Prospective follow-up, matched controls411; all HBeAg+ (208 versus 203)14.3% of the adult treated patients; no data for controlsRecombinant IFNDot-blot hybridization, followed by PCR assay; ∼1,600 copies/mL8.9 versus 9.02.9 versus 0.0NS• Highly mixed treatment regimes
12-24 weeks ± pred• More liver-related complication with IFN treatment versus controls: 4.3% versus 1.0% (P = 0.062)

The three studies showing IFN-α to be of use in preventing HBV-related HCC will be discussed first. In one of the earliest reports, Lin et al. followed up 101 HBeAg-positive male Taiwan patients.7 This is one of two randomized studies on the long-term effects of IFN-α treatment. Baseline median alanine transaminase (ALT) levels were unusually high for an Asian population, being 221 U/mL for the treated patients and 256 U/mL for the controls. Of importance with reference to the next study, 10.3% of treated patients and 14.7% of the controls had cirrhosis on recruitment. There were no differences in the development of new cirrhosis and cirrhosis complications between the two groups of patients. However, HCC developed in only 1 of the 67 (1.5%) treated patients, compared to 4 of the 34 controls (12%) (P = 0.043). Survival was also significantly better in the treated patients (P = 0.018). Multivariate analysis showed that IFN-α treatment, preexisting cirrhosis, and age were independent factors affecting survival.

The same group of investigators subsequently extended their study population by including patients from other nonrandomized studies in Taiwan, making a total of 233 IFN-treated patients and 233 matched controls.8 Cirrhosis was present in 8.1% of the treated patients and 10.7% of the controls. Overall, the incidence of HCC was lower in the treated population (2.7%) than in the controls (12%) (P < 0.01). However, when patients were subdivided into those with cirrhosis and those without cirrhosis, reduction in HCC was only observed in patients with preexisting cirrhosis (19.7% for treated patients and 58.9% for controls; P = 0.0086), but not in the patients without cirrhosis (2.1% for treated patients and 2.3% for controls). This calls into question the earlier study from the same group of investigators. Did the reduction in HCC development of their previous study also apply only to the 10%-15% of patients with preexisting cirrhosis? In addition, the earlier study showed that IFN-α treatment had no effect on the development of new cirrhosis and cirrhosis complications. So, the apparent effect of IFN-α in reducing the development of HCC is paradoxical, especially because the second study shows that IFN-α only reduced HCC in patients with preexisting cirrhosis. In both studies, the relationship between HBV DNA suppression and HCC development with IFN-α treatment is unclear.

The third and last study showing IFN-α to be of use in reducing HCC incidence was conducted only in patients with cirrhosis.9 Ninety-four (30.0%) were treated with lymphoblastoid/recombinant IFN-α “intermittently” for a median of 10 months. Some patients were treated for over 5 years when treatment was regarded as effective in normalizing ALT levels. IFN-α treatment was found, by both uni- and multivariate analysis, to significantly decrease the rate of HCC development, being 4.5%, 7.0%, and 17% at the end of 3, 5, and 10 years, respectively, for the treated patients and 13.3%, 19.6%, and 30.8% for the untreated patients (P = 0.0124). In this study, HCC reduction with IFN-α treatment was much more significant in patients with viral load higher than 106 copies/mL. Of the three studies, this is the only study showing the possible relationship between IFN-α treatment, HBV DNA suppression, and HCC development. But, the study is marred by the lack of a definite IFN-α treatment protocol as well as the lack of patients without cirrhosis.

All the remaining seven studies did not show any beneficial effect in HCC prevention. Krogsgaard followed up 469 HBeAg-positive patients from three multicenter European trials for survival (340 patient) and response (308 patients).10 The trials were individually randomized to receive differing doses of IFN-α or no treatment. There was no difference in progression to cirrhosis in the treated and untreated groups (18 of 187 [treated; 10%] versus 9 of 89 [untreated; 10%]). Three (1%) patients developed HCC during follow-up: 2 received treatment, and 1 was untreated. The investigators commented that they failed to establish a presumed beneficial effect of IFN-α on disease progression and survival because of the incomplete follow-up (66%-73%), the variable follow-up period, and the substantial percentage of nonresponders. This last reason is of obvious importance, because the response rate of the IFN-α trials was only 28% at the end of the original trials (compared to 22% in the control patients). With long-term follow-up of a median of 4.7 years, the response rate of the treated and untreated patients was 86% and 91%, respectively. In this study, IFN-α treatment did not have any effect on response rate, progression to cirrhosis, or development of HCC.

The International Interferon-alpha Hepatocellular Carcinoma Study Group recruited 913 patients positive for antibody (Ab) against the hepatitis C virus (anti-HCV) and/or hepatitis B surface antigen (HBsAg) from 21 centers in Italy and Argentina retrospectively.11 They found that IFN-α lowered the rate of HCC development for patients with HCV with no HBV markers (5% of treated patients versus 20% of untreated; RR, 6.28; 95% confidence interval [CI]: 1.65-23.8). This benefit was not observed in HCV patients positive for Ab against hepatitis B core antigen (anti-HBc). In the 146 HBV patients, 16% of treated and 19% of untreated patients developed HCC. The RR of HCC in untreated versus treated HBV patients was 0.98 (CI: 0.33-2.92).

Mazzella et al. followed up 64 HBeAg-positive Italian patients.12 Though IFN-α treatment accelerated loss of HBeAg (90.1% in treated and 61.3% in control patients; P < 0.007), 1 treated patient with cirrhosis and 2 controls with cirrhosis developed HCC. The investigators postulated that the similar long-term outcome for the two groups of patients might be the result of persistence of the virus, with HBV DNA being positive by polymerase chain reaction (PCR) assay in 80% of treated patients who remained HBsAg positive and 10% in those who cleared HBsAg.

In a retrospective study, Tangkijvanich et al. followed up 139 HBeAg-positive patients in Thailand.13 HBV DNA levels were apparently only measured in treated patients before treatment. The two populations were unmatched for ALT levels, being higher in the IFN-treated group (P = 0.001). New cirrhosis occurred in 4.2% of sustained responders (n = 24), 14% of nonresponders (n = 43), and 22.2% of controls (P = 0.04 for sustained responders versus controls). Two IFN-α nonresponders and 9 controls developed HCC (P > 0.05). Though the investigators claimed that IFN-α reduced the incidence of HCC in sustained responders, the unmatched and retrospective nature of the study, the relatively small number of responders, and the lack of significant P value render any firm conclusions impossible.

Papatheodoridis et al. followed up prospectively 406 HBeAg-negative patients in Greece.14 The patients were not randomized. Treated patients had significantly higher baseline ALT levels (P < 0.001). IFN-α treatment did not have any apparent effect on survival and complication-free survival, compared to untreated patients (P = 0.62 and 0.14, respectively). HCC occurred in 17 (8.1%) treated (16 in nonresponders) and 15 (7.75%) untreated patients. The investigators noted that most of the treated and untreated patients had residual HBV DNA, even in IFN-α responders with HBsAg seroclearance.

Truong et al. followed up 62 Japanese patients.15 Once again, the IFN-treated group had significantly higher baseline ALT levels, compared to untreated patients (P < 0.05). There was no difference between the two groups of patients with respect to HBeAg seroconversion, ALT normalization, and undetectable HBV DNA. Three treated patients and 2 controls developed cirrhosis. One treated patient developed HCC; none of the controls developed HCC.

The final study is the only case-controlled study of IFN-α versus no treatment. Yuen et al. followed up 208 HBeAg-positive patients (27 of 189 [14.3%] with biopsy-proven cirrhosis) treated with IFN-α for 12-24 weeks and compared them to 203 controls who fulfilled the inclusion criteria of the treated patients, but were not treated.16 As in the study of Truong et al.,15 there was no difference between the two groups of patients in terms of HBeAg seroconversion rate and undetectable HBV DNA by PCR assay in the patients who became anti-HBe positive (88.6% PCR positive in the treated group and 91.1% in the control group). Nine treated patients (4.3%) developed cirrhosis complications, including 5 with HCC, whereas only 2 untreated patients (1.0%) developed cirrhosis complications; none had HCC (P = 0.062).

To date, there are six meta-analyses of the effect of IFN-α treatment on the development of HBV-related HCC. (Where HCV-related HCCs are also analyzed, only the HBV-related HCC sections are selectively considered.) The studies of Cammà et al.17 and Miyaki et al.18 calculated the risk difference, whereas the other four studies calculated the RR.19-22 Details of the six meta-analyses are shown in Table 2.

Table 2. Summary of Six Meta-analyses on the Effect of IFN-αVersus no Treatment on Development of HBV-Related HCC
AuthorsNumber of StudiesNumber of Patients Treated Versus ControlsRR/Risk Difference* (95% CI)P ValueComments
  1. NS, not significant.

Cammà et al.177853 versus 652 (all cirrhotic patients)4.8%* (0.11-0.015)NSSignificant inconsistency in Oriental studies; the European studies showed no preventive effect with IFN-α.
Miyake et al.188553 versus 7505.0%* (9.4-0.5)0.028Effect not shown in European studies
Sung et al.19121,292 versus 1,4580.66 (0.48-0.89)0.006No effect in patients without cirrhosis
Yang et al.20111,006 versus 1,0760.59 (0.43-0.81)0.001Patients with healthy ALT excluded
Zhang et al.212176 versus 1710.23 (0.05-1.04)NS (0.056)Only included two randomized control trials for analysis
Jin et al.2291,291 versus 1,0480.274 (0.059-1.031)NS 

The conclusions show inconsistent effects of IFN-α treatment on the reduction of HCC, most probably related to its moderate suppression of HBV replication. At least four studies with patients without cirrhosis show that IFN-α has no effect on the development of cirrhosis.7, 10, 15, 16 The lack of beneficial effect in patients without cirrhosis may be partly attributed to the lower rate of development of HCC in these patients. However, in the studies where cirrhosis status was known, there were over 740 treated patients and 640 controls who did not have cirrhosis. Much larger studies with greater statistical power will be required to demonstrate any positive benefits in patients without cirrhosis. IFN-α treatment may be of benefit in preventing HCC in patients with preexisting cirrhosis who are also sustained responders to the treatment. It should be noted that IFN-α may lead to severe liver decompensation when used in patients with cirrhosis, and that less than 35% of treated patients are sustained responders.

One final study without a control arm confirms the above-mentioned findings. In this study, 33% of 165 HBeAg-positive patients were responders (with some patients receiving up to five courses of IFN-α). Seven of eight patients who developed HCC were nonresponders/relapsers after IFN therapy.23 All current guidelines suggest that nonresponders to IFN-α should receive nucleoside/nucleotide analog therapy.

Prevention of HBV-Related HCC by Nucleoside/Nucleotide Analog Treatment

The beneficial effect of nucleoside/nucleotide analog treatment in the prevention of HBV-related HCC is more clear cut than with IFN-α. However, the only published results to date are with the earlier generation of drugs (lamivudine and adefovir). No trials have been published with the two current first-line agents (entecavir and tenofovir). Any studies published in the future concerning the prevention of HCC with entecavir and tenofovir is unlikely to be controlled because of the great proven efficacy of these two agents in improving viral and clinical parameters other than HCC. It would not be ethically justifiable to perform placebo-controlled studies or even comparison studies with lamivudine and adefovir.

Table 3 summarizes the details of five studies of the effect of lamivudine (and adefovir) treatment on the development of HCC. In three of the studies,25, 27, 28 sensitive assays were used for the measurement of HBV DNA levels, showing that even with a relatively weak nucleoside analog such as lamivudine, the HBV DNA suppression was potent. The effect is obviously marred by the development of the lamivudine-resistant tyrosine-methionine-aspartate-aspartate (YMDD) mutations. Further critical comments on each study are given below.

Table 3. Summary of Six Studies of Nucleoside/Nucleotide Analog Treatment on Development of HBV-Related HCC
AuthorsNature of StudyNumber of Patients (Treated Versus Controls)% With Cirrhosis (Treated Versus Controls)Treatment DetailsHBV DNA Assay and Detection LimitDuration of Follow-up in YearsHCC Development (% Treated Versus % Controls)P ValueComments
(Treated Versus Controls)
  1. NA, not available.

Liaw et al.24Prospective, randomized, placebo controlled651; 58% HBeAg+ (436 versus 215)31 versus 39 (all had Ishak's fibrosis score > 4 of 6)Lamivudine 100 mg versus placeboBranched-chain hybridization assay; 0.7 × 106 copies/mL2.7 (no separate data for treated patients and controls)3.9 versus 7.40.047 (HR, 0.49)• Prospective controlled trial
• When the 5 patients with HCC in the first year were excluded, HR changed to 0.47 and P = 0.052.
49% treated patients had YMDD mutations.
Yuen et al.25Matched controls266; all HBeAg+ (142 versus 124)0 versus 0Lamivudine 100 mg (some received lamivudine 25 mg for 1 year) versus no treatmentPCR assay; 35 copies/mL7.49 versus 8.990.7 versus 2.40.005P value was for both cirrhosis development and HCC development.
• 76.3% treated patients had YMDD mutations.
Matsumotoet al.26Retrospective, matched case-control study754; 54.7% HBeAg+ (377 versus 377)17.2 versus 17.8Lamivudine 100 mg versus no treatmentNA2.7 versus 5.31.1 versus 13.3<0.001Neither HBV DNA nor YMDD mutations were mentioned.
Eun et al.27Retrospective, historical controls1,501; 88.7% HBeAg+ (872 versus 699)47.4 versus 37.2Lamivudine 100 mg versus no treatmentBranched DNA assay; 2,000 copies/mL; later PCR assay; 70 copies/mL4.7 versus 5.76.65 versus 9.870.02• Controls not matched for age, ALT, % HBeAg+, HBV DNA, and duration of follow-up
47.6% (415) had virologic breakthrough; YMDD present in 83.8% of the 253 patients tested
Papatheodoridis et al.28Retrospective, historical controls404; all HBeAg (209 versus 195 control versus 209 IFN-α patients)31.8 versus 34..0 (versus 27.3 for IFN-α patients)Lamivudine 100 mg (with adefovir rescue) versus no treatment (also versus IFN-α)PCR assay; 1,000 copies/mL3.8 versus 6.1 (versus 6.0 for IFN-α patients)1.49 versus 7.69 (versus 6.69 for IFN-α patients)0.04 (0.036 versus IFN-α non responders)P values are for liver-related survival.
• Survival same for lamivudine patients and IFNα sustained responders (i.e., 27.3% of IFN-α-treated patients)

The first study of the long-term effect on lamivudine was the only prospective placebo-controlled study in 651 HBV patients with biopsy-proven cirrhosis or advanced fibrosis.24 The study was terminated early (at 32.4 months) because of significant beneficial effects in the treatment group, with 7.8% developing cirrhosis complications (versus 17.7% in the placebo group; hazard ratio [HR], 0.45; P = 0.001). The effect on HCC prevention was less obvious, probably because of the early termination of the trial, especially when the 5 patients who developed HCC within the first year of the study were excluded, assuming that these patients might have preexisting undetected HCC on recruitment to the study.

In a similar study of the long-term effects of lamivudine treatment in 142 HBeAg-positive Hong Kong Chinese controls without cirrhosis versus 124 untreated matched controls, the results were almost identical, though with a longer median follow-up of 8 years, a longer follow-up being necessary for patients without cirrhosis to develop complications.25 Though YMDD mutations developed in 76.6% of patients over the 8 years of follow-up (detected in 7 of the 8 treated patients with complications), patients with the YMDD mutations still had significantly less complications than the untreated patients (P = 0.024).

There are three more studies. In a retrospective study of 657 lamivudine-treated Japanese patients versus 2,138 untreated patients, Matsumoto et al. found seven factors that predisposed patients to HCC development: no treatment, male sex, family history of HBV carriage, age >40 years, fibrosis of over grade 2 of 4, albumin level of <40 g/L, and platelet count of <150,000/mm3.26 All patients had liver biopsies, but HBV DNA measurement was not mentioned. In a further matched case-control study (summarized in Table 3), the annual incidence rate of HCC was reduced from 2.5% in the untreated to 0.4% in the treated patients (P < 0.001).

In a retrospective study of 872 Korean patients versus 699 historical controls, annual incidence of HCC was reduced from 4.1% in the controls to 0.95% in the sustained responders to lamivudine (P = 0.005).27 Annual incidence of HCC rose to 2.18% in patients with viral breakthrough and 5.26% in those with suboptimal response. The investigators found that lamivudine had no protective effect in patients with no cirrhosis or decompensated cirrhosis. The same center published a preliminary abstract 5 years earlier,28 from which the final report was presumably based. (The significance of the abstract and the full report will be mentioned below.)

Another study followed up 201 lamivudine-treated HBeAg-negative patients from Greece and compared them with historical cohorts, 209 treated with IFN-α, and 195 untreated patients.29 Adefovir was given as rescue therapy in 79 of 109 lamivudine-treated patients without virological remission. The event-free survival for the lamivudine-treated patients at 60 months of follow-up was 92%. The liver-related survival of the lamivudine-treated patients was better than the untreated cohort (P = 0.05) and the nonsustained responders to IFN-α (P = 0.36), but similar to the IFN-α sustained responders.

There are two meta-analyses and one systematic review with analysis of pooled data of the long-term effect of nucleosides/nucleotides on HBV-related HCC. In the meta-analysis of Sung et al., five studies were analyzed, with 1,267 treated and 1,022 untreated patients. (The data from the abstract of the Korean center was used.) They found that the use of nucleoside analogs (mostly with lamivudine, with some patients receiving adefovir rescue), treatment reduced the development of HCC from 11.7% in controls to 2.5% (P = 0.01; RR, 0.22; CI: 0.10-0.50).19 All five studies showed benefit with nucleoside/nucleotide treatment. The protective effect applied to patients with cirrhosis (3.9% in treated patients and 22.4% in controls), patients without cirrhosis (1.8% in treated patients and 8% in controls), and even to patients with drug resistance (3.3% in treated patients and 6.4% in controls). There are no studies directly comparing nucleoside analogs and IFN-α in reduction of the risk of HCC. Compared to controls, the risk reduction for nucleoside analog treatment was 78% and for IFN-α treatment was 34%. This greater reduction may be related to the more-profound viral suppression with nucleoside analog therapy. However, comparisons of the rate of reduction across studies (and meta-analyses) are inevitably limited by differences in the baseline characteristics of different study cohorts.

The second meta-analysis included six studies, with 2,035 treated and 1,609 untreated patients.30 They, however, mistakenly included both the data from the preliminary abstract28 and the subsequent full report from the same Korean center27 as two separate studies. The results from this meta-analysis therefore will not be separately discussed.

In the systematic review with pooled data analysis, two randomized and 19 cohort studies were analyzed, with 3,881 treated and 534 untreated patients.31 HCC occurred less frequently in the treated than in the untreated patients (2.8% and 6.4%, respectively; P = 0.003). This review included seven studies deemed to be of low quality (with inadequate criteria for the diagnosis of HCC) and 11 small studies (with less than 100 patients). Among the treated patients, HCC occurred more frequently in patients with cirrhosis than in those without cirrhosis, 10.8% and 0.5%, respectively, for treatment-naïve patients (P < 0.001), and 17.6% and 0%, respectively, for lamivudine-resistant patients (P < 0.001). Treatment-naïve patients with virologic response were less likely to develop HCC than those without virologic response (2.3% and 7.5%, respectively; P < 0.001). Though nucleoside/nucleotide analog treatment lowered the frequency of HCC development, HCC still occurred in some patients during virologic remission.

A final recent study, without a control arm, emphasized the same point. This is a nationwide study from Greece retrospectively analyzing 818 HBeAg-negative patients on lamivudine treatment with various forms of rescue therapy in 402 patients after virologic breakthrough or when there was lack of response to lamivudine.32 HCC developed in 49 (6.0%) patients with a median follow-up of 4.7 years. Risk of HCC was higher in patients with cirrhosis and in older patients. The chief message of this study is that on-therapy virologic remission did not significantly affect the incidence of HCC (though there was a trend for lower risk of HCC development in patients with no evidence of cirrhosis; P = 0.076). It should be noted that in the patients with on-therapy virologic remission who developed HCC (8 of 228; 3.6%) all developed HCC within 30 months of the start of treatment. It is possible that the HCC in these patients had already developed before they were started on treatment. However, with the early integration of HBV into the host genome and with cirrhosis as an independent factor for the development of HCC, it is to be expected that even with very effective suppression of viral replication with nucleoside/nucleotide analogs, HCC may continue to develop.

The beneficial effect of lamivudine (and adefovir) is observed in CHB patients at different stages of the disease, including asymptomatic patients, those without cirrhosis, and those with cirrhosis. However, treatment with nucleoside/nucleotide analogs should be given on a long-term basis. Although emergence of drug-resistant HBV blunts the outcome of long-term lamivudine (and adefovir) treatment, drug resistance can be markedly reduced with the use of the newer, highly potent nucleoside/nucleotide analogs.

Other Risk Factors for Development of HCC

Other than viral replication, there are other important viral and host factors associated with development of HCC. These are not readily amendable to therapeutic interventions. Several viral genomic changes/mutations/polymorphisms have been found to be associated with a higher risk of development of HCC. These include HBV genotypes (especially genotype C), pre-S deletions, enhancer II mutations (T1653), and core promoter mutations (V1753, T1762, and A1764).33, 34

Host factors associated with higher risk of development of HCC include increasing age and male gender. Host response to HBV infection that results in progression to cirrhosis is also a major contributory risk factor. However, cirrhosis can currently be prevented and even reversed, to a certain extent, by prolonged antiviral therapy.35 Host genomic constitution may also influence the risk of development of HCC. Several recent genome-wide association studies (GWAS) have shown that single-nucleotide polymorphisms at different human genomic loci (e.g., chromosome 1p36.22, chromosome 6 of human leukocyte antigen [HLA]-DP and HLA-DQ loci, and chromosome 8p12) are associated with a higher chance of development of HCC in CHB patients.36, 37

The combination of viral and host factors exert synergistic effects on the development of HCC.34 Though total amelioration of HCC occurrence in CHB disease may not be achievable, maximal viral suppression by potent antiviral treatment is still the most effective way to reduce the occurrence of HCC.


Successful treatment of CHB can decrease the risk of HCC. The protective effect of IFN-α is likely to be limited to patients with cirrhosis who are sustained responders, a relatively small proportion of patients. The effect of IFN-α in patients without cirrhosis is unclear. Treatment with nucleoside analogs appears more effective in lowering the risk of HCC development, probably through more potent and persistent suppression of viral replication, though the effect may be blunted with the occurrence of resistance. Whether the current two first-line agents (entacavir and tenofovir), which have greater potency and lower resistance rates than lamivudine and adefovir, can further reduce the risk of development of HCC have yet to be proven.