Mortality, Liver Transplantation, And Hepatocellular Carcinoma

Gastroenterology 2014;147:152–161

Mortality, Liver Transplantation, and Hepatocellular Carcinoma
Among Patients With Chronic Hepatitis B Treated With
Entecavir vs Lamivudine
Young–Suk Lim,1 Seungbong Han,2 Nae–Yun Heo,3 Ju Hyun Shim,1 Han Chu Lee,1
and Dong Jin Suh1
1

CLINICAL LIVER

Department of Gastroenterology, Liver Center, 2Department of Clinical Epidemiology and Biostatistics, Asan Medical Center,
Seoul, Republic of Korea; 3Department of Internal Medicine, Haeundae Paik Hospital, Inje University College of Medicine,
Busan, Republic of Korea

See Covering the Cover synopsis on page 2; see
editorial on page 24.
BACKGROUND & AIMS: Little is known about whether the
antiviral agent entecavir is more effective than a less potent
drug, lamivudine, in reducing the risk of death and hepatocellular carcinoma (HCC) in patients with chronic hepatitis B.
METHODS: We performed a retrospective analysis of data from
5374 consecutive adult patients with chronic hepatitis B, treated
with entecavir (n ¼ 2000) or lamivudine (n ¼ 3374), at a tertiary referral hospital in Seoul, Korea, from November 1, 1999,
through December 31, 2011. Data were collected from patients
for up to 6 years and analyzed by a multivariable Cox proportional hazards model for the entire cohort and for propensity
score-matched cohorts. RESULTS: During the study period, 302
patients (5.6%) died, 169 (3.1%) received a liver transplant, and
525 (9.8%) developed HCC. Multivariable analyses showed that
compared with lamivudine, entecavir therapy was associated
with a significantly lower risk of death or transplantation (hazard ratio [HR], 0.49; 95% confidence interval [CI], 0.38–0.64),
but a similar risk of HCC (HR, 1.08; 95% CI, 0.87–1.34). In the
1792 overall propensity-matched pairs, entecavir again was
associated with a significantly lower risk of death or transplantation (HR, 0.49; 95% CI, 0.37–0.64) and a similar risk of
HCC (HR, 1.01; 95% CI, 0.80–1.27). Entecavir also reduced the
risk of death or transplantation, compared with lamivudine, in
860 pairs of patients with cirrhosis (HR, 0.42; 95% CI,
0.31–0.57) but there were no differences in risk for HCC (HR,
1.00; 95% CI, 0.78–1.28). However, entecavir and lamivudine did
not have significantly different effects on clinical outcome in 878
pairs of patients without cirrhosis. CONCLUSIONS: In a retrospective study of 5374 patients with chronic hepatitis B virus
infection, entecavir therapy was associated with a significantly
lower risk of death or transplantation than lamivudine. However,
the drugs did not have different effects on HCC risk.

Keywords: Comparative Effectiveness; HBV; Liver Cancer;
Survival.

C

hronic hepatitis B virus (HBV) infection is a major
global health burden. Approximately 400 million

people worldwide are chronically infected with HBV. These
patients have a substantially increased risk of cirrhosis
and hepatocellular carcinoma (HCC), which together are
responsible for approximately 1 million deaths worldwide
every year.1,2
Since lamivudine first was introduced for the treatment
of chronic hepatitis B (CHB), 4 additional oral antiviral
agents, namely, adefovir, telbivudine, entecavir, and tenofovir, have been approved. A randomized placebo-controlled
trial and several cohort studies have shown that when
compared with no treatment, lamivudine significantly reduces mortality and prevents HCC in patients with CHB.3–6
However, entecavir and tenofovir are more potent than
lamivudine in terms of suppressing HBV replication with a
minimal risk of resistance.7–10 Thus, entecavir and tenofovir
currently are being recommended as first-line agents for
CHB.11–13 However, whether entecavir or tenofovir reduce
the risk of death or HCC to a greater extent than lamivudine
is unknown.
Although a randomized controlled trial provides the
highest level of evidence, such trials often have limited power to evaluate efficacy in reducing long-term clinical
events. They also may be unethical because of the proven
differences in intermediate end points, such as virologic
responses. Furthermore, their results may not be generalizable because the patients often are selected carefully.
Observational studies from clinical databases may better
reflect real-world practice and, consequently, may complement the data from clinical trials if they are based on a large
population and if biases and confounders are controlled
adequately.
The aim of this cohort study was to compare the efficacy
of entecavir and lamivudine in terms of the risk of death or
transplantation and HCC in CHB patients.

Abbreviations used in this paper: CHB, chronic hepatitis B; CI, confidence
interval; HBeAg, hepatitis B e antigen; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HR, hazard ratio.
© 2014 by the AGA Institute
0016-5085/$36.00
http://dx.doi.org/10.1053/j.gastro.2014.02.033

Materials and Methods
Study Subjects
The study population was obtained from a historical cohort
of 9615 consecutive adult CHB patients who were treated with
entecavir (0.5 mg/day) or lamivudine (100 mg/day) at Asan
Medical Center, a 2700-bed academic tertiary referral hospital
in Seoul, Korea, between November 1, 1999, and December 31,
2011 (Figure 1). Patients were excluded if they met any of the
following criteria: younger than age 20 or older than age 80;
the patient died or underwent liver transplantation within
6 months of treatment; the patient developed HCC within 1
year of treatment; the patient achieved hepatitis B surface antigen seroclearance within 6 months of treatment; the patient
was co-infected with other hepatotrophic viruses; the duration
of therapy with the study drugs was shorter than 6 months; the
patient was treated with any other antiviral agents for more
than 2 weeks previously; or the patient had serum HBV-DNA
levels less than 2000 IU/mL (or undetectable) at baseline.
Before January 2007, when entecavir first was approved, lamivudine was used exclusively for the treatment of CHB. Thereafter, entecavir was used in the majority (82.5%) of patients in
the study cohort.
This study was approved by the Institutional Review Board
of Asan Medical Center, and the requirement for informed
consent from the patients was waived.

Outcomes and Follow-up Evaluation
The outcomes of interest in this study were all-cause
mortality or liver transplantation and the development of
HCC. The index date was defined as the first date that the patient took entecavir or lamivudine. The patients were followed
up from the index date to the time of death, transplantation, or
the last follow-up evaluation (March 31, 2013). Information
regarding baseline patient characteristics and the major clinical
outcomes was obtained from complete inpatient and outpatient

Figure 1. Patient flow diagram. HBsAg, hepatitis B
surface antigen; HCV, hepatitis C virus; HDV, hepatitis
delta virus; HIV, human immunodeficiency virus.

Entecavir vs Lamivudine and Clinical Outcome

153

medical records. To validate the complete set of follow-up data,
information about vital status through March 31, 2013, was
obtained from the National Population Registry of the Korea
National Statistical Office using unique personal identification
numbers. Information regarding the primary diagnosis of HCC
was validated by accessing the Korean National Health Insurance Service database, which covers more than 99% of the
entire Korean population, has a high HCC registration rate
(96.5%) and highly accurate diagnoses, and has been validated
as a valid resource for research.14
The patients regularly underwent clinical examinations,
liver function tests, and measurement of serum HBV-DNA levels
every 3–6 months. Because the early discovery that hepatitis
B e antigen (HBeAg) seroconversion is poorly durable,15 patients were advised to continue their treatment even after
HBeAg seroconversion.
Two HCC risk assessment scores that were developed for
CHB patients, the CU-HCC score and the GAG-HCC score,16–18
were used for this study to assess the risk of developing HCC
at baseline in patients treated with lamivudine and entecavir.
Patients were screened for cirrhosis and HCC by ultrasonography and serum a-fetoprotein level at baseline and at
every 6 months during the follow-up period. Cirrhosis was
diagnosed when the liver echotexture was coarse and nodular
and/or when features of portal hypertension (eg, ascites,
splenomegaly, and varices) were noted. The diagnosis of HCC
was based on radiologic findings (dynamic computed tomography and/or magnetic resonance imaging), as recommended.19 When the hepatic nodule did not show imaging
features typical of HCC, a biopsy specimen was taken and
subjected to pathologic examination.

Serum Assays
Before 2007, serum HBV-DNA levels were measured using a
hybrid capture assay (lower limit of detection, 20,000 IU/mL;
Digene Diagnostics, Gaithersburg, MD); thereafter, levels were

CLINICAL LIVER

July 2014

154

Lim et al

measured in a real-time polymerase chain reaction assay
(linear dynamic detection range, 15 IU/mL 1  109 IU/mL;
Abbott Laboratories; Chicago, IL). The results were expressed
in international units per milliliter after appropriate transformation.20 HBV genotype was not determined because more
than 98% of Korean patients with CHB have HBV genotype
C2.21 Restriction fragment mass polymorphism assays were
performed routinely22 to identify HBV resistance mutations if
the patient showed a primary nonresponse (a reduction in
serum HBV-DNA level of less than 2 log10 IU/mL after 24 weeks
of therapy), an inadequate virologic response (HBV-DNA level,
>4 log10 IU/mL after 24 weeks of therapy), or virologic
breakthrough (1 log10 IU/mL increase in HBV-DNA level from
the nadir on 2 consecutive tests) during treatment.11,13
CLINICAL LIVER

Statistical Analysis
All patients who met the eligibility criteria at baseline were
included in the analyses.
To reduce the effect of treatment selection bias and potential confounding, propensity score-matching was performed.
The use of propensity scores and matching produced 2 groups
of similar patients except for the treatment of interest and deal
with extraneous differences that were confounding the results.23 Propensity score-matching analyses allow simultaneous
adjustment for many covariates in situations in which conventional multivariable models might not be appropriate owing
to the multiple confounders and rare outcomes.24 The propensity scores for each cohort (the overall cohort and subcohorts comprising patients without cirrhosis, those with
cirrhosis, those with compensated cirrhosis, those with
decompensated cirrhosis, and those showing a complete virologic response) were derived nonparametrically using the
following variables: age; sex; HBeAg positivity; serum levels of
HBV DNA, alanine aminotransferase, albumin, total bilirubin,
and creatinine; international normalized ratio; platelet count;
diabetes mellitus; hypertension; cirrhosis; ascites; and Child–
Turcotte–Pugh scores. The same set of covariates was used for
propensity score modeling in the entire cohort and in the
subcohorts (cirrhosis was not included as a covariate in the
noncirrhosis and cirrhosis subcohorts). Propensity scorematching was performed by nearest-neighbor matching with
a caliper width of 0.1 multiplied by the standard deviation for
the linearly transformed propensity scores (logit-transformation). The propensity score-matching method is described
in detail in the Supplementary Materials and Methods section.
Basal laboratory data were missing in 0.02%–6.22% of patients: HBeAg status was missing in 1 patient (0.02%), creatinine levels were missing in 8 patients (0.15%), platelet counts
were missing in 15 patients (0.28%), and INR was missing in
334 patients (6.22%). Missing values were imputed using
linear interpolation.25 Imputed values were constrained within
plausible ranges.
The baseline characteristics of the 2 treatment groups were
compared using the chi-square test and the t test for categoric
and continuous variables, respectively. For the propensity
score-matched cohorts, the baseline characteristics of the 2
groups were compared using the McNemar test and the paired
t test for categoric and continuous variables, respectively.
The differences between the 2 groups in terms of clinical
outcome were assessed using a multivariable Cox proportional
hazards model. Multivariable models were fitted using the

Gastroenterology Vol. 147, No. 1
backward variable selection approach, and the least significant
variables were discarded one by one after fitting the full model.
Variables included in the multivariable analyses were the
same as those used in the propensity score matching as listed
earlier. The virologic response (serum HBV-DNA level, <20,000
IU/mL) at 1 year of treatment, the need for rescue therapy and
occurrence of drug-resistance during treatment, and overall
treatment duration also were included in the multivariable
analyses as time-dependent covariates.
For the propensity score-matched cohorts, differences in
clinical outcome between the 2 groups were assessed adjusting
for overall treatment duration using multivariable Cox regression models, with robust standard errors that accounted for
the clustering of matched pairs.26–28 Curves representing the
cumulative incidence rates were constructed based on
Kaplan–Meier estimates and then compared with the use of
methods described by Klein and Moeschberger.27 Competingrisks survival analysis was performed to adjust for the risks
associated with HCC (ie, death or transplantation) in the overall
propensity score-matched cohort. A detailed description of the
competing risks analysis is provided in the Supplementary
Materials and Methods section.
All reported P values are 2-sided, and P values less than .05
were considered significant. SAS (version 9.1; SAS, Cary, NC)
and R (version 3.0; available: http://cran.r-project.org/) software were used for statistical analyses.

Results
Characteristics of the Overall Study Population
The study population comprised 5374 CHB patients who
started their first treatment with entecavir (n ¼ 2000) or
lamivudine (n ¼ 3374) and who met the inclusion criteria
(Figure 1). The 2 groups differed significantly in terms of
baseline characteristics (Table 1). The patients in the entecavir group were significantly older (mean, 47 vs 43 y),
were less likely to be HBeAg positive (58.4% vs 71.8%), had
lower levels of serum HBV DNA (median, 7.14 vs 7.49 log10
IU/mL), and were more likely to have cirrhosis (53.6% vs
48%). The HCC risk assessment scores for the 2 groups also
were compared. The CU-HCC score was similar between the
groups (P ¼ .99), whereas the GAG-HCC score was significantly higher for the entecavir group than for the lamivudine group (mean, 95.3 vs 91.2; P < .001) (Table 1).

Follow-up Evaluation
Complete follow-up evaluation was defined as following
up a patient until the time of death, transplantation, or the
last follow-up date (March 31, 2013). These data were
available for 99.8% of the entire cohort. For the lamivudine
and entecavir groups, the median follow-up periods were
8.7 (interquartile range, 6.5–11.5) and 3.1 (interquartile
range, 2.2–4.3) years, respectively. To minimize inequalities
in follow-up duration between the 2 groups, patients who
were followed up for more than 6 years were censored at
6 years of follow-up evaluation. In the resulting follow-up
period (ie, up to 6 years), 302 patients (5.6%) died, 169
patients (3.1%) received a transplant, and 525 patients
(9.8%) developed HCC.

July 2014

Entecavir vs Lamivudine and Clinical Outcome

155

Table 1.Characteristics of the Entire Patient Cohort
Entire cohort
Entecavir
(n ¼ 2000)

Lamivudine
(n ¼ 3374)

Age, mean  standard deviation, y
Male sex
HBeAg
HBV-DNA level, mean  standard deviation, log10 IU/mL
ALT level, median (IQR), IU/mL
Albumin level, median (IQR), g/dL
Total bilirubin level, median (IQR), mg/dL
INR, median (IQR)
Creatinine, median (IQR), mg/dL
Platelets, median (IQR), 1000/mm3
Diabetes mellitus
Hypertension
Cirrhosis
Compensated
Decompensated
Ascites
CTP score, median (IQR)
CU-HCC score17
GAG-HCC score18
Duration of overall treatment, median (IQR), y
Duration of overall follow-up period, median (IQR), y

47  11
1288 (64.4%)
1168 (58.4%)
7.14  1.64
101 (53–190)
3.8 (3.4–4.1)
1.2 (0.9–1.6)
1.10 (1.00–1.20)
0.80 (0.70–1.00)
142 (96–183)
77 (3.9%)
102 (5.1%)
1071 (53.6%)
758 (70.8%)
313 (29.2%)
253 (12.7%)
5 (5–6)
19.8  13.9
95.3  22.1
2.6 (1.8–3.9)
3.1 (2.2–4.3)

43  11
2386 (70.7%)
2421 (71.8%)
7.49  1.17
128 (68–244)
3.8 (3.2–4.1)
1.1 (0.9–1.6)
1.10 (1.00–1.30)
0.90 (0.80–1.00)
147 (96–195)
122 (3.6%)
134 (4.0%)
1621 (48.0%)
958 (59.1%)
663 (40.9%)
532 (15.8%)
5 (5–6)
19.8  14.8
91.2  22.7
6.1 (2.7–9.0)
8.7 (6.5–11.5)

P value
<.001
<.001
<.001
<.001
.13
<.001
.79
<.001
.02
.03
.66
.05
<.001
<.001
.002
<.001
.99
<.001
<.001
<.001

ALT, alanine aminotransferase; CTP, Child–Turcotte–Pugh; INR, international normalized ratio; IQR, interquartile range.

The cumulative rates of documented genotypic resistance mutations to lamivudine and/or entecavir during
the follow-up period (up to 6 years) were 1.5% and 50.8%
for the entecavir and lamivudine groups, respectively
(P < .001). Overall, 1.8% of the entecavir-treated patients
and 39.3% of the lamivudine-treated patients needed rescue
therapy (P < .001). Most of these patients were switched to
a combination therapy that included adefovir. Tenofovir was
not available during the study period.

Multivariable Analyses
Multivariable analyses to identify factors predictive of
clinical outcomes were performed for the entire cohort
(Table 2). Entecavir treatment was associated independently
with a significantly lower risk of death or transplantation
(hazard ratio [HR], 0.49; 95% confidence interval [CI],
0.38–0.64; P < .001). However, the risk of HCC was similar
between the 2 treatment groups (HR, 1.08; 95% CI,
0.87–1.34; P ¼ .48).

Table 2.Multivariable Analyses Showing Significant Predictive Factors of Death or Transplantation and Hepatocellular
Carcinoma in the Entire Cohort
Death or transplantationa

Hepatocellular carcinomab

Variables

HR

95% CI

P value

HR

95% CI

P value

Treatment with entecavir
Age
Male sex
Albumin level, g/dL
Platelets, 103/mL
Diabetes mellitus
Cirrhosis
Need for rescue therapy
Overall treatment yearsc

0.49
1.02
1.58
0.73
0.989
2.18
2.51
2.14
0.72

0.38–0.64
1.00–1.03
1.27–1.96
0.57–0.93
0.986–0.991
1.59–2.99
1.71–3.68
1.69–2.70
0.69–0.75

<.001
.006
<.001
.01
<.001
<.001
<.001
<.001
<.001

1.08
1.06
1.81
0.70
0.995
1.51
2.59
1.53
0.94

0.87–1.34
1.05–1.07
1.48–2.20
0.61–0.80
0.993–0.997
1.10–2.08
1.97–3.41
1.23–1.91
0.92–0.97

.48
<.001
<.001
<.001
<.001
.01
<.001
<.001
<.001

NOTE. A time-dependent multivariable Cox proportional hazards model with a backward variable selection approach was used
for all analyses.
a
Total number of patients, 5374; number of events, 457.
b
Total number of patients, 5374; number of events, 525.
c
Overall duration of treatment with any nucleoside/nucleotide analogue.

CLINICAL LIVER

Characteristics

156

Lim et al

Propensity Score-Matching Analyses

CLINICAL LIVER

Propensity score matching of the entire study population
yielded 1792 matched pairs of patients. The entecavir and
lamivudine groups within this matched overall cohort did
not differ significantly in terms of their baseline characteristics (Table 3 and Supplementary Figure 1).
The risk of death or transplantation and HCC was
assessed by multivariable Cox regression model adjusting
for the overall treatment duration in each propensity scorematched pair. For the matched pairs of the overall cohort,
the risk of death or transplantation diverged after 3 years of
follow-up evaluation, and was significantly lower in the
entecavir group than in the lamivudine group (HR, 0.49;
95% CI, 0.37–0.64; P < .001) (Figure 2A and Table 4). By
contrast, the 2 groups did not differ significantly in the risk
of HCC throughout the follow-up period (HR, 1.01; 95% CI,
0.80–1.27; P ¼ .95) (Figure 2B and Table 4).

Competing Risks Analysis
The 3 competing outcomes in this study were death,
transplantation, and HCC. It is possible that the more
frequent occurrence of death or transplantation in
lamivudine-treated patients may have reduced the number
of patients at risk of HCC in this group. Thus, the risks were
adjusted using competing risks analysis (death and transplantation). Again, the entecavir- and lamivudine-treated
groups had a similar risk of HCC (HR, 1.10; 95% CI,
0.89–1.36; P ¼ .39) (Supplementary Figure 2).

Subcohort Analyses According to Cirrhosis
The entire cohort was subdivided according to the
presence of cirrhosis at baseline and separate propensity
score matching was performed for the 2 subcohorts. There
were 860 and 878 matched pairs of patients with and
without cirrhosis, respectively. The entecavir- and
lamivudine-treated groups in each matched cohort did not
differ significantly in terms of baseline characteristics
(Table 3 and Supplementary Figure 1).
Analysis of matched patients without cirrhosis showed
that entecavir and lamivudine did not differ in terms of risk
of death or transplantation (HR, 0.86; 95% CI, 0.39–1.89;
P ¼ .71) (Figure 3A) or HCC (HR, 1.26; 95% CI, 0.68–2.34;
P ¼ .46) (Figure 3B) after adjusting for overall treatment
duration. The incidence of HCC was 0.71% per year and
0.80% per year for the entecavir and lamivudine groups,
respectively (Table 4).
Adjusted analysis of matched pairs of patients with
cirrhosis showed that entecavir was associated with a
significantly lower risk of death or transplantation (HR,
0.42; 95% CI, 0.31–0.57; P < .001) (Figure 3C and Table 4)
than lamivudine. However, the 2 treatment groups did not
differ in terms of the risk of HCC (HR, 1.00; 95% CI,
0.78–1.28; P ¼ .999) (Figure 3D and Table 4), with an
incidence rate of 4.10% per year and 4.35% per year for the
entecavir and lamivudine groups, respectively.
The cirrhosis subcohort then was divided further
according to the presence of decompensation at baseline
and separate propensity score-matching analyses were

Gastroenterology Vol. 147, No. 1

performed. There were 575 and 275 matched pairs of patients with compensated and decompensated cirrhosis,
respectively. The entecavir- and lamivudine-treated groups
within each matched cohort did not differ significantly in
terms of baseline characteristics (Supplementary Table and
Supplementary Figure 1).
In matched pairs of patients with compensated cirrhosis
and decompensated cirrhosis, entecavir was associated with
a significantly lower risk of death or transplantation than
lamivudine (HR, 0.37; 95% CI, 0.21–0.65; P < .001; and HR,
0.43; 95% CI, 0.30–0.63; P < .001, respectively) (Table 4
and Supplementary Figure 3A and B). However, none of
the treatment groups within any of the subcohorts differed
in terms of HCC risk (P ¼ .66 and P ¼ .38 for the
compensated and decompensated cirrhosis subcohorts,
respectively) (Table 4 and Supplementary Figure 3A and B).

Subcohort Analyses According to Complete
Virologic Response
Polymerase chain reaction data regarding complete
virologic responses (serum HBV-DNA level, <60 IU/mL)
were available after 2007. Therefore, we performed a
separate propensity score-matched analysis in patients who
achieved a complete virologic response (serum HBV-DNA
level, <60 IU/mL) at 1 year of treatment. A total of 275
matched pairs of patients showed a complete virologic
response and there were no significant differences in the
baseline characteristics between the 2 treatment groups
(Supplementary Table and Supplementary Figure 1). There
was no difference between the entecavir and lamivudine
groups in terms of the incidence of death or transplantation
(HR, 0.61, 95% CI, 0.32–1.17; P ¼ .14) and HCC (HR, 1.06,
95% CI, 0.65–1.74; P ¼ .81) (Table 4 and Supplementary
Figure 3C).

Discussion
Here, we compared midterm clinical outcomes for a
large cohort of CHB patients after they started treatment
with entecavir or lamivudine. This observational study
showed that entecavir was associated with a minimal risk of
drug resistance, a minimal need for rescue therapy, and a
significantly lower risk of death or transplantation. By
contrast, entecavir was no better than lamivudine at
reducing the risk of HCC. These results were observed
consistently in the analyses by multivariable Cox proportional hazards model for the entire cohort and for propensity score-matched cohorts.
Many earlier studies including a randomized trial and
cohort studies have shown that treatment with nucleos(t)
ide analogues improves clinical outcomes in patients with
CHB compared with no treatment.3–6,29,30 Two recent
nationwide cohort studies also confirmed these findings in
geographically and racially diverse patients with or without
cirrhosis.31,32 Among different nucleos(t)ide analogues,
entecavir and tenofovir show greater antiviral potency and
markedly lower resistance rates than lamivudine and
adefovir7–10; thus, they have been recommended as the

July 2014

Table 3.Characteristics of the Propensity Score-Matched Cohorts
Overall cohort
Characteristics

Lamivudine
(n ¼ 1792)

P value

46.1  10.1
1193 (66.6%)
1133 (63.2%)
7.28  1.55

46.1  10.9
1179 (65.8%)
1107 (61.8%)
7.30  1.18

.98
.64
.32
.57

42.2
581
643
7.72

103 (55–195)
3.8 (3.3–4.1)
1.2 (0.9–1.6)
1.10 (1.00–1.20)
0.86 (0.70–1.00)
143 (95–184)
68 (3.8%)
77 (4.3%)
933 (52.1%)
632 (67.7%)
301 (32.3%)
244 (13.6%)
5 (5–6)
19.8  14.2
94.4  22.1

118 (65–223)
3.8 (3.3–4.1)
1.1 (0.9–1.6)
1.10 (1.00–1.20)
0.90 (0.80–1.00)
140 (96–184)
73 (4.1%)
91 (5.1%)
934 (52.1%)
634 (67.9%)
300 (32.1%)
239 (13.3%)
5 (5–6)
20.1  14.4
94.4  22.9

.93
.86
.59
.85
.59
.68
.73
.31
1.00
.95

156
3.9
1.1
1.07
0.9
175
25
28

.85
.65
.56
.97

Entecavir
(n ¼ 878)
 10.6
(66.2%)
(73.2%)
 1.61

(95–290)
(3.6–4.1)
(0.9–1.4)
(1.00–1.10)
(0.7–1.0)
(143–209)
(2.8%)
(3.2%)
0
0
5 (5–5)
8.8  8.4
74.6  12.3

Cirrhosis cohort

Lamivudine
(n ¼ 878)
42.2
578
656
7.68

P value

 10.6
(65.8%)
(74.7%)
 1.08

.93
.91
.45
.47

172
3.9
1.0
1.08
0.9
176
24
32

(96–321)
(3.6–4.1)
(0.8–1.3)
(1.00–1.10)
(0.7–1.0)
(140–208)
(2.7%)
(3.6%)
0
0
5 (5–5)
9.5  8.8
74.5  11.8

.77
.82
.52
.77
.79
.84
1.00
.98
NA

NA
NA
.09
.81

ALT, alanine aminotransferase; CTP, Child–Turcotte–Pugh; INR, international normalized ratio; IQR, interquartile range.

Entecavir
(n ¼ 860)
49.9
565
437
6.92

 8.5
(65.7%)
(50.8%)
 1.31

Lamivudine
(n ¼ 860)
49.7
572
447
6.91

 9.6
(66.5%)
(52.0%)
 1.14

73 (44–124)
91 (55–153)
3.6 (3.0–4.0)
3.6 (3.0–4.0)
1.4 (1.0–1.9)
1.3 (0.9–1.9)
1.20 (1.10–1.30) 1.20 (1.10–1.30)
0.8 (0.7–0.9)
0.9 (0.7–1.0)
107 (73–146)
104 (72–140)
49 (5.7%)
43 (5.0%)
55 (6.4%)
49 (5.7%)
860 (100%)
860 (100%)
589 (68.5%)
572 (66.5%)
271 (31.5%)
288 (33.5%)
221 (25.7%)
216 (25.1%)
6 (5–7)
6 (5–7)
30.1  10.4
30.3  10.6
112.9  10.5
112.6  10.7

P value
.61
.76
.63
.72
.48
.76
.95
.98
.83
.51
.59
.62
NA
.38
.82
.66
.63
.49

Entecavir vs Lamivudine and Clinical Outcome

Age, mean  standard deviation, y
Male sex
HBeAg level
HBV-DNA level, mean  standard
deviation, log10 IU/mL
ALT level, median (IQR), IU/mL
Albumin level, median (IQR), g/dL
Total bilirubin level, median (IQR), mg/dL
INR, median (IQR),
Creatinine level, median (IQR), mg/dL
Platelets, median (IQR), 1000/mm3
Diabetes mellitus
Hypertension
Cirrhosis
Compensated
Decompensated
Ascites
CTP score, median (IQR)
CU-HCC score17
GAG-HCC score18

Entecavir
(n ¼ 1792)

Noncirrhosis cohort

157

CLINICAL LIVER

158

Lim et al

CLINICAL LIVER

Figure 2. Cumulative incidence of death or transplantation
and HCC in the overall propensity score-matched cohort.
Propensity score matching of the entire cohort created 1792
matched pairs of patients. (A) Cumulative incidence of death
or transplantation. (B) Cumulative incidence of HCC.

first-line agents for the treatment of CHB.11–13 However,
few studies have examined whether treatment with entecavir or tenofovir rather than earlier-generation drugs results in a better clinical outcome. Recently, 2 cohort studies
showed that, compared with no treatment, treatment with
entecavir reduced the risk of HCC and death in patients
with cirrhosis.29,30 One of these studies also suggested that
entecavir was associated with a lower risk of HCC than
lamivudine.29 Although this was a risk-adjusted analysis, the
comparison was underpowered (it included a small number
of patients and events); only 12 and 19 patients developed
HCC in the entecavir and lamivudine groups, respectively.
Moreover, the data were prone to bias because patients
who received rescue therapy were excluded. By contrast, a
recent meta-analysis found no difference in the incidence
of HCC between antiviral agents.33 This observation is
consistent with our own.
A reduction in hepatic necroinflammation is associated with
a virologic response.7,9,34 Patients harboring drug-resistant

Gastroenterology Vol. 147, No. 1

mutants are more likely to suffer hepatitis flares, disease progression, and to die than those without resistant mutants.3,5,6
Thus, our results showing that patients treated with entecavir
have a significantly lower risk of death or transplantation than
those treated with lamivudine may reflect the higher virologic
response rates and markedly lower resistance rates in entecavirtreated patients. These factors may reduce and prevent hepatitis
flares and hepatic necroinflammation, thereby slowing progression to decompensation, which then may result in a further
reduction in the risk of liver failure and death.
By contrast, treatment with entecavir did not reduce the
risk of HCC to a greater extent than lamivudine in any of the
cohorts in any of the analyses. This may be explained, at
least partly, by the fact that rescue therapies were widely
used during treatment with lamivudine. However, other
viral or host factors that are not readily amenable to the
suppression of viral replication may be associated with a
continued risk of HCC. These factors may include the early
integration of HBV DNA into the host genome, which results
in genomic alterations and/or chromosomal instability, and
pre-existing cirrhosis.35–38 In patients who have been
infected with HBV for a long time, or who already have
progressed to cirrhosis, many transformed hepatocyte
clones carrying genetic abnormalities that predispose to
cancer may be present before the onset of treatment.38,39
Thus, our results suggest that even after very effective
suppression of viral replication by oral antiviral agents, the
risk of HCC may not be reduced dramatically in CHB patients, particularly in those with cirrhosis. Further studies
are needed to determine whether these results are applicable to other potent antiviral agents, such as tenofovir.
The current consensus is that regular surveillance for
the early detection of HCC may be cost effective when the
annual risk exceeds 1.5% in patients with cirrhosis and 0.2%
in patients without cirrhosis.19,40 We showed that the annual
incidence of HCC in both the entecavir and lamivudine
cohorts was higher than 4% in patients with cirrhosis and
0.7% in those without cirrhosis. These observations, and
those of others,33 suggest that the risk of HCC persists
regardless of the drugs being used to treat CHB patients;
therefore, surveillance for early detection may be justified.
This study had several strengths. First, entecavir was
compared directly with lamivudine in a large cohort with a
high number of events. Second, there were very few missing
data and follow-up evaluation was almost complete. Third,
strict inclusion criteria were applied. Patients who died or
underwent liver transplantation within 6 months of treatment were assumed to be too sick to observe the clinical
benefit from any antiviral treatment, and were excluded
from the study. Patients who were diagnosed as HCC within
the first year of treatment also were excluded, based on the
assumption that these patients may have had pre-existing
undetected HCC on recruitment to the study. Finally, multiple strategies were used to minimize confounders (multivariable analyses, propensity score-matching analyses
adjusted for overall treatment duration) and competing
risks analyses were used to rigorously adjust for all risks.
The major limitation of this study was that it was based
on observational data. Thus, our findings are potentially

July 2014

Entecavir vs Lamivudine and Clinical Outcome

159

Table 4.Cumulative Incidence and Hazards Ratios for the Clinical Outcomes in the Propensity Score-Matched Cohorts

No.
No. of /100 patientyears
Outcomes Patient-years events
Overall cohort (N ¼ 1792 pairs)
Entecavir
5874
76
1.29
Lamivudine
9920
188
1.90
Noncirrhosis subcohort (N ¼ 878 pairs)
Entecavir
2975
9
0.30
Lamivudine
5109
17
0.33
Cirrhosis subcohort (N ¼ 860 pairs)
Entecavir
2779
60
2.16
Lamivudine
4530
162
3.58
Compensated cirrhosis subcohort (N ¼ 575 pairs)
Entecavir
1830
16
0.87
Lamivudine
3166
64
2.02
Decompensated cirrhosis subcohort (N ¼ 275 pairs)
Entecavir
900
44
4.89
Lamivudine
1286
101
7.86
Complete virologic responsea subcohort (N ¼ 275 pairs)
Entecavir
930
13
1.40
Lamivudine
1349
30
2.22

HR
(95% CI)

Hepatocellular carcinoma
No.
Patient- No. of /100 patientyears
P value years events

HR
(95% CI)

P value

0.49
(0.37–0.64)

<.001

5687
9503

137
234

2.41
2.46

1.01
(0.80–1.27)

.95

0.86
(0.39–1.89)

.71

2947
5031

21
40

0.71
0.80

1.26
(0.68–2.34)

.46

0.42
(0.31–0.57)

<.001

2635
4211

108
183

4.10
4.35

1.00
(0.78–1.28)

.999

0.37
(0.21–0.65)

<.001

1747
2957

58
105

3.32
3.55

0.92
(0.65–1.31)

.66

0.43
(0.30-0.63)

<.001

830
1182

49
75

5.90
6.34

0.84
(0.56-1.25)

.38

0.61
(0.32–1.17)

.14

875
1273

33
42

3.77
3.30

1.06
(0.65–1.74)

.81

NOTE. The hazards ratios and P values represent the entecavir group compared with the lamivudine group, adjusted for overall
treatment duration using a multivariable Cox regression model.
a
Serum HBV-DNA level <60 IU/mL at 1 year of treatment.

Figure 3. Cumulative incidence of death or transplantation and HCC in
propensity score-matched
cohorts with and without
cirrhosis. Propensity score
matching created 860 and
878 matched pairs of patients with and without
cirrhosis, respectively. (A)
Cumulative incidence of
death or transplantation in
the
noncirrhosis subcohort. (B) Cumulative
incidence of HCC in the
noncirrhosis
subcohort.
(C) Cumulative incidence
of death or transplantation
in the cirrhosis subcohort.
(D) Cumulative incidence
of HCC in the cirrhosis
subcohort.

CLINICAL LIVER

Death or transplantation

160

Lim et al

CLINICAL LIVER

subject to selection bias and confounding. The propensity
score matching showed an excellent balance between the
entecavir and lamivudine groups and eliminated many of
the baseline differences. However, there still may be some
hidden bias as a result of unmeasured confounders. Despite
these limitations, the present observational study is arguably the best way to compare lamivudine and entecavir
because a randomized trial was regarded as being unethical
given the proven superior efficacy of entecavir over
lamivudine in terms of virologic response.7,34 Another limitation was the limited duration of overall treatment and
follow-up in the entecavir group. Although the results did
not change after adjusting for treatment duration between
the groups, the shorter treatment duration of the entecavir
group might have prevented us from detecting a significant
difference in HCC incidence between this group and the
lamivudine group. It is possible that continuous treatment
with entecavir or tenofovir for more than a decade eventually may reduce the incidence of HCC by inducing
regression of cirrhosis.41,42 Finally, as a single-center study,
our study may have limitations for generalization of results.
The predominant population included in this study had
genotype C HBV, acquired through a vertical mode of
transmission,21 which might have been associated with a
high risk of HCC.43,44
In conclusion, the present midterm study of a large
cohort of CHB patients consistently showed that treatment
with entecavir was associated with a significantly lower risk
of death or transplantation than treatment with lamivudine;
however, the risk of HCC was high and did not differ between the 2 treatment groups. These results advocate current practice guidelines that recommend the use of 2 potent
antiviral agents (ie, entecavir and tenofovir), as first-line
drugs for the treatment of CHB.11–13 Our data also may
support recommendations that surveillance for HCC should
be continued in CHB patients regardless of the treatment
used.19,40,45 However, the risk for HCC may not be the same
for CHB patients with different host and viral characteristics; therefore, recommendations for HCC surveillance may
not be generalized in noncirrhotic CHB patients. Further
longer-term studies on patients of other races or ethnicity,
and studies using other potent antiviral agents such as
tenofovir, are warranted.

Supplementary Material
Note: To access the supplementary material accompanying
this article, visit the online version of Gastroenterology at
www.gastrojournal.org, and at http://dx.doi.org/10.1053/
j.gastro.2014.02.033.

References
1. Perz JF, Armstrong GL, Farrington LA, et al. The contributions of hepatitis B virus and hepatitis C virus infections to cirrhosis and primary liver cancer worldwide.
J Hepatol 2006;45:529–538.
2. Schiff ER. Prevention of mortality from hepatitis B and
hepatitis C. Lancet 2006;368:896–897.

Gastroenterology Vol. 147, No. 1
3. Liaw YF, Sung JJ, Chow WC, et al. Lamivudine for patients with chronic hepatitis B and advanced liver disease. N Engl J Med 2004;351:1521–1531.
4. Eun JR, Lee HJ, Kim TN, et al. Risk assessment for the
development of hepatocellular carcinoma: according to
on-treatment viral response during long-term lamivudine
therapy in hepatitis B virus-related liver disease.
J Hepatol 2010;53:118–125.
5. Papatheodoridis GV, Dimou E, Dimakopoulos K, et al.
Outcome of hepatitis B e antigen-negative chronic hepatitis B on long-term nucleos(t)ide analog therapy starting
with lamivudine. Hepatology 2005;42:121–129.
6. Yuen MF, Seto WK, Chow DH, et al. Long-term lamivudine
therapy reduces the risk of long-term complications of
chronic hepatitis B infection even in patients without
advanced disease. Antivir Ther 2007;12:1295–1303.
7. Chang TT, Gish RG, de Man R, et al. A comparison of
entecavir and lamivudine for HBeAg-positive chronic
hepatitis B. N Engl J Med 2006;354:1001–1010.
8. Chang TT, Lai CL, Yoon SK, et al. Entecavir treatment for
up to 5 years in patients with hepatitis B e antigen-positive
chronic hepatitis B. Hepatology 2010;51:422–430.
9. Marcellin P, Heathcote EJ, Buti M, et al. Tenofovir disoproxil fumarate versus adefovir dipivoxil for chronic
hepatitis B. N Engl J Med 2008;359:2442–2455.
10. Kitrinos KM, Corsa A, Liu Y, et al. No detectable resistance to tenofovir disoproxil fumarate after 6 years of
therapy in patients with chronic hepatitis B. Hepatology
2014;59:434–442.
11. European Association for the Study of the Liver. EASL
Clinical Practice Guidelines: management of chronic
hepatitis B virus infection. J Hepatol 2012;57:167–185.
12. Korean Association for the Study of the Liver. KASL
Clinical Practice Guidelines: management of chronic
hepatitis B. Clin Mol Hepatol 2012;18:109–162.
13. Lok AS, McMahon BJ. Chronic hepatitis B: update 2009.
Hepatology 2009;50:661–662.
14. Seo HJ, Oh IH, Yoon SJ. A comparison of the cancer
incidence rates between the National Cancer Registry
and insurance claims data in Korea. Asian Pac J Cancer
Prev 2012;13:6163–6168.
15. Song BC, Suh DJ, Lee HC, et al. Hepatitis B e antigen
seroconversion after lamivudine therapy is not durable in
patients with chronic hepatitis B in Korea. Hepatology
2000;32:803–806.
16. Wong GL, Chan HL, Chan HY, et al. Accuracy of risk scores
for patients with chronic hepatitis B receiving entecavir
treatment. Gastroenterology 2013;144:933–944.
17. Wong VW, Chan SL, Mo F, et al. Clinical scoring system
to predict hepatocellular carcinoma in chronic hepatitis B
carriers. J Clin Oncol 2010;28:1660–1665.
18. Yuen MF, Tanaka Y, Fong DY, et al. Independent risk
factors and predictive score for the development of hepatocellular carcinoma in chronic hepatitis B. J Hepatol
2009;50:80–88.
19. Bruix J, Sherman M. Management of hepatocellular
carcinoma: an update. Hepatology 2011;53:1020–1022.
20. Saldanha J, Gerlich W, Lelie N, et al. An international
collaborative study to establish a World Health Organization international standard for hepatitis B virus DNA

21.

22.

23.

24.

25.

26.
27.

28.
29.

30.

31.

32.

33.

34.

35.

36.

nucleic acid amplification techniques. Vox Sang 2001;
80:63–71.
Kim H, Jee YM, Song BC, et al. Molecular epidemiology
of hepatitis B virus (HBV) genotypes and serotypes in
patients with chronic HBV infection in Korea. Intervirology 2007;50:52–57.
Hong SP, Kim NK, Hwang SG, et al. Detection of hepatitis B virus YMDD variants using mass spectrometric
analysis of oligonucleotide fragments. J Hepatol 2004;
40:837–844.
Fletcher RH, Fletcher SW. Clinical epidemiology: the
essentials. Philadelphia, PA: Lippincott Williams & Wilkins, 2005.
Cepeda MS, Boston R, Farrar JT, et al. Comparison of
logistic regression versus propensity score when the
number of events is low and there are multiple confounders. Am J Epidemiol 2003;158:280–287.
Hamilton BH, Ko CY, Richards K, et al. Missing data in
the American College of Surgeons National Surgical
Quality Improvement Program are not missing at
random: implications and potential impact on quality
assessments. J Am Coll Surg 2010;210:125–139.
Therneau TM, Grambsch PM. Modeling survival data:
extending the Cox model. New York, NY: Springer, 2000.
Klein JP, Moeschberger ML. Survival analysis: techniques for censored and truncated data. New York, NY:
Springer, 2003.
Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci 2010;25:1–21.
Hosaka T, Suzuki F, Kobayashi M, et al. Long-term
entecavir treatment reduces hepatocellular carcinoma
incidence in patients with hepatitis B virus infection.
Hepatology 2013;58:98–107.
Wong GL, Chan HL, Mak CW, et al. Entecavir treatment reduces hepatic events and deaths in chronic hepatitis B patients with liver cirrhosis. Hepatology 2013;58:1537–1547.
Gordon SC, Lamerato LE, Rupp LB, et al. Antiviral Therapy
for Chronic Hepatitis B Virus Infection and Development
of Hepatocellular Carcinoma in a US Population. Clin
Gastroenterol Hepatol 2014;12:885–893.
Wu CY, Lin JT, Ho HJ, et al. Association of Nucleos(t)ide
Analogue Therapy with Reduced Risk of Hepatocellular
Carcinoma in Patients with Chronic Hepatitis B—a
Nationwide Cohort Study. Gastroenterology 2014;147:
143–151.
Singal AK, Salameh H, Kuo YF, et al. Meta-analysis: the
impact of oral anti-viral agents on the incidence of hepatocellular carcinoma in chronic hepatitis B. Aliment
Pharmacol Ther 2013;38:98–106.
Lai CL, Shouval D, Lok AS, et al. Entecavir versus lamivudine for patients with HBeAg-negative chronic hepatitis B. N Engl J Med 2006;354:1011–1020.
Levrero M, Pollicino T, Petersen J, et al. Control of
cccDNA function in hepatitis B virus infection. J Hepatol
2009;51:581–592.
Neuveut C, Wei Y, Buendia MA. Mechanisms of HBVrelated hepatocarcinogenesis. J Hepatol 2010;52:594–604.

Entecavir vs Lamivudine and Clinical Outcome

161

37. Brechot C. Pathogenesis of hepatitis B virus-related
hepatocellular carcinoma: old and new paradigms.
Gastroenterology 2004;127:S56–S61.
38. Papatheodoridis GV, Lampertico P, Manolakopoulos S,
et al. Incidence of hepatocellular carcinoma in chronic
hepatitis B patients receiving nucleos(t)ide therapy: a
systematic review. J Hepatol 2010;53:348–356.
39. Hoshida Y, Villanueva A, Kobayashi M, et al. Gene
expression in fixed tissues and outcome in hepatocellular
carcinoma. N Engl J Med 2008;359:1995–2004.
40. European Association for the Study of the Liver. EASLEORTC clinical practice guidelines: management of hepatocellular carcinoma. J Hepatol 2012;56:908–943.
41. Chang TT, Liaw YF, Wu SS, et al. Long-term entecavir
therapy results in the reversal of fibrosis/cirrhosis and
continued histological improvement in patients with
chronic hepatitis B. Hepatology 2010;52:886–893.
42. Marcellin P, Gane E, Buti M, et al. Regression of cirrhosis
during treatment with tenofovir disoproxil fumarate for
chronic hepatitis B: a 5-year open-label follow-up study.
Lancet 2013;381:468–475.
43. Lin CL, Kao JH. The clinical implications of hepatitis B
virus genotype: recent advances. J Gastroenterol Hepatol 2011;26(Suppl 1):123–130.
44. McGlynn KA, London WT. The global epidemiology of
hepatocellular carcinoma: present and future. Clin Liver
Dis 2011;15:223–243.
45. Korean Liver Cancer Study Group. National Cancer
Center. Practice guidelines for management of hepatocellular carcinoma 2009. Korean J Hepatol 2009;15:
391–423.

Author names in bold designate shared co-first authorship.
Received September 29, 2013. Accepted February 4, 2014.
Reprint requests
Address requests for reprints to: Young-Suk Lim, MD, PhD, Associate Professor,
Department of Gastroenterology, Asan Medical Center, 88, Olympic-Ro 43-Gil,
Songpa-Gu, Seoul 138-736, Korea. e-mail: [email protected]; fax: (82) 02485-5782.
Acknowledgments
The authors thank the Information Technology Service Management team of
Asan Medical Center and Drs G. A. Kim, J. An, D. B. Lee, K. M. Kim, Y. H.
Chung, and Y. S. Lee for their help in data collection, and Drs D. W. Park
and S. C. Yun for critically reviewing the statistical analyses. None received
compensation for their work.
Presented in part at the annual meeting of the American Association for the
Study of the Liver, Washington, DC, November 03, 2013.
Conflicts of interest
These authors disclose the following: Young-Suk Lim is an advisory board
member of Bayer Healthcare, Bristol-Myers Squibb, and Gilead Science, and
receives investigator-initiated research funding from Bayer Healthcare,
Bristol-Myers Squibb, Gilead Science, and Novartis. The remaining authors
disclose no conflicts.
The interpretation and reporting of the data were the sole responsibility of the
authors.
Funding
This research was supported by grants from the Asan Institute for Life Sciences
of Asan Medical Center and the Korean Association for the Study of Liver.

CLINICAL LIVER

July 2014