J Antimicrob Chemother 2016; 71: 1601 – 1608

doi:10.1093/jac/dkw019 Advance Access publication 2 March 2016

a-Fetoprotein level-dependent early hepatitis B surface antigen decline

during entecavir therapy in chronic hepatitis B with hepatitis flare
Wen-Juei Jeng, Yi-Cheng Chen, Ming-Ling Chang and Yun-Fan Liaw*

Liver Research Unit, Chang Gung Memorial Hospital, Chang Gung University College of Medicine, Taipei, Taiwan

Downloaded from https://academic.oup.com/jac/article-abstract/71/6/1601/1751338 by guest on 18 August 2019

*Corresponding author. Tel: +886-3-3281200, ext. 8120; Fax: +886-3-3282824; E-mail: liveryfl@gmail.com

Received 31 July 2015; returned 16 October 2015; revised 11 January 2016; accepted 15 January 2016

Objectives: Hepatitis B surface antigen (HBsAg) reduction during nucleos(t)ide analogue therapy is related to ALT
level. ALT reflects hepatocytolysis while a-fetoprotein (AFP) ≥100 ng/mL during hepatitis flare reflects more
extensive hepatocytolysis (bridging hepatic necrosis). The impact of AFP levels on early HBsAg kinetics during
entecavir therapy was investigated.
Methods: HBsAg level was measured at baseline and months 6 and 12 of entecavir therapy in 149 chronic
hepatitis B patients with hepatitis flare, defined as ALT ≥5× upper limit of normal (ULN), and 58 patients with
ALT ,5× ULN.
Results: There was a significantly greater HBsAg reduction in an ALT (,5, 5– 10, 10 –20 and ≥20× ULN, P ¼ 0.001)
and AFP (,20, 20–99 and ≥100 ng/mL, P ¼0.000) level-dependent manner. In hepatitis flares with a peak AFP
level ≥20 ng/mL, the differences in HBsAg reduction across all ALT levels became non-significant. HBsAg reduc-
tion was greater in genotype B- than genotype C-infected patients with baseline ALT ≥20× ULN, but the differ-
ence became non-significant in those with peak AFP ≥100 ng/mL. Multivariate linear regression analysis showed
that AFP level ≥100 ng/mL, baseline HBsAg level and genotype B were independent significant factors for greater
HBsAg decline at month 6 of entecavir therapy.
Conclusions: During entecavir therapy, early HBsAg reduction increased in an AFP and ALT level-dependent
manner, suggesting the impact of hepatocytolysis rather than nucleos(t)ide analogue per se. Notably, a higher
AFP level during hepatitis flare, reflecting more extensive hepatic necrosis, was a more powerful factor than ALT
and genotype for greater HBsAg decline.

Introduction decline in patients with higher ALT levels.5 – 7 A sharp contrast

Serum hepatitis B surface antigen (HBsAg) level correlates with
the intrahepatic transcriptional activity of covalently closed circu-
lar DNA (cccDNA) and is considered a surrogate marker of infected
cells.1,2 A longitudinal study on the natural course of chronic hepa-
titis B virus (HBV) infection showed that a slow overall decrease
in HBsAg levels and a .1 log10 IU/mL HBsAg decline reflects
improved immune control, which was associated with a higher
HBsAg seroclearance rate.3 Subsequent studies have further
shown that antiviral therapy may enhance HBsAg decline, greater
with IFN-based therapy than with nucleos(t)ide analogue (Nuc)
therapy.2 A study showed that the HBsAg decline in hepatitis B e
antigen (HBeAg)-positive patients during entecavir therapy was
primarily confined to patients with baseline ALT levels greater
than two times the upper limit of normal (ULN) and suggested
that an active pre-existing immune activity might be required to
lower HBsAg levels during entecavir therapy.4 Subsequent studies
have confirmed this finding and further shown greater HBsAg
was demonstrated in a mean decline of 0.58 log10 IU/mL in
HBeAg-negative chronic hepatitis B patients with severe reactiva-
tion (ALT .10× ULN) compared with 0.01 log10 IU/mL in patients
with ALT ≤3×ULN at month 6 of Nuc therapy.7 These findings sug-
gest that greater hepatocytolysis, as reflected in higher ALT levels,
is associated with greater HBsAg decline during Nuc therapy. This
is consistent with the notion that cccDNA may be reduced by hep-
atocyte killing and further diluted by subsequent regenerations.8
Hence, it is possible that patients with more extensive hepatocy-
tolysis, such as bridging hepatic necrosis (BHN), may have even
greater HBsAg decline during Nuc therapy.
It is well known that ALT levels reflect the degree or extent of
hepatocytolysis. However, it seems much less noticed that serum
a-fetoprotein (AFP) is not only a seromarker of hepatocellular car-
cinoma, but a properly measured peak AFP level during ALT flare
≥5×ULN is also closely correlated with BHN.9 A clinicopathological
study involving 182 chronic hepatitis B patients with hepatitis flare
showed that BHN was found in 84% of those with AFP .100 ng/mL,

# The Author 2016. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved.
For Permissions, please e-mail: journals.permissions@oup.com

1601
Jeng et al.
46% of those with AFP 21 –100 ng/mL and only 9% of those with
AFP ≤20 ng/mL.9,10 Given these findings, together with a high
degree of AFP-producing progenitor oval cell activation in patients
with BHN, a recent review suggests that AFP .100 ng/mL during
hepatitis B flare can be used as a surrogate marker of BHN.11
Asian-Pacific HBV guidelines recommend that patients with ALT
.5× ULN should be monitored closely with weekly or biweekly
serum ALT, bilirubin and prothrombin time measurement.12 In
addition to these tests, we have included serial AFP assays to
detect the peak level of AFP since the relation of AFP and BHN dur-
ing hepatitis flare was reported three decades ago.9 With these
and the approval of the Ethics Committee of our hospital (CGMF
IRB no. 100-1848C), we were able to examine the impact of differ-
ent pretherapy serum ALT and AFP levels on HBsAg decline during
the early phase of entecavir therapy in patients with hepa-
titis flare.
Patients and methods
Patients
This retrospectively analysed study with prospective sample collection
included 207 chronic hepatitis B patients receiving entecavir treatment:
58 with ALT ,5× ULN and 149 with ALT ≥5× ULN, who had undergone
‘proper measurement of AFP’ and had HBsAg levels measured at baseline
and months 6 and 12 of entecavir therapy. When HBsAg data were miss-
ing, stored serum samples were used for retrospective HBsAg quantifica-
tion with patients’ informed consent. Those with serum samples at only
two timepoints (month 6 not available in 80 and month 12 not available
in 69) were excluded. ‘Hepatitis B flare’ was defined as an event with ALT
level ≥5× ULN.11 ‘Proper measurement of AFP’ was defined as at least
three serum AFP assays at 1 – 2 week intervals during hepatitis flare to
detect peak AFP levels.9 – 11,13 ‘Rapid HBsAg decline’ was defined as
≥0.5 log10 IU/mL reduction at month 6 or ≥1.0 log10 IU/mL reduction at
month 12 of Nuc therapy.14 Cirrhosis was diagnosed by histological find-
ings or repeated ultrasonography findings consistent with cirrhosis, sup-
plemented with clinical features such as varices and thrombocytopenia.
Those with evidence of other liver disease(s), such as concurrent chronic
infection with other virus(es), alcoholic, metabolic or autoimmune liver
disease and hepatocellular carcinoma, were excluded.
Laboratory methods
Biochemical tests were performed using automated techniques at our
clinical pathology laboratories. The serum ALT ULN was set by the labora-
tory at 36 U/L for both males and females. Commercial kits were used for
serum HBsAg, HBeAg, Anti-HCV assay (Abbott Laboratories, North Chicago,
IL, USA), AFP level (Abbott Laboratories, lower limit of detection: 2 ng/mL),
serum HBV DNA level (COBAS TaqManw; Roche Diagnostics, Indianapolis,
IN, USA; lower limit of detection: 20 IU/mL) and serum HBsAg level
(lower limit of detection: 0.05 IU/mL, Elecsys HBsAg II Quant assay,
Roche Diagnostics, Mannheim, Germany). HBV genotype was determined
using PCR-RFLP of the surface gene of HBV. All biochemistry tests and
serum HBV DNA assays were conducted at the time of outpatient visit.
Less than 40% HBsAg quantification were assayed using recently (within
1 year) stored serum samples.
Statistical analysis
Baseline factors analysed for HBsAg reduction at months 6 and 12 of ente-
cavir therapy included: HBV DNA and HBsAg levels; HBeAg status; ALT and
peak AFP levels; evidence of liver cirrhosis; age; and gender. Further, the
correlation between ALT and peak AFP was analysed by the stratification
1602
method. Statistical analysis was performed with the x2 test or Fisher’s
exact test and independent Student’s t-test or Mann–Whitney U-test for
the categorical and continuous variables, respectively, among patients
with hepatitis B flares with different ALT (5 – 10×, 10 – 20× and .20×
ULN) and AFP (,20, 20–99 and ≥100 ng/mL) levels.9 – 11 Continuous vari-
ables are shown as the median (range). Multiple linear regression analysis
was performed to find the predictor of HBsAg log reduction. Variables that
were previously identified as significant factors in univariate analysis
(P, 0.05) and factors reported to be significant in earlier studies were
included in multivariate analyses.
Downloaded from https://academic.oup.com/jac/article-abstract/71/6/1601/1751338 by guest on 18 August 2019
Results
Baseline characteristics
The demographic features of the 149 chronic hepatitis B patients
with hepatitis flare and 58 patients with baseline ALT ,5× ULN are
shown in Table 1. Of the 149 patients with hepatitis flare, 30.2% had
a baseline ALT ≥20× ULN, 34.9% ALT 10 – 20× ULN and 34.9%
5 –10× ULN. About half (57%) of them had a peak AFP level
,20 ng/mL and 39 patients (26.2%) had AFP ≥100 ng/mL. Of the
17 patients who had liver biopsy within 2 months prior to or after
starting therapy, 3 of the 5 patients with AFP ≥20 ng/mL were
reported to have BHN while none of the 12 patients with AFP
,20 ng/mL had BHN (P¼0.015).
HBsAg kinetics during the first year of entecavir
therapy
Baseline serum HBV DNA levels of the patients with hepatitis B
flare were significantly higher than those with ALT ≤5× ULN,
and became undetectable (,20 IU/mL) in 48.3% by month 6
and 77.9% by month 12 of entecavir therapy. The corresponding
figures were 67.2% and 82.8%, respectively, in those with baseline
ALT ,5×ULN. The time to undetectable HBV DNA was significantly
longer in patients with hepatitis flare compared to those without
flare (median time: 285 days versus 92 days, P ¼ 0.000).
In patients with pretherapy ALT ≥5× ULN, there was a steep
and significant HBsAg reduction (20.59 log10 IU/mL) in the first
6 months (P ¼ 0.000), but only small and non-significant further
HBsAg decline from month 6 to month 12 of entecavir therapy.
In contrast, patients with pretherapy ALT ,5× ULN showed
small and non-significant HBsAg declines of 20.06 and
20.09 log10 IU/mL from baseline to months 6 and 12, respect-
ively (Table 1 and Figure 1). As shown in Table 1, 54.4% and
34.5% of the patients with hepatitis flare showed ‘rapid HBsAg
decline’ at months 6 and 12 of entecavir therapy, respectively,
both significantly greater than that of patients with baseline ALT
,5× ULN. There was no significant difference in achieving ‘rapid
HBsAg decline’ at month 6 between HBeAg-positive and -negative
patients (60.8% versus 51%, P ¼ 0.256, in patients with hepatitis
flare and 22.2% versus 4.1%, P ¼ 0.110, in those without flare).
In HBeAg-positive patients, there was no statistical difference
in the HBsAg decline between ALT ≥20× ULN and ,20× ULN
(20.75 versus 20.47 log10 IU/mL, P¼0.263), but it was significantly
greater in HBeAg-negative patients with ALT ≥20× ULN than in
those with ALT ,20× ULN (20.8 versus 20.14 log10 IU/mL,
P ¼ 0.000). AFP ≥100 showed greater HBsAg reduction in both
HBeAg-positive (21.085 versus 20.44 log10 IU/mL, P¼0.006) and
HBeAg-negative (21.38 versus 20.145 log10 IU/mL, P ¼ 0.000)
patients.
AFP-dependent rapid HBsAg decline JAC
Table 1. Baseline demographic features and HBsAg reduction during entecavir therapy in patients with and those without hepatitis B flare

Hepatitis B flare

Baseline yes (n ¼149) no (n¼58) P

Age 53.9+11.9 53.6+11.4 0.504

Male 114 (76.5%) 53 (91.4%) 0.015
Genotype B/C 110/38 45/13 0.73
HBeAg positive 51 (34.2%) 9 (15.5%) 0.008
Cirrhosis 47 (31.5%) 31 (53.4%) 0.003

Downloaded from https://academic.oup.com/jac/article-abstract/71/6/1601/1751338 by guest on 18 August 2019

ALT (ULN) 455 (181– 4498) 85.5 (22– 174) 0.000
,5× 0 58 (100%) 0.000
5 –10× 52 (34.9%) 0
10– 20× 52 (34.9%) 0
≥20× 45 (30.2%) 0

AFPa (ng/mL) 10.2 (1.7 –3324) 3.45 (1.6– 17.8) 0.000

,20 85 (57%) 58 (100%) 0.000
20– 99 25 (16.8%) 0
≥100 39 (26.2%) 0

Platelets (103)/L 165 (67 –375) 173 (13 –285) 0.063

INR 1.1 (1–2.2) 1.1 (0.9– 1.4) 0.000
HBV DNA (log10 IU/mL) 7.55 (1.83– 9.73) 5.59 (3.45–8.54) 0.000

HBsAg (log10 IU/mL) 3.58 (0.48– 5.44) 3.02 (1.56–4.72) 0.000

decline, month 6 20.59 (22.94 –1.28) 20.06 (21.02– 0.54) 0.000
≥0.5 log10 81 (54.4%) 4 (6.9%) 0.000
≥1 log10 49 (32.9%) 2 (3.4%) 0.000
decline, month 12 20.64 (23.05 –1.62) 20.09 (21.34– 0.47) 0.000
≥0.5 log10 86 (57.7%) 5 (8.6%) 0.000
≥1 log10 51 (34.5%) 2 (3.4%) 0.000

INR, international normalized ratio.

Data expressed as n (%), mean+SD or median (range).
a
Peak level during flare episode.

HBsAg kinetics in relation to pretherapy ALT and peak AFP
level in patients with hepatitis B flare
Patients with higher pretherapy ALT, especially those with higher
AFP levels, showed greater HBsAg reduction in an ALT and AFP level-
dependent manner (Table 2). Patients with higher ALT (≥10×versus
5–10×ULN) also had more frequent ‘rapid HBsAg decline’ at month
6 and month 12 of entecavir therapy, being greatest in patients
with ALT ≥10× (ALT 5 –10×¼ 30.8%, 10– 20×¼ 63.5% and ≥20×
ULN ¼ 71.1%, P ¼ 0.000) and AFP ≥20 ng/mL (,20 ¼ 41.2%,
20 – 99 ¼ 60% and ≥100 ¼ 79.5% in patients with ALT ≥5× ULN,
P ¼ 0.000). The clinical courses of two representative patients
with hepatitis flare and a patient with pretherapy ,5× ULN are
shown in Figure 2. HBsAg reduction during the first year of entecavir
therapy occurred primarily by month 6. Comparisons of data at
month 6 showed greater reduction in patients with ALT levels
≥10× ULN than in those with ALT 5 – 10× ULN if peak AFP was
,20 ng/mL (P ¼ 0.012), but the differences were non-significant
in those with peak AFP ≥20 ng/mL. Compared with patients with
peak AFP ,20 ng/mL, patients with peak AFP ≥20 ng/mL showed
significantly greater HBsAg decline at month 6 of entecavir therapy
in patients with ALT ≥10× ULN, but not in those with ALT 5–10×
ULN. The differences were not significant across all categories of
ALT flares with AFP ≥20 ng/mL (Table 2).
HBsAg kinetics in relation to HBV genotypes
HBV genotype data were available in 149 patients with hepatitis B
flare (genotype B¼110, genotype C¼38 and not typeable¼1) and
58 patients without flare (genotype B¼45 and genotype C¼13).
Taking all patients together, HBsAg decline at month 6 of entecavir
therapy was greater in genotype B-infected patients (20.27 versus
20.16 log10 IU/mL, P¼0.097). In patients without hepatitis flare,
there was similar HBsAg reduction between genotypes B and C
(20.06 versus 20.08 log10 IU/mL, P¼0.867) while greater HBsAg
reduction was observed in genotype B patients with hepatitis
flare (20.715 versus 20.295 log10 IU/mL in genotype C,
P¼0.028), which was significantly greater in both HBeAg-positive
and -negative patients. However, breakdown data showed that
the difference was significant only in patients with ALT ≥20× ULN
(genotype B : genotype C ¼ 20.965 versus 0.04 log10 IU/mL,
P¼0.002). Of the patients with ALT ≥20× ULN, the different impact
of genotypes B and C was observed only in patients with
AFP ≥20 ng/mL (genotype B : genotype C ¼ 21.19 versus 0.07

1603
Jeng et al.

ALT and
4.00 AFP
category
1
2
3
3.75 4
5
Median HBsAg (log10 IU/mL)

3.50

Downloaded from https://academic.oup.com/jac/article-abstract/71/6/1601/1751338 by guest on 18 August 2019

3.25

Group 2
3.00

Group 1
Group 5
2.75 Group 3
Group 4

Baseline Month 6 Month 12

Month of entecavir therapy

HBsAg, median (range) log10 IU/mL

ALT AFP
Group Number
(ULN) (ng/mL) Baselinea Month 6b Month 12c D6M a vs b a vs c
1 <5× <20 58 3.03 (1.56–4.72) 2.97 (1.62–4.62) 2.89 (1.3–4.45) –0.06 0.483 0.209

2 <20 43 3.52 (1.44–4.38) 3 (2.1–4.62) 3.04 (2.16–4.22) –0.18 0.004 0.000

5–10×
3 ≥20 9 3.46 (2.27–4.62) 2.92 (–0.3–3.62) 2.74 (0.66–3.68) –0.45 0.136 0.113
4 <20 42 3.344 (1.15–4.98) 2.78 (1.69–3.76) 2.72 (1.61–3.83) –0.57 0.000 0.000
≥10×
5 ≥20 55 3.99 (0.48–5.44) 2.94 (–1.3–4.60) 2.78 (–1.12–4.61) –1.09 0.000 0.000
All groups comparing b vs c P > 0.1

Figure 1. Reduction of serum HBsAg from baseline to month 6 was highly significant, but the reduction from month 6 to 12 of entecavir therapy was not
significant in patients regardless of ALT and AFP levels. Greater reduction of HBsAg was noted in those with ALT ≥10× and AFP ≥20 ng/mL (21.09 log
reduction at month 6). vs, versus.

log10 IU/mL, P ¼ 0.008), among whom the difference became
non-significant in patients with AFP ≥100 ng/mL. Among patients
with flare who had AFP ,100 ng/mL, median HBsAg reduction
in genotype B was greater than in genotype C (20.545 versus
20.240 log10 IU/mL, P ¼ 0.042). In contrast, the difference was
not statistically significant in patients with AFP ≥100 ng/mL (geno-
type B:genotype C¼21.34 versus 20.92 log10 IU/mL, P¼0.399).
Factors predicting HBsAg decline at month 6 of
entecavir therapy
Since there was only a small and non-significant decline between
months 6 and 12 of entecavir therapy, only factors predicting
‘rapid HBsAg decline’ at month 6 were analysed. By multiple linear
regression analysis, genotype B, higher baseline HBsAg level,
higher baseline HBV DNA level, pretherapy ALT ≥10× ULN and
AFP ≥100 ng/mL were factors for HBsAg reduction in the univariate
analysis. In the multivariate analysis, genotype B [adjusted b (95%
CI): 0.164 (0.079–0.517), P¼0.008], higher baseline HBsAg level
[adjusted b (95% CI): 20.474 (20.642 to 20.335), P¼0.000] and
AFP ≥100 ng/mL [adjusted b (95% CI): 20.254 (20.705 to 20.215),
P¼0.000] were significant factors while ALT ≥10× ULN [adjusted b
(95% CI): 20.123 (20.41720.009), P¼0.060] was of borderline sig-
nificance for HBsAg reduction at month 6 of therapy (Table 3).
Of these independent factors for HBsAg reduction at month 6,
higher baseline HBsAg level [adjusted b 20.709 (20.809 to
20.444), P ¼ 0.000] and ALT ≥10× ULN [adjusted b 20.287
(20.496 to 20.147), P ¼0.000] were factors for HBsAg reduction
in patients with hepatitis flare whose AFP was ,20 ng/mL, while
higher baseline HBsAg level was the only factor in those with AFP
≥20 ng/mL [adjusted b 20.414 (20.712 to 20.168), P ¼ 0.002].

1604
AFP-dependent rapid HBsAg decline
Table 2. HBsAg reduction at month 6 of entecavir therapy in relation to pretherapy ALT and peak AFP levels in patients with hepatitis B flare
log10 IU/mL
ALT (ULN)
5 –10× (n¼52) 20.18 (22.8– 1.05)
≥10× (n ¼97) 20.80 (22.94 – 1.28)
AFPa (ng/mL)
,20 (n ¼85) 20.30 (21.82 – 1.28)
20– 99 (n¼25) 20.72 (22.51 – 1.27)
≥100 (n ¼39) 21.19 (22.94 – 0.61)
P 0.000
ALT (ULN)/AFPa (ng/mL) ,20
5 –10×b 20.18 (21.66 to +1.05)d
≥10×c 20.57 (21.82 to +1.28)g
P 0.012
Data expressed as median (range) log10 IU/mL or n (%).
a
Peak level during flare episode.
Linear trends of band care both P¼0.000.
d
versus f: P¼0.007.
e
versus f: P¼0.063.
g
versus i: P¼0.001.
Others: P.0.05.
Long-term outcome in patients with ‘rapid HBsAg decline’
Of the 60 HBeAg-positive patients in this study, 35 (68.6%) of
the 51 patients with hepatitis flare at baseline achieved HBeAg
seroconversion, much higher than 2 (22.2%) of the 9 with baseline
ALT ,5× ULN (P ¼ 0.021). Of these HBeAg-positive patients,
the 3 year cumulative HBeAg seroconversion rate was 83% in
patients with ‘rapid HBsAg decline’ and 45% in those without
rapid decline (P¼ 0.03).
HBsAg loss was documented in 3 (2%) of the 149 patients with
baseline hepatitis flare and 2 (3.4%) of the patients without flare,
at 1303 – 2015 (median 1590) days after the start of therapy
(P ¼ 0.616). The median time from baseline to HBsAg loss was
also similar between these two groups (with flare¼1590 days ver-
sus without flare¼1641 days). Of the five patients with HBsAg loss,
none had prior IFN therapy and one had liver cirrhosis at baseline.
Of the 149 patients with baseline hepatitis flare, HBsAg
seroclearance was observed in 3 (3.7%) of the 81 patients
with ‘rapid HBsAg decline’ and none of the patients without
(P ¼ 0.251). In addition, two of the three patients with HBsAg
loss had peak AFP .100 ng/mL. The average annual HBsAg reduc-
tion in patients with ‘rapid HBsAg decline’ was greater than in
those without it, in both HBeAg-positive and -negative patients
(20.439 versus 20.041 log10 IU/mL, P ¼ 0.000 and 20.379 versus
20.099 log10 IU/mL, P ¼0.000, respectively). Average reduction of
HBsAg from baseline to the end of treatment was greater in those
with ‘rapid HBsAg decline’ than in those without (median: 21.092
versus 20.183 log10 IU/mL, P ¼ 0.000).
Of the 53 HBeAg-negative patients with hepatitis flare at base-
line who had stopped entecavir therapy, 45 patients had stopped
therapy by the Asian Pacific Association for the Study of the Liver
JAC
HBsAg decline at month 6
.0.5 log10 decline
16 (30.8) 
P¼0.000
65 (67)

Downloaded from https://academic.oup.com/jac/article-abstract/71/6/1601/1751338 by guest on 18 August 2019
35 (41.2) 
P¼0.097
15 (60)  P¼0.000
P¼0.091
31 (79.5)
0.000
20– 99 ≥100
20.1 (21.32 to +0.14)e 21.34 (22.8 to 20.45)f
20.76 (22.51 to +1.27)h 21.19 (22.94 to +0.61)i
0.169 0.806
(APASL) stopping rule.12 The end-of-treatment HBsAg level in
patients with and without ‘rapid HBsAg decline’ was similar
(29.6 – 7515 versus 14 – 2086 IU/mL, P ¼ 0.716); however, more
achieved an HBsAg level ,100 IU/mL, though the difference
was non-significant (15% versus 8%, P¼ 0.654).
Discussion
The results of the present study showed that patients with hepatitis
B flare had a steep and significantly greater HBsAg reduction and
more frequent ‘rapid HBsAg decline’, in an ALT level-dependent
manner, and patients with ‘rapid HBsAg decline’ during the early
phase of entecavir therapy had a significantly higher HBeAg sero-
conversion rate. Previous studies showed: a ‘rapid HBsAg decline’ at
week 24 (≥0.5 log10 IU/mL decline) and year 1 (≥1.0 log10 IU/mL
decline) during telbivudine therapy in 33% and 20%, respectively,
of 162 HBeAg-positive patients with a mean pretherapy ALT of
4 – 5×ULN;14 a ‘rapid HBsAg decline’ at month 6 during lamivudine
therapy in 30% of 86 HBeAg-positive patients with a mean
pretherapy ALT of 4.58× ULN;15 and a ‘rapid HBsAg decline’ at
month 6 of Nuc therapy in 51% of 74 HBeAg-negative patients
with severe reactivation (ALT ≥10× ULN).7 The present study has
further shown that the proportion of patients with ‘rapid HBsAg
decline’ increases significantly along with the increasing ALT
level, being 6.9% in patients with baseline ALT ,5× ULN
(Table 1), 30.8% in those with ALT of 5 – 10× ULN, 63.5% in
those with 10– 20× ULN and 71.1% in those with ALT ≥20× ULN
at month 6. Although previous studies demonstrated that higher
pretherapy ALT level is a factor for greater HBsAg decline during
Nuc therapy,4 – 7,16,17 the data in the present study may be the
1605
Jeng et al.

(IU/mL) phase of entecavir therapy in both HBeAg-positive and -negative

(a) 1000 108 patients with pretherapy ALT ≥5× ULN. In particular, the differ-
AFP (ng/mL)
Therapy
ences in HBsAg reduction at month 6 of entecavir therapy across
800 all categories of ALT flare became non-significant in patients with
106
103 peak AFP ≥20 ng/mL (Tables 2 and 3). All these findings in HBsAg

HBV DNA
ALT (U/L)

600 reduction are consistent with the observation in an earlier study

104 that .70% of hepatitis B flare with AFP ≥100 ng/mL were fol-

HBsAg
102
400 lowed by HBeAg seroclearance, in contrast to ,20% in flares
with AFP ,100 ng/mL, including those with ALT .25× ULN.11,13
101 102 Another study also reported AFP elevation of 52 –1636 ng/mL in
200
seven patients with an ALT level .5× ULN (217 – 480 IU/L), of

Downloaded from https://academic.oup.com/jac/article-abstract/71/6/1601/1751338 by guest on 18 August 2019

whom HBV DNA fell in all patients and HBeAg was lost in two of
0 0
the four HBeAg-positive patients when ALT and AFP returned to
–16 –8 0 8 16 24 32 40 48 56
normal.18 To the best of our knowledge, the role of AFP level in
(b) 1000 108 HBsAg kinetics during Nuc therapy has never been explored.
Therapy
Since hepatitis flare with AFP .100 ng/mL is closely associated
800
106
with BHN,9 – 11 the influence of greater AFP in ‘rapid HBsAg decline’
AFP

103
HBV DNA is probably related to a greater extent of hepatocytolysis and sub-
HBsAg

600 sequent regeneration that may reduce and dilute intrahepatic

ALT

102 104 cccDNA.8 If AFP levels were also properly measured in patients
400 with baseline ≥5× ULN in the above-mentioned earlier studies,
perhaps we might explain why the early and ‘rapid HBsAg decline’
101 102 rates differ among different studies and being much higher in the
200
present study as compared with a lower rate in patients with ALT
0 0 .10× ULN in the Hong Kong study.7
0 8 16 24 32 40 48 56 60 There are variable findings on HBsAg reduction during Nuc
(c) 1000 108 therapy among patients infected with different HBV genotypes.
Therapy 104
Greater HBsAg decline was observed in patients with genotype
A than in patients with genotypes B, C and D HBV infec-
AFP

800
103 106 tion.5,14,19,20 Between genotypes B and C HBV-infected patients,
HBV DNA
HBsAg

600 greater HBsAg decline in the first year of entecavir therapy was
documented in patients with genotype B HBV infection,5,19,21,22
ALT

102
400
101
200
0 0
–8 0 8 16 24 32 40 48 56 60
Weeks of therapy
Figure 2. Clinical course during antiviral therapy in three representative
patients. (a) ‘Rapid HBsAg decline’ in a patient with high ALT and peak
AFP .100 ng/mL. (b) Slow HBsAg decline in a patient with high ALT, but
peak AFP ,20 ng/mL. (c) Slow HBsAg decline in a patient with ALT ,5×
ULN and AFP ,20 ng/mL.
first showing the influence of different ALT levels ≥5× ULN. Since
ALT elevation in chronic hepatitis B has been considered to be the
result of endogenous immune response against HBV and the ALT
levels reflect the strength of the immune response,10,11 the
above-mentioned results showing HBsAg reduction in an ALT
level-dependent manner are consistent with the concept that
greater HBsAg decline during entecavir therapy is linked to stron-
ger pre-existing immune activity.4,7 This is also in line with the
finding of association between greater HBsAg reduction and
stronger immune modulation reflected in higher IFN-g-inducible
protein of 10 kDa (IP-10).6,16
Perhaps the most important new finding of the present study
is that the peak AFP level is a more powerful baseline factor than
ALT level and HBV genotype for HBsAg decline during the early
104
but there was no difference in patients treated with tenofovir.20
The results of the present study showed significantly greater
102 HBsAg decline in genotype B HBV-infected patients with hepatitis
flare as a whole group, but the significant difference was actually
limited to those with ALT ≥20× ULN. In addition, the difference
became non-significant in patients with AFP ≥100 ng/mL during
hepatitis flare regardless of ALT levels, although genotype B
remained a significant independent factor for HBsAg reduction
at month 6 of therapy in the multivariate analysis. These findings
suggest that the impact of HBV genotype in HBsAg decline is over-
powered by the more extensive hepatocytolysis, as reflected in an
AFP level ≥100 ng/mL.9,10 Perhaps the different effects of HBV
genotype in patient populations with different ALT and/or AFP
levels may explain the different findings related to HBV genotype
reported in previous studies.5,14,19 – 21
The current study has several limitations. First, only a small
number of patients had liver biopsy and an even smaller number
of patients with AFP ≥20 ng/mL showed evidence of BHN (3 of 5)
and none of those with AFP ,20 ng/mL had BHN (0 of 12).
Therefore, the greater HBsAg decline could only be speculated
to be the result of more extensive hepatocytolysis based on higher
AFP levels. However, previous studies have provided strong
evidence that elevated AFP, especially ≥100 ng/mL, during hepa-
titis B flare is closely associated with biopsy findings of BHN9,10 and
AFP ≥100 ng/mL can be considered a surrogate marker of BHN.11
Second, the study aimed to examine the early HBsAg kinetics and
therefore did not include long-term HBsAg kinetics. However, earl-
ier studies have shown that HBsAg declines mainly during HBV

1606
AFP-dependent rapid HBsAg decline JAC
Table 3. Linear regression analysis of baseline factors in 149 patients with hepatitis B flare for HBsAg reduction at month 6 of entecavir therapy

Univariate Multivariate

Variable regression coefficient (b) (95% CI) P regression coefficient (b) (95% CI) P

Gender
female 1.0
male 20.005 (20.312– 0.295) 0.956

HBeAg positive 20.084 (20.41– 0.13) 0.308

Downloaded from https://academic.oup.com/jac/article-abstract/71/6/1601/1751338 by guest on 18 August 2019

Age 0.067 (20.006– 0.015) 0.417
Cirrhosis 0.06 (20.174– 0.379) 0.464

Genotype (n¼148)
B 1.0 1.0
C 0.177 (0.029–0.613) 0.031 0.164 (0.079– 0.517) 0.008

HBsAg (log10 IU/mL) 20.586 (20.74 to 20.467) 0.000 20.474 (20.642 to 20.335) 0.000
HBV DNA (log10 IU/mL) 20.418 (20.372 to 20.178) 0.000 20.127 (20.181– 0.015) 0.095

ALT
5 –10× 1.0 1.0
≥10× 20.276 (20.717 to 20.198) 0.001 20.123 (20.417– 0.009) 0.060
a
AFP
,20 1.0 1.0
20– 99 20.133 (20.607– 0.044) 0.090 0.037 (20.197– 0.354) 0.576
≥100 20.437 (21.062 to 20.509) 0.000 20.254 (20.705 to 20.215) 0.000

a
Peak level during flare episode.

DNA declining phase, which is mostly in the first year of Nuc ther-
apy. Third, the observation period was not long enough to show
differences in long-term efficacy. Nevertheless, the current
study showed that the HBeAg seroconversion rate was much
higher in patients with hepatitis flare than in those without flare
(68.6% versus 22.2%, P ¼ 0.021). For patients with ‘rapid HBsAg
decline’, the 3-year cumulative HBeAg seroconversion rate was
83%, significantly higher than 45% (P ¼ 0.03) in patients without
‘rapid HBsAg decline’. Their HBsAg loss rate was two times higher
(3.5% versus 1.7%; P ¼ 0.332) though the difference was not sig-
nificant statistically. Previous studies also suggest that ‘rapid
HBsAg decline’ may predict long-term outcomes of Nuc therapy
such as HBsAg seroclearance.7,14,16,20,23 Given a very low rate of
HBsAg seroclearance (,1%) in Asian patients (with genotype B
or C HBV infection) in previous studies with entecavir therapy over
3 – 5 years,24,25 longer follow-up is needed to see if these initial
dynamics translate into clinically meaningful events. Fourth, the
study patients were exclusively infected with genotype B or C HBV.
Perhaps similar findings are also applicable in patients with predom-
inant HBV genotypes other than B and C, especially genotypes A and
D. However, further studies are required to confirm this.
In conclusion, HBsAg kinetics during Nuc therapy is a complex
issue because it involves multiple factors. The small HBsAg
reduction of 0.09 log10 IU/mL at month 12 of entecavir therapy
in patients with baseline ALT ,5× ULN in the current study
(Table 1) implies that Nuc per se has limited impact on HBsAg
decline during the early phase of therapy. Notably, there was a
greater HBsAg reduction in an ALT and AFP level-dependent
manner during the early phase of entecavir therapy in patients
with pretherapy ALT ≥5× ULN. Further, AFP ≥100 ng/mL is a
more powerful factor that overwhelms ALT and genotype. This
AFP factor may reflect the effects of more extensive immune-
mediated hepatocyte killing as HBsAg decline was greatest in flares
with AFP ≥100 mg/mL, which is considered to be a surrogate
marker of BHN.11 Therefore, both ALT and AFP factors should be
taken into account when comparing HBsAg reduction between dif-
ferent studies or different patient populations with high ALT levels.
Measurement of AFP during hepatitis flare has long been neglected
since it was reported 30 years ago.9 Perhaps proper measurement
of AFP levels during hepatitis B flare deserves more attention in clin-
ical practice in the era of HBsAg quantification during natural course
and antiviral therapy.
Acknowledgements
We thank Ms Li-Hua Lu for laboratory work, Ms Shin-Huei Yang for
database construction, Ms Su-Chiung Chu for manuscript preparation
and Dr Hwai-I Yang for statistical consultation.
Funding
This study was supported by grants from the Chang Gung Medical
Research Fund (SMRPG1005, OMRPG380061 and CMRPG3A0901-3) and
the Prosperous Foundation, Taipei, Taiwan.
Transparency declarations
Y.-F. L. has been involved with clinical trials and served as a global advisory
board member of Roche. All other authors: none to declare.

1607
Jeng et al.
References
1 Chan HL, Thompson A, Martinot-Peignoux M et al. Hepatitis B surface
antigen quantification: why and how to use it in 2011 —a core group
report. J Hepatol 2011; 55: 1121–31.
2 Liaw YF. Clinical utility of hepatitis B surface antigen quantitation in
patients with chronic hepatitis B: a review. Hepatology 2011; 54: E1 –9.
3 Chan HL, Wong VW, Wong GL et al. A longitudinal study on the natural
history of serum hepatitis B surface antigen changes in chronic hepatitis B.
Hepatology 2010; 52: 1232– 41.
4 Reijnders JG, Rijckborst V, Sonneveld MJ et al. Kinetics of hepatitis B sur-
face antigen differ between treatment with peginterferon and entecavir.
J Hepatol 2011; 54: 449–54.
5 Zoulim F, Carosi G, Greenbloom S et al. Quantification of HBsAg
in nucleos(t)ide-naı̈ve patients treated for chronic hepatitis B with enteca-
vir with or without tenofovir in the BE-LOW study. J Hepatol 2015; 62:
56–63.
6 Papatheodoridis G, Goulis J, Manolakopoulos S et al. Changes of HBsAg
and interferon-inducible protein 10 serum levels in naive HBeAg-negative
chronic hepatitis B patients under 4-year entecavir therapy. J Hepatol
2014; 60: 62– 8.
7 Wong GL, Chan HL, Lo AO et al. Serum hepatitis B surface antigen
kinetics in severe reactivation of hepatitis B e antigen negative chronic
hepatitis B patients receiving nucleoside/nucleotide analogues. Antivir
Ther 2013; 18: 979–86.
8 Levrero M, Pollicino T, Petersen J et al. Control of cccDNA function in
hepatitis B virus infection. J Hepatol 2009; 51: 581–92.
9 Liaw YF, Tai DI, Chen TJ et al. a-Fetoprotein changes in the course of
chronic hepatitis: relation to bridging hepatic necrosis and hepatocellular
carcinoma. Liver 1986; 6: 133– 7.
10 Liaw YF, Tsai SL. Pathogenesis and clinical significance of spontaneous
exacerbation and remissions in chronic HBV infection. Viral Hepatitis Rev
1997; 3: 143–54.
11 Chang ML, Liaw YF. Hepatitis B flares in chronic hepatitis B: patho-
genesis, natural course and management. J Hepatol 2014; 61:
1407 – 17.
12 Liaw YF, Kao JH, Piratvisuth T et al. Asian-Pacific consensus statement
on the management of chronic hepatitis B: a 2012 update. Hepatol Int
2012; 6: 531–61.
13 Liaw YF, Chu CM, Huang MJ et al. Determinants for hepatitis B e antigen
clearance in chronic type B hepatitis. Liver 1984; 4: 301–6.
1608
14 Wursthorn K, Jung M, Riva A et al. Kinetics of hepatitis B surface antigen
decline during 3 years of telbivudine treatment in hepatitis B e antigen-
positive patients. Hepatology 2010; 52: 1611– 20.
15 Chen J, Wang Z, Zhou B et al. Factors associated with serum hepatitis B
surface antigen levels and its on-treatment changes in patients under
lamivudine therapy. Antivir Ther 2012; 17: 71 –9.
16 Jaroszewicz J, Ho H, Markova A et al. Hepatitis B surface antigen
(HBsAg) decrease and serum interferon-inducible protein-10 levels as
predictive markers for HBsAg loss during treatment with nucleoside/
nucleotide analogues. Antivir Ther 2011; 16: 915–24.
Downloaded from https://academic.oup.com/jac/article-abstract/71/6/1601/1751338 by guest on 18 August 2019
17 Zoutendijk R, Hansen BE, van Vuuren AJ et al. Serum HBsAg
decline during long-term potent nucleos(t)ide analogue therapy
for chronic hepatitis B and prediction of HBsAg loss. J Infect Dis
2011; 204: 415 – 8.
18 Lok AS, Lai CL. a-Fetoprotein monitoring in Chinese patients with
chronic hepatitis B virus infection: role in the early detection of hepatocel-
lular carcinoma. Hepatology 1989; 9: 110–5.
19 Gish RG, Chang TT, Lai CL et al. Quantitative hepatitis B surface antigen
analysis in hepatitis B e antigen-positive nucleoside-naive patients treated
with entecavir. Antivir Ther 2013; 18: 691– 8.
20 Marcellin P, Buti M, Krastev Z et al. Kinetics of hepatitis B surface anti-
gen loss in patients with HBeAg-positive chronic hepatitis B treated with
tenofovir disoproxil fumarate. J Hepatol 2014; 61: 1228– 37.
21 Chen EQ, Wang TT, Bai L et al. Quantitative hepatitis B surface antigen
titres in Chinese chronic hepatitis B patients over 4 years of entecavir treat-
ment. Antivir Ther 2013; 18: 955–65.
22 Seto WK, Lam YF, Fung J et al. Changes of HBsAg and HBV DNA levels in
Chinese chronic hepatitis B patients after 5 years of entecavir treatment.
J Gastroenterol Hepatol 2014; 29: 1028 –34.
23 Hosaka T, Suzuki F, Kobayashi M et al. Clearance of hepatitis B surface
antigen during long-term nucleot(s)ide analog treatment in chronic hepa-
titis B: results from a nine-year longitudinal study. J Gastroenterol
2013; 48: 930–41.
24 Yuen MF, Seto WK, Fung J et al. Three years of continuous entecavir
therapy in treatment-naı̈ve chronic hepatitis B patients: VIRAL suppres-
sion, viral resistance, and clinical safety. Am J Gastroenterol 2011; 106:
1264– 71.
25 Chen YC, Peng CY, Jeng WJ et al. Clinical outcomes after interrup-
tion of entecavir therapy in HBeAg-negative chronic hepatitis B
patients with compensated cirrhosis. Aliment Pharmacol Ther 2015; 42:
1182 – 91.