Académique Documents
Professionnel Documents
Culture Documents
Background & Aims: This study evaluates trends in hepatocellu- Conclusions: Individual-level risk of HBV-related HCC has
lar carcinoma (HCC) among people with hepatitis B virus (HBV) declined, suggesting an impact of more effective antiviral therapy
or hepatitis C virus (HCV) infection in New South Wales (NSW), from mid-2000s. In contrast, the interferon-containing HCV treat-
Australia between 2000 and 2014. ment era had no impact on individual-level HCV-related HCC risk
Methods: Data on HBV and HCV notifications between January and has seen escalating population-level HCC burden.
1993 and December 2012 were linked to the NSW Admitted Lay summary: Individual-level risk of HBV-related HCC has
Patients Data Collection database between July 2000 and June declined, suggesting an impact of more effective antiviral therapy
2014 and NSW Registry of Births Deaths and Marriages. The bur- from mid-2000s. In contrast, the interferon-containing HCV treat-
den, crude and age-standardised incidence of HCC based on first ment era had no impact on individual-level HCV-related HCC risk
hospitalization were calculated. and has seen escalating population-level HCC burden.
Results: In NSW between 20002014, there were 54,399, 93,099 2016 European Association for the Study of the Liver. Published
and 3,809 individuals notified with HBV, HCV and HBV/HCV coin- by Elsevier B.V. All rights reserved.
fection respectively. There were 725 (1.3%) with HCC among
those with HBV notification as compared to 1,309 with HCC
(1.4%) in those with HCV notification. The population-level bur-
den of new HCC cases per year has stabilised in the HBV cohort Introduction
(53 in 2001 and 44 in 2013), but increased markedly in the
HCV cohort (49 in 2001 to 151 in 2013). The age-standardised Hepatocellular carcinoma (HCC) is the third leading cause of can-
incidence rates of HCC (per 1,000 person-years) declined from cer death worldwide [14]. Chronic hepatitis B virus (HBV) and
2.3 (95% confidence interval (CI) 1.4, 3.1) in 2001 to 0.9 (95% CI hepatitis C virus (HCV) infections are the major causes of HCC
0.6, 1.2) in 2012 among those with HBV and remained stable [58], responsible for around 80% of cases [9]. The prognosis of
between 2001 (1.4; 95% CI 0.8, 1.9) and 2012 (1.5; 95% CI 1.2, HCC is extremely poor [3,10], with curative strategies such as
1.7) in those with HCV. Main factors associated with HCC in those liver resection and liver transplantation feasible only in a small
with HBV included later study period (20052009; 20102014) minority of cases [3,4,11,12].
(hazard ratio (HR) = 0.54, 95% CI 0.42, 0.70), male gender Increasing chronic viral hepatitis prevalence, ageing popula-
(HR = 4.50, 95% CI 3.6, 5.6), Asia-Pacific country of birth tions of those infected and sub-optimal treatment uptake and
(HR = 3.84, 95% CI 2.58, 5.71) and alcohol dependency outcomes (particularly for HCV) have led to escalating HCC bur-
(HR = 2.84, 95% CI 1.95, 4.13). Main factors associated with HCC den and mortality globally [2,10,12,13]. Improved antiviral ther-
in those with HCV included male gender (HR = 2.56, 95% CI apy for chronic HBV (from mid-2000s) and HCV (more recent
2.20, 2.98), rural place of residence (HR = 0.73, 95% CI 0.62, development of direct acting antiviral agents) may have impacted
0.86), Asia-Pacific country of birth (HR = 2.37, 95% CI 1.99, 2.82) HCC at the individual-level (age-adjusted risk) and population-
and alcohol dependency (HR = 3.90, 95% CI 3.39, 4.49). level (disease burden).
The mandatory notification of HBV and HCV diagnoses in Aus-
tralia and well-established surveillance systems provide the
opportunity to evaluate HCC burden at the individual and popu-
Keywords: Australia; Data linkage; Epidemiology; Hepatitis B virus; Hepatitis C lation levels, including temporal trends and factors associated
virus; Hepatocellular carcinoma.
Received 28 April 2016; received in revised form 23 June 2016; accepted 4 August
with HCC. The aims of this study were: 1) to assess trends in
2016; available online 26 August 2016 HCC incidence and burden among people with HBV or HCV infec-
Corresponding author. Address: Viral Hepatitis Clinical Research Program, The tion, and 2) identify factors associated with HBV- and HCV-
Kirby Institute for Infection and Immunity in Society, University of New South related HCC.
Wales, Australia. Tel.: +61 2 9385 9273.
E-mail address: rwaziry@kirby.unsw.edu.au (R. Waziry).
A case of HCC was defined by hospitalization with a HCC code (C22.0) as principal
or additional diagnosis. Hospital admissions were obtained from the NSW Admit- Ethics approval
ted Patient Data Collection (APDC), which includes in-patient hospitalizations
from all public and private hospitals in NSW between 1 July 2000 and 30 June Ethics approval for the study was granted by the NSW Population and Health Ser-
2014. Data on each hospitalization is recorded at separation and includes demo- vices Research Ethics Committee.
graphic and administrative data, the principle and any additional diagnoses coded
according to the 10th revision of the International Classification of Diseases-
Australian Modification (ICD-10-AM). The validity of ascertainment of HCC cases
through hospitalization-based codes was assessed by linkage to the NSW Cancer Results
Council Registry (CCR), with data available for the period between 1 Jan 1994 to
31 Dec 2009.
Participant characteristics
Other data sources and definitions
A total of 151,307 individuals were notified with HBV or HCV
infections in NSW between 1993 and 2012, including 54,399 with
Data on HIV was obtained from the National HIV Registry (NHR) that includes all
individuals notified with HIV between 1 January 1993 and 31 December 2013. HBV, 93,099 with HCV and 3809 with HBV/HCV coinfection. Par-
Data on deaths among those with HBV or HCV notification was obtained from ticipant characteristics are shown in Table 1. Compared to people
the NSW Registry for Births, Deaths and Marriages (RBDM) between 1 January with HCV, people notified with HBV were younger (median year
1993 and 18 June 2014. HCV and HBV mono-infections were defined according of birth = 1966 vs. 1965), more often female (45% vs. 37%), and
to the earliest notification record available. HBV/HCV co-infections were defined
according to date of notification of the latest infection and included in the HCV
more often born in Asia (30% vs. 5%) (Table 1).
cohort. Alcohol dependency was defined according to hospitalization with any
of the following principal or additional ICD-10 codes: alcohol abuse counselling
Burden and incidence of HCC in people with HBV and HCV
and surveillance (Z71.4), alcoholic cardiomyopathy (I42.6), alcohol-induced
pseudo-Cushings syndrome (E24.4), alcoholic myopathy (G72.1), alcoholic notification in NSW 20002014
polyneuropathy (G62.1), alcohol rehabilitation (Z50.2), degeneration of nervous
system due to alcohol (G31.2) or mental and behavioural disorders due to alcohol Between 2000 and 2014, the total number of people with HCC
(F10). For HCC and alcohol dependency diagnoses, the first hospitalization as
was 1309 (1.4%) among those with an HCV notification and 725
principle or additional diagnosis was used.
(1.3%) among those with an HBV notification. Among people with
HBV, HCC cases per year remained relatively stable between 2001
Linkage process
(n = 53) and 2013 (n = 44). In comparison, among people with
HCV, HCC cases per year increased from 49 in 2001, to 151 in
Data linkage was completed in two stages. First, HBV and HCV notifications were
linked internally to identify people with HBV/HCV coinfection. All notifications 2013 (Fig. 1A, B).
were then matched to other datasets (hospitalizations, cancer registry, death reg- Among people with HBV, age-standardised HCC incidence
istry) using probabilistic record linkage methods based on matching demographic decreased from 2.3 (95% confidence interval (CI) 1.4, 3.1) per
data, using ChoiceMaker software [15,16]. The second stage involved HBV and
1000 person-years in 2001 to 0.9 (95% CI 0.6, 1.2) per 1000
HCV notifications linkage to the HIV notifications using deterministic methods
based on 100% match on name code, sex and date of birth. Data linkage was com-
person-years in 2012. Among people with HCV, age-
pleted by the Centre for Health Record Linkage (CHeReL) [15,16]. standardised HCC incidence remained relatively constant at 1.4
(95% CI 0.8, 1.9) per 1000 person-years in 2001 and 1.5 (95% CI
Statistical analysis 1.2, 1.7) in 2012 (Fig. 1C, D).
A B C HBV-HCC age
D HCV-HCC age
Age standardised incidence rates
HBV-HCC burden HCV-HCC burden standardised incidence rates standardised incidence rates
(per 1000 person-years)
200 4 4
First hospitalizations, n
200
150 150 3 3
100 100 2 2
50 50 1 1
0 0 0 0
01 0 3 05 07 0 9 11 13 01 0 0 3 0 0 5 007 0 0 9 0 1 1 013
01
03
05
07
09
11
13
01
03
05
07
09
11
13
20 2 0 20 20 2 0 20 20 20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
2 2 2 2 2 2
Fig. 1. First hospitalization for HCC. Data shown for patients with (A) HBV notification, (B) HCV notification and age-standardized incidence rates for first HCC
hospitalization among those with (C) HBV notification, and (D) HCV notification.
Person-years HCC (n = 552) Rate (/1000 person-years) HR (95% CI) Adjusted HR (95% CI) p value
Study period
2000-2004 126,501 134 1.1 1.00 1.00 --
2005-2009 196,827 195 0.9 0.94 (0.74, 1.18) 0.75 (0.59, 0.94) 0.01
2010-2014 219,230 223 1.0 0.99 (0.78, 1.26) 0.54 (0.42, 0.70) <0.001
Age -- -- -- 1.06 (1.06, 1.07) 1.06 (1.05, 1.06) <0.001
Gender
Female 248,993 94 0.4 1.00 1.00
Male 290,337 458 1.6 4.17 (3.34, 5.21) 4.50 (3.6, 5.6) <0.001
Place of residence
Metro 243,626 257 1.1 1.00 1.00 --
Outer-metro 247,880 263 1.1 1.01 (0.84, 1.19) 0.94 (0.79, 1.13) 0.54
Rural 46,993 32 0.6 0.64 (0.45, 0.93) 0.85 (0.57, 1.27) 0.44
Country of birth
Australia 42,674 31 0.7 1.00 1.00 --
Asia Pacific 159,955 400 2.5 3.44 (2.38, 4.96) 3.84 (2.58, 5.71) <0.001
Europe 15,246 52 3.4 4.69 (3.01, 7.32) 2.14 (1.34, 3.39) <0.01
Other 7724 12 1.6 2.13 (1.09, 4.15) 2.91 (1.48, 5.74) <0.01
Not specified 3737 21 5.6 7.75 (4.45, 13.48) 7.02 (3.96, 12.43) <0.001
HIV
No 539,652 545 1.0 1.00 1.00 --
Yes 2907 7 2.4 2.38 (1.13, 5.02) 2.25 (0.99, 5.10) 0.05
Alcohol-related hospitalization
No 533,353 514 0.9 1.00 1.00 --
Yes 9206 38 4.1 4.27 (3.07, 5.94) 2.84 (1.95, 4.13) <0.001
y
Includes people co-infected with HBV/HIV and excluding HCC hospitalizations within 6 months after HBV notification; HCC, Hepatocellular carcinoma; The overall p
value for place of residence = 0.06, country of birth <0.001 and study period <0.001; Age is calculated based on 5-year interval.
treatment uptake could lead to pronounced declines in HBV-HCC tralia, and the particularly enhanced efficacy and tolerability of
cases per year, given that 43% of people living with chronic HBV these regimens in the setting of advanced liver disease [3134]
in Australia are estimated to remain undiagnosed [19,20]. Thus, provides considerable optimism for a major impact on HCV-
there is potential to further reduce the individual-level HCC risk HCC burden.
for the estimated 183,000 people with chronic HBV in Australia Although later study period (20052009, 20102014) was
and to reduce the population-level burden [20]. associated with reduced risk of HBV-HCC, other factors associated
The individual-level risk of HCC among people with HCV in with HCC were similar among those with HBV and HCV. These
NSW has remained stable over the period 2001 to 2013, while included older age, male gender, Asia-Pacific and other non-
the population-level HCV-HCC burden continues to escalate. This Australian countries of birth, and hospitalization for alcohol-
is despite the improvements in IFN-based therapy during the related disorder (surrogate for alcohol dependency). Older age,
2000s and early 2010s, including the addition of first generation male gender and alcohol dependency are all associated with
protease inhibitors [23,24]. However, HCV treatment rates in accelerated fibrosis progression or more advanced fibrosis, and
Australia have remained low and IFN-based therapy has have been clearly associated with HCC development in prior
remained particularly problematic for many people with studies [9,10,35,36]. It is unclear why non-Australian country of
advanced liver disease [5,25,26]. Thus, the apparent lack of an birth is associated with development of both HBV-HCC and
impact of HCV antiviral therapy on individual-level HCC risk HCV-HCC. Recent evidence suggests that Asian ethnicity may be
relates to sub-optimal treatment outcomes, particularly in those associated with fibrosis progression through a higher prevalence
with cirrhosis [27,28] and low HCV treatment uptake. Although of IFNk3 gene polymorphisms [37,38]. Among people with HCV-
the estimated number of people receiving treatment increased related cirrhosis, reported rates of HCC development are consis-
from about 1,100 in 1997 to 2,800 in 2014 [29], the latter fig. tently higher in Asian settings, particularly Japan [36]. It is also
remains less than 2% of the estimated 230,000 people with unclear why risk of HCV-HCC is reduced in rural areas. Differen-
chronic HCV infection [30]. tial HCC screening or hospitalization coding in different areas
The recent development of highly effective IFN-free DAA ther- could influence HCC diagnosis, however, the latter seems unlikely
apy, and Australian Government subsidization of several regi- given the high concordance with cancer registry.
mens from March 2016 (without disease stage or drug and The apparent chemoprevention of HCC through improved
alcohol restrictions) provides the opportunity to rapidly scale- HBV antiviral therapy has important implications for public
up HCV antiviral therapy and provide individual-level and health policy and clinical management. The specific impact needs
population-level benefits in terms of HCC and other liver disease to be further evaluated through linkage to individual-level HBV
burden reductions. The relatively high HCV diagnosis rate in Aus- treatment data. Such linkage should be possible in NSW in the
Person-years HCC (n = 1168) Rate (/1000 person-years) HR (95% CI) Adjusted HR (95% CI) p value
Study period
2000-2004 249,848 184 0.7 1.00 1.00 --
2005-2009 360,922 350 0.9 1.29 (1.07, 1.56) 0.86 (0.71, 1.04) 0.12
2010-2014 373,389 634 1.6 2.20 (1.83, 2.65) 0.85 (0.69, 1.03) 0.10
Age -- -- -- 1.07 (1.06, 1.07) 1.1 (1.08, 1.09) <0.001
Gender
Female 372,446 231 0.6 1.00 1.00 --
Male 608,106 937 1.5 2.49 (2.15, 2.88) 2.56 (2.20, 2.98) <0.001
Place of residence
Metro 323,233 440 1.4 1.00 1.00 --
Outer-metro 323,048 445 1.4 1.04 (0.91, 1.19) 0.91 (0.79, 1.05) 0.20
Rural 327,573 280 0.8 0.64 (0.55, 0.74) 0.73 (0.62, 0.86) <0.001
Country of birth
Australia 503,627 569 1.1 1.00 1.00 --
Asia Pacific 49,684 236 4.7 4.30 (3.69, 5.1) 2.37 (1.99, 2.82) <0.001
Europe 46,998 188 4.0 3.49 (2.96, 4.13) 1.59 (1.33, 1.90) <0.001
Other 16,174 60 3.7 3.38 (2.59, 4.42) 2.00 (1.52, 2.64) <0.001
Not specified 16,307 12 0.7 0.59 (0.34, 1.06) 0.64 (0.35, 1.13) 0.12
HBV
No 948,235 1104 1.2 1.00 1.00 --
Yes 35,925 64 1.8 1.58 (1.22, 2.03) 1.3 (0.97, 1.65) 0.07
HIV
No 975,577 1161 1.2 1.00 1.00 --
Yes 8584 7 0.8 0.71 (0.34, 1.49) 0.76 (0.36, 1.62) 0.48
Alcohol-related hospitalization
No 845,590 779 0.9 1.00 1.00 --
Yes 138,570 389 2.8 3.12 (2.76, 3.52) 3.90 (3.39, 4.49) <0.001
y
Includes HCV/HBV co-infections and excluding HCC hospitalizations within 6 months after HCV notification; HCC, Hepatocellular carcinoma; The overall p value for place
of residence <0.001, country of birth <0.001 and study period = 0.1; Age is calculated based on 5-year interval.
near future, following procedures to enable access to of hospitalization coding data for diagnosis of HCC could intro-
government-subsidized treatment (vast majority of treatment duce misclassification (over-diagnosis) or missed cases (under-
in Australia) datasets. These procedures will also enable evalua- diagnosis). Some patients with HCC, particularly those with early
tion of individual-level HCV treatment impact, following the stage disease, do not require hospitalization for HCC manage-
anticipated rapid scale-up of DAA therapy. ment. However, the availability of data on HCC through the Cen-
Our findings are likely relevant to other settings. In Taiwan, tral Cancer Registry allowed for validation of HCC cases captured
enhanced therapeutic intervention through the national hepatitis through hospitalization data for most of the study period (2000
treatment program (for HBV and HCV) appears to have reduced 2009). Around 90% of HCC cases were recorded on both datasets,
HCC incidence from the early 2000s to recent years, although providing confidence that hospitalization-based HCC burden and
the separate impact of HBV and HCV was not analysed [35]. trends are valid measures at the population-level. Third, symp-
The low HCV treatment uptake in the IFN-containing era is con- tomatic presentation of HCC could inflate incidence rates, partic-
sistent with most settings [5,3941], and the limited ularly among people with HBV in whom a considerable
population-level impact of therapy on HCV-HCC incidence has proportion remain undiagnosed. This is less an issue for HCV,
been previously described. Among a population of US veterans, where an estimated 75% are diagnosed [20,29]. Further, cases of
IFN-containing HCV therapy had a negligible impact on the inci- HCC within six months of HBV or HCV diagnosis were excluded
dence of HCC related to HCV, even though the majority of HCV for the analyses of incidence. Finally, the use of alcohol hospital-
patients have been diagnosed (80%) and about 23% have izations likely underestimates the real burden of alcohol depen-
received treatment [5,42]. dency in this cohort. However, the use of ICD diagnostic
This study has several limitations. First, HCV diagnosis for definitions from hospital databases has shown great validity with
surveillance reporting is generally based on anti-HCV antibody physicians diagnoses and has been extensively used in epidemi-
detection and does not require HCV RNA confirmation. Thus, an ological studies [5,12,4347].
estimated 25% of HCV notifications would have undergone spon- In conclusion, this population-level data linkage study in
taneous HCV clearance. The lack of confirmation of chronic HCV NSW, Australia provides evidence for declining risk of HBV-
infection should, however, not have a major impact on the study HCC, a likely result of improving HBV antiviral therapy. The
findings, given the surveillance definition and systems in NSW surveillance system we have developed, particularly with the
have been stable throughout the study period. Second, the use addition of individual-level HBV and HCV treatment data will
02
03
04
05
06
07
08
09
20
20
20
20
20
20
20
20
20
Conflict of interest
Year of HCC diagnosis
B 100
HCV Jason Grebely has received research support and is a consultant
for Gilead Sciences and Merck. Jason Grebely has received
HCC diagnosis, proportion
80
research support from Bristol-Myers Squibb and Abbvie. Jacob
George is on the speakers bureau for Gilead Sciences, Merck,
60 Janssen, Roche and Pharmaxis. Jacob George is a member of advi-
sory board for Gilead Sciences, Merck, Janssen, Bristol-Myers
40 Squibb, Abbvie, Roche, GlaxoSmithKline and Pharmaxis. Jacob
George has received travel support from Gilead Sciences, Merck,
20 Bristol-Myers Squibb, Abbvie and Roche. Gail V Matthews has
received advisory board payments from Gilead, speaker fees
0 and honoraria from Gilead, Bristol-Myers Squibb and Abbvie
01
02
03
04
05
06
07
08
09
and research grant funding from Janssen, Gilead and Abbvie. Mat-
20
20
20
20
20
20
20
20
20
Year of HCC diagnosis thew G Law has received research support and is a consultant for
Admitted Patient Data Collection only Central Cancer Registry only Both
Gilead Sciences. Matthew G Law has received research support
from Merck, Bristol-Myers Squibb, Boehringer Ingelheim,
Fig. 2. Agreement between the Central Cancer Registry and the Admitted Janssen-Cilag and ViiV HealthCare. Matthew G Law has received
Patients Data Collection. Data shown for patients with (A) HBV and (B) HCV Data Safety Monitoring Board Committee fees from Sirtex Pty
notification.
Ltd. Gregory J Dore has received research support and is a consul-
tant for Gilead Sciences, Merck and Janssen. Gregory J Dore has
received research support from Bristol-Myers Squibb, Abbvie
be a valuable tool for evaluation of enhanced strategies for viral and Roche. Gregory J Dore is on the speakers bureau for Gilead
hepatitis diagnosis and treatment in the future. It will also pro- Sciences, Merck, Janssen and Roche. Gregory J Dore is a member
vide the opportunity to compare population and individual- of advisory board for Gilead Sciences, Merck, Janssen, Bristol-
level impacts in different settings, particularly those with manda- Myers Squibb, Abbvie, Roche, GlaxoSmithKline and Abbott Diag-
tory HBV and HCV notification and capacity for data linkage to a nostics. Gregory J Dore has received travel support from Gilead
range of administrative datasets. Sciences, Merck, Bristol-Myers Squibb, Abbvie and Roche. Other
authors have no commercial relationships that might pose a con-
flict of interest in connection with this manuscript.
Financial support