Potential conflict of interest: Dr. Manolakopoulos is a consultant for Gilead. Dr. Kitis is a consultant for and has received a research grant from Gilead. Dr. Papatheodoridis is on the advisory board for Roche and is the principal investigator in clinical trials for Idenix. Dr. Manesis has received a research grant from Schering-Plough. Dr. Hadziyannis is a consultant, on the speakers' bureau, and on the advisory board for Gilead, Roche, and Bristol-Myers Squibb and has received research grants from Gilead and Roche.
We determined the clinical outcome of hepatitis e antigen (HBeAg)-negative chronic hepatitis B patients treated with long-term nucleos(t)ide analog therapy starting with lamivudine. We evaluated 201 such patients treated for 3.8 ± 1.4 years and 2 historical similar cohorts: 1 treated with interferon-alfa (n = 209) and 1 untreated (n = 195). Virological or biochemical remission rate at 48 months under lamivudine was 34% or 36%, respectively, whereas adefovir was administered in 79 patients with virological–biochemical breakthroughs or no response. Of the lamivudine-treated patients, 4 died, 1 underwent a transplantation, and another 8 developed major events, all having advanced fibrosis at baseline and all but 1 having experienced breakthroughs or no response. At 5 years, survival was 96%, and major event–free survival was 93%. The major event–free survival was significantly better in patients with than in those without virological remission under lamivudine. At the end of follow-up, both survival and major event–free survival were independently associated with type of and response to treatment, being significantly better in patients under long-term antiviral therapy or interferon sustained responders than in interferon non-sustained responders or untreated cases (5-year survival: 96% or 98% vs. 88% or 90%, respectively). In conclusion, in HBeAg-negative chronic hepatitis B, long-term nucleos(t)ide analog therapy starting with lamivudine significantly improves survival and reduces the risk of major complications, compared with interferon non-sustained responders or untreated patients. In such patients with advanced fibrosis, close follow-up for lamivudine resistance and prompt onset of additional antiviral therapy is required or the ab initio use of agent(s) with low resistance rates should be considered. (HEPATOLOGY 2005;42:121–129.)
Hepatitis e antigen (HBeAg)-negative chronic hepatitis B (CHB) represents a rather late phase in the course of chronic hepatitis B virus (HBV) infection1, 2 usually associated with replication of HBV variants that are unable to produce HBeAg because of mutations either at the precore or basic core promoter region of the viral genome.2, 3 HBeAg-negative CHB is a potentially severe and progressive form of chronic liver disease with rare spontaneous remissions, frequent progression to cirrhosis, and increased risk of hepatocellular carcinoma (HCC).4–7
The management of CHB has improved over the last years with the addition of lamivudine and more recently adefovir dipivoxil (ADV), but it is still suboptimal in achieving sustained off-therapy responses.8–11 In particular in HBeAg-negative CHB, interferon-alfa (IFNα), the first available effective therapeutic option in CHB, is the agent offering the higher chances for sustained-off therapy response ranging between 18% and 30%.10–13 Conversely, both lamivudine and ADV, 2 oral, safe, and well-tolerated drugs, are mainly considered and used as agents for long-term therapy in an effort to maintain on-therapy remission and prevent progression of liver disease, which currently represents the most realistic therapeutic option for most HBeAg-negative CHB patients.11, 14, 15
Lamivudine, which was the first agent used as long-term maintenance therapy in HBeAg-negative CHB since the late 1990s, is associated with high initial on-therapy remission rates, but with progressively increasing rates of viral resistance usually followed by biochemical breakthrough phenomena.16–18 The recent introduction of ADV, an agent active against both wild and mutant lamivudine-resistant HBV strains, offered an effective therapeutic option for the increasing numbers of patients with resistance to lamivudine. However, data on the clinical efficacy and particularly on the incidence of major clinical events in patients under long-term antiviral therapy are rather limited. Besides the initial partially conflicting results on the impact of emergence of YMDD HBV mutants under lamivudine,17–23 cases of liver failure have only occasionally been described, mostly in patients with preexisting cirrhosis,21 and hepatic decompensation was recently reported to develop in fewer than 1% of 998 patients with HBeAg-positive CHB treated with lamivudine for a median of 4 years.24
The aim of this study was to determine the incidence of major events, such as death, HCC, and liver decompensation, among patients with HBeAg-negative CHB treated with long-term nucleos(t)ide analog therapy starting with lamivudine and to compare it with the same incidence in 2 historical cohorts of HBeAg-negative CHB patients, one treated with IFNα and the other remaining untreated.
This retrospective study included 201 adult patients with HBeAg-negative CHB who started long-term lamivudine therapy between January 1997 and December 2001 at 4 liver centers in Greece: Academic Department of Medicine at Hippokration Hospital of Athens, Department of Medicine and Hepatology at Henry Dunant Hospital of Athens, Department of Gastroenterology at George Papanikolaou Hospital of Thessaloniki, and Department of Gastroenterology at Polyclinic Hospital of Athens. Initial results for 58 of these patients have been previously published.16, 17 Lamivudine was given at a dose of 100 to 150 mg daily.
Inclusion and exclusion criteria for starting lamivudine were similar at the 4 centers. All patients had: (1) hepatitis B surface antigen positive and HBeAg negative for 6 months or longer; (2) elevated alanine aminotransferase (ALT) activity on 2 separate monthly occasions; (3) detectable serum HBV-DNA within the last month before therapy onset; and (4) histological findings of chronic hepatitis or cirrhosis (Child-Pugh class A) within the last 12 months preceding therapy. No patient had decompensated liver disease, HCC [focal lesions at ultrasonography or alpha-fetoprotein (AFP) >50 ng/mL] or detectable antibody to hepatitis C or D virus or to human immunodeficiency virus, and none had received a liver transplant or any antiviral drug other than IFNα in the past or any antiviral or immunosuppressive therapy within the last 6 months. Patients who were treated with IFNα or any anti-HBV agent other than nucleos(t)ide analog after the onset of lamivudine were also excluded. Patients may have been treated with any nucleos(t)ide analog after development of resistance to lamivudine. All patients gave written informed consent to participate in these studies. The study protocols conformed to the ethical guidelines of the 1975 Declaration of Helsinki.
Follow-up at all 4 centers included clinical examination and routine laboratory tests at least every 3 months and determination of serum HBV-DNA levels at least every 6 months and on any biochemical breakthrough. Ultrasonography and measurement of AFP levels were performed at baseline in all patients and at least every 6 months during follow-up in patients with cirrhosis. Serum HBV-DNA levels were evaluated by using a commercially available quantitative polymerase chain reaction (PCR) assay (Amplicor™ HBV Monitor Test, Roche Diagnostics, GmbH Mannheim, Germany); the sensitivity of this assay according to the manufacturer was approximately 1,000 copies/mL.25 Initial virological response was considered as the undetectability of serum HBV-DNA by PCR and initial biochemical response as the decline of ALT below the upper limit of normal (40 IU/L) in ≥2 consecutive determinations during therapy. Virological breakthrough was considered as the reappearance of serum HBV-DNA by PCR after initially undetectable HBV-DNA and biochemical breakthrough as the increase of ALT above 1.5 × upper limit of normal after initial biochemical response. Virological or biochemical remission was considered as the maintenance of initial virological or biochemical response without development of virological or biochemical breakthrough throughout follow-up.
Historical Cohorts of IFNα-Treated and Untreated Patients.
Two other cohorts of HBeAg-negative CHB patients, who constituted the population in another published study,26 were also evaluated. Two hundred nine IFNα-treated and 195 untreated patients fulfilled inclusion and exclusion criteria described previously in detail26 and being similar to the criteria used for the main population. No patient of these 2 historical cohorts was treated with any nucleos(t)ide agent during the follow-up period of our previous study,26 which ended in December 1998, when such agents were not approved for the treatment of CHB. In IFNα-treated patients, sustained off-therapy biochemical response was observed in 57 (sustained responders) and biochemical relapse or no response in the remaining 152 cases (non-sustained responders). In summary, follow-up included clinical examination and liver function tests at least every 6 months in all patients as well as ultrasonography and AFP determinations every 6 months in patients with cirrhosis.
Clinical complications considered as major events were: death, orthotopic liver transplantation (OLT), development of HCC, and development of liver decompensation. HCC diagnosis was based on histological findings or high serum AFP levels (>400 ng/mL) with compatible radiological findings. Liver decompensation was diagnosed in case of development of ascites, variceal bleeding, hepatic encephalopathy, or jaundice (serum bilirubin > 3 mg/dL) of non-extrahepatic origin.
Start of follow-up was considered to be the lamivudine onset for the main population, the IFNα onset for IFNα-treated, and the first outpatient visit for untreated patients. The follow-up ended at the time of death, OLT, or at the last patient visit. Fourteen (7.0%) patients of the main population, 9 (4.3%) IFNα-treated, and 12 (6.2%) untreated patients did not attend their liver clinics for the last 1-2 years of their follow-up period and were considered as lost to follow-up. “Lost” patients of all groups were considered to have the clinical status of their last visit and were censored at that time for survival analysis. However, all “lost” patients of the main population were contacted by phone in March 2005 and reported to be alive and without any major event, except for 1 patient in whom HCC was diagnosed.
All patients of the main population and the IFNα-treated cohort had a baseline liver biopsy within the last 12 months preceding therapy, and all untreated cases had a biopsy within the first year of follow-up. The histological lesions were classified into grading and staging according to the system proposed by Ishak et al.27
All data were analyzed using the statistical package SPSS (version 11.0, SPSS Inc., Chicago, IL). Corrected chi-square or two-tailed Fisher's exact test and t test or Mann-Whitney test were used for qualitative and quantitative data, when appropriate. The Kaplan-Meier method was used to estimate on lamivudine remission and survival rates and the log-rank as a test for significance. Any death or OLT were combined as end-points of survival and liver-related death or OLT as end-points of liver related survival. For the calculation of major event-free survival, any major event was taken into account and follow-up ended at the time of development of such a complication, if any. Cox regression analysis was used for evaluation of the independent effects of several factors on survival and major event-free survival. A two-tailed P value < .05 was considered statistically significant.
The baseline characteristics of all patients appear in Table 1. Patients of the main population were older than both IFNα-treated and untreated patients, whereas they had higher staging scores than IFNα-treated and higher ALT levels than untreated patients. There was no other significant difference in the baseline characteristics of the main population and the 2 historical cohorts, which had longer follow-up. Sixty-four (31.8%) patients of the main population had histological cirrhosis (fibrosis score = 5-6; score 5: 34, score 6: 30).
Table 1. Baseline Characteristics of Three Cohorts of Patients With HBeAg-Negative Chronic Hepatitis B: One Treated With Maintenance Antiviral Therapy Starting With Lamivudine (Main Study Population), One Treated With Interferon Alfa (IFNα), and One Remaining Untreated
Main Study Population (n = 201)
IFNα-Treated Patients (n = 209)
Untreated Patients (n = 195)
Versus ‡, P < .001; †versus §, P = .007. There was no other significant difference in the characteristics of the main study population and the IFNα-treated or untreated patients. ‖IgM anti-HBc was available in 126 of the 201 patients of the main study population.
Up to April 2004, the mean follow-up was 3.8 ± 1.4 (1.0-7.9) years. In total, virological no response was observed in 17 (8.5%) and breakthroughs in another 92 (45.8%) patients, whereas biochemical no response was observed in 11 (5.5%) and breakthroughs in another 86 (42.8%) patients. The probability of virological remission under lamivudine monotherapy was 73%, 52%, 40%, and 34% and of biochemical remission 84%, 64%, 50%, and 36% at 12, 24, 36, and 48 months after lamivudine onset, respectively.
ADV (10 mg/day) was administered in 79 (72.5%) of the 109 patients without virological remission. It was added to lamivudine in 53 and replaced lamivudine in another 26 patients. The median duration of ADV therapy was 14 (1-45) months. ADV started at a median of 13 (1-45) months after the onset of virological breakthroughs in 68 patients with such breakthroughs and at a median of 36 (15-91) months after the lamivudine onset in 11 patients with virological no response. In relation to biochemical responses, ADV was started at a median of 7 (0-65) months after the onset of biochemical breakthroughs in 69 patients with such breakthroughs and at a median of 30 (15-91) months after the lamivudine onset in 7 patients with biochemical no response. ADV was also given in 3 patients with virological breakthroughs but in biochemical remission.
Major Events in the Main Population.
During follow-up, 4 patients died, and 1 underwent OLT at a median of 39 (12-42) months after the onset of lamivudine. The cause of death was liver decompensation in 2, HCC in 1, and unrelated to the liver disease (myocardial infarction) in 1 patient, whereas OLT was performed because of HCC in 1 patient (Fig. 1). Overall or liver-related survival was excellent, being 99% for the first 3 years and 96% or 97% at 4 and 5 years (Fig. 2). All patients who died of liver-related complications or underwent transplantation had histological cirrhosis (fibrosis score = 6) at baseline and had developed both virological and biochemical breakthroughs during lamivudine. The first of the 2 patients who died because of decompensation received no additional therapy, because no antiviral agent was available at that time. The second patient was treated with ADV (plus lamivudine), which started at 5.5 months after the development of decompensation, but he died 2 months after the addition of ADV.
Liver decompensation developed in 9 patients, including those 2 who eventually died at a median of 18 (12-69) months after lamivudine onset (Fig. 1). All 9 patients had histological cirrhosis at baseline (fibrosis score 5 and 6: 1 and 8 cases, respectively) and had developed both virological and biochemical breakthroughs (n = 8) or no response (n = 1) during lamivudine. In particular, decompensation developed at a median of 6 (3-24) months after the detection of virological or of 5 (0-19) months after the onset of biochemical breakthroughs in the 8 patients with such breakthroughs and at 13 months after the onset of lamivudine in the one patient with no response. ADV was added to lamivudine in 5 patients at a median of 4 (0-13) months after (n = 4) or at 2 months before (n = 1) development of decompensation. The duration of ADV plus lamivudine therapy in these patients was 15 (7-36) months. The addition of ADV resulted in reversion of decompensation in 4 patients, whereas 1 patient eventually died. A new agent was not added in 4 patients who developed decompensation in the pre-ADV era; one died at 3 months after decompensation, whereas the remaining 3 patients were alive but lost to follow-up at 1 to 5 months after decompensation (the latter 3 patients were treated with ADV at another center and were alive without signs of decompensation 1 to 2 years later).
HCC developed in 3 patients with baseline histological cirrhosis (fibrosis score = 6) (Fig. 1). It was diagnosed in 2 patients with breakthroughs at 33 and 54 months after lamivudine onset (1 died at 10 months and the other is still alive at 11 months after HCC diagnosis). In relation to the timing of breakthroughs, HCC was diagnosed at 9 and 15 months after the onset of virological or at 6 and 10 months after the onset of biochemical breakthroughs. HCC was also diagnosed in 1 patient remaining in both virological and biochemical remission at 10 months after lamivudine onset (underwent OLT at 5 months after HCC diagnosis). HCC was diagnosed at another center in a fourth patient 21 months after his last follow-up visit in this study. The classification of the latter patient into those with major events did not significantly change any of the results (data not shown).
When all cases of liver decompensation were taken into account, the major event-free survival was 99%, 97%, 94%, 94% and 92% at 12, 24, 36, 48, and 60 months, respectively. The major event-free survival was significantly better in patients remaining in virological remission (99% at 12 to 60 months) than in those with virological breakthroughs or no response under lamivudine (99%, 95%, 91%, 90%, and 88% at 12, 24, 36, 48, and 60 months, respectively) (P = .016) (Fig. 3A). At the end of follow-up, when the 4 patients who had achieved reversion of decompensation after ADV addition were considered as cases without major events, the major event-free survival slightly improved, being 99.5%, 98%, 96%, 95% and 93% at 12, 24, 36, 48, and 60 months, respectively (Fig. 4). There was a trend for better major event-free survival at the end of follow-up in patients remaining in virological remission than in those with virological breakthroughs or no response under lamivudine (P = .07) (Fig. 3B).
Comparison With IFNα-Treated and Untreated Patients.
The overall survival was better in the main population than in untreated (P = .05), but not than in IFNα-treated patients (P = .20) irrespective of their response to IFNα. The liver-related survival in the main population was significantly better compared with untreated patients (P = .04) or non-sustained responders to IFNα (P = .036), relatively better compared with total IFNα-treated patients (P = .11) and similar compared with IFNα sustained responders (Fig. 2).
The major event–free survival in the main population at any time point was also significantly better than in untreated (P = .03), but not than in IFNα-treated patients (P = .47) irrespective of their response to IFNα. When the reversion of decompensation after ADV addition was taken into account, the difference in the major event–free survival between the main population and untreated patients increased (P = .003), and there was a trend for difference between the main population and IFNα-treated patients (P = .10) (Fig. 4A). Moreover, the major event–free survival in the main population at the end of follow-up was better than in IFNα non-sustained responders (P = .02) and similar to that of IFNα sustained responders (P = .65) (Fig. 4B).
Cox regression analysis showed that younger age (P < .001), lower fibrosis score or absence of cirrhosis at baseline (P < .0001) and type of therapy were independently associated with survival and major event–free survival. In particular, the adjusted survival was significantly better in the main population than in IFN-treated [hazard ratio (HR): 1.80, 95%; confidence interval (CI): 1.00-3.25; P = .047) or untreated patients (HR: 1.65, 95%; CI: 1.00-2.72;; P = .05). Moreover, the adjusted survival in the main population was significantly better than in IFNα non-sustained responders (HR: 1.80, 95%; CI: 1.06-3.04;; P = .03) and similar with IFNα sustained responders. Similar associations were observed for the major event–free survival at the end of follow-up. The adjusted major event–free survival in the main population at any time, however, was significantly better than in untreated patients or IFNα non-sustained responders, but not than in total IFNα-treated patients (Table 2).
Table 2. The Independent Effect of Maintenance Antiviral Therapy Starting With Lamivudine (Main Study Population) on the Outcomes of Patients With HBeAg-Negative Chronic Hepatitis B
Main Study Population Versus
HR (95% CI)
HR (95% CI)
HR (95% CI)
HR (95% CI)
NOTE. Comparisons with a cohort of interferon-alfa (IFNα) treated and a cohort of untreated patients using Cox regression analysis. Hazard ratios (HR) and their confidence intervals (CI) are adjusted for age and presence of cirrhosis.
Abbreviations: IFNα-SR, IFNα-no-SR: sustained responders, non-sustained responders after IFNα therapy; OLT: orthotopic liver transplantation. Major events included death, OLT, hepatocellular carcinoma, or liver decompensation.
Survival (any death/OLT)
Liver related survival (liver-related death/OLT)
Major event-free survival (at any time of follow-up)
Major event-free survival (at the end of follow-up)
Survival and major event–free survival rates were not significantly different between patients who remained in virological remission under lamivudine and sustained responders to IFNα, whereas they were almost identical in patients of the main population with (n = 131) or without (n = 70) previous unsuccessful IFNα therapy (data not shown).
Our study, which evaluated approximately 200 patients with HBeAg-negative CHB followed for up to 8 years, shows that long-term nucleos(t)ide analogs maintenance therapy starting with lamivudine is associated with an excellent survival of over 95% at 5-years. Such a survival rate was observed in a patient population with 2 baseline factors, old age (mean of 52 years), and frequent (>30%) presence of cirrhosis, which have been repeatedly shown to be strongly associated with development of liver-related complications and poor survival.26, 28 These data may reflect the results of long-term antiviral therapy in attentive clinical practice, because we included patients from 4 liver centers treated with similar therapeutic strategies, but not according to a strict therapeutic protocol. The efficacy of lamivudine was similar to what has been previously reported in other studies from Greece16, 17 and Italy29 with approximately one third of patients remaining in remission at 4 years of therapy. There was no specific guideline for the management of patients with lamivudine breakthroughs, which was mostly influenced by time period and drug availability.
The 5-year probability of developing liver decompensation or HCC was found to be 7% to 8%. In a recent study, decompensation or HCC were observed in 6.3% (35/551) of patients with HBeAg-negative CHB or compensated cirrhosis treated with lamivudine monotherapy for a mean of 2 years.29 In contrast in a large study of long-term lamivudine monotherapy in HBeAg-positive CHB patients, lower death and decompensation rates (<1%) during a median follow-up of 4 years were reported.24 However, patients of the latter study differed from our main population in 2 crucial baseline characteristics, age and presence of cirrhosis, because they were approximately 2 decades younger (mean age of 32 years) and had cirrhosis in a minority of cases (10%).24 Liver decompensation, HCC, and liver-related death or OLT developed only in patients with baseline cirrhosis (all but 1 had fibrosis score 6 and one score 5) and almost exclusively in patients with no response or breakthroughs under lamivudine (except for 1 patient with HCC). The probability of developing a major event was significantly lower in our patients with than in those without virological remission under lamivudine (Fig. 3A), which is compatible with data of other recent studies.29, 30
The introduction of ADV resulted in biochemical and virological remission in most cases with breakthroughs or no response under lamivudine, despite the fact that ADV was administered with delay mainly due to non-availability. In particular, ADV was given at a median of 13 or 7 months after detection of virological or biochemical breakthroughs in the cases with such breakthroughs and at a median of 30 to 36 months after the lamivudine onset in the cases with no response. Conversely, the addition of ADV at 5.5 months after decompensation in a patient with lamivudine breakthrough did not prevent the deterioration of liver disease and eventually the patient's death. The administration of ADV improved the major event–free survival in our main population, which, however, remained relatively better in patients with than without virological remission under lamivudine (Fig. 3B). Because earlier administration of ADV might have prevented such an outcome in the latter patient, close monitoring for breakthroughs should be advised in all CHB patients under lamivudine monotherapy, and particularly in patients with cirrhosis who are at high risk of deterioration when breakthroughs occur.
The survival and complication rates in our main study population were compared with those of two historical cohorts, 1 including IFNα-treated and the other untreated patients. The 3 cohorts had similar inclusion and exclusion criteria, whereas older age and presence of cirrhosis were strong factors associated with survival and probability of complications in all of them. Finding similar control groups would be nearly impossible, because it is unethical not to offer treatment to naïve CHB patients or those who fail to respond to IFNα. In fact, a very recent randomized placebo-controlled trial of lamivudine monotherapy in CHB was discontinued prematurely because of the strong beneficial effect of lamivudine on patients' outcome.30
According to our findings, long-term antiviral therapy was independently associated with better outcomes (in both survival and liver-related complications) compared with IFNα or no therapy. In the large randomized trial of Liaw et al.,30 in which patients with advanced fibrosis or cirrhosis due to either HBeAg-positive or HBeAg-negative CHB were included, long-term lamivudine monotherapy also was shown to significantly improve the outcome compared with placebo. In addition, our patients under long-term antiviral therapy had similar outcomes with IFNα sustained responders and significantly better outcomes than IFNα non-sustained responders. One limitation of our study is that the proportion of patients at risk after the first 3 to 4 years of follow-up, when most of the major events occurred, was substantially lower in the main population than in the IFNα-treated or untreated patients. Thus, although the follow-up length was taken into account in our analysis, it would be possible that as more patients in the main population are followed for longer, more events may develop and affect the survival rates. Conversely, it should be stressed that the outcomes of patients under long-term nucleos(t)ide analogs therapy are expected to further improve in the current era of ADV availability with the prompter administration of effective treatment after development of lamivudine breakthroughs. Similarly, the outcomes of IFNα-treated patients must have also improved, because non-sustained responders to IFNα are now treated with nucleos(t)ide analogs and they usually do not remain untreated, as was the usual case during the follow-up period of our historical cohorts.
Whether maintenance of on-therapy remission can prevent or reduce the risk of HCC development is unclear. One of our 3 patients with HCC was in short-lived remission at 10 months after the lamivudine onset, when HCC was diagnosed. Despite absence of suspicious finding at ultrasonography and normal baseline AFP levels, however, the interval between lamivudine onset and HCC detection was so short that the pre-existence of HCC foci is most likely. Conversely, the probability of HCC in our main study population was relatively lower than in untreated or total IFNα-treated patients, significantly lower than in IFNα–non-sustained responders and similar with that in IFNα sustained responders. The results did not significantly change after adjustment for age and presence of cirrhosis (data not shown).
In conclusion, in patients with HBeAg-negative CHB, long-term nucleos(t)ide analog therapy starting with lamivudine significantly improves survival and reduces the risk of liver-related major complications. Such a management is associated with outcomes similar to those observed after sustained response to IFNα, both being significantly better than the outcomes in patients without sustained response after IFNα or in untreated cases.26 In HBeAg-negative CHB patients under long-term lamivudine monotherapy, major liver complications usually develop in the presence of baseline cirrhosis or advanced fibrosis following the emergence of breakthroughs. Thus, in such patients with advanced fibrosis, very close follow-up for lamivudine resistance and prompt onset of effective additional antiviral therapy is required in an effort to reduce the rate of complications, or alternatively the ab initio use of agent(s) with low resistance rates should be considered.