Trends in Incidence of Hepatocellular Carcinoma After Diagnosis of Hepatitis B or C Infection
A Population-based Cohort Study, 1992-2007
H.-H. Thein; S. R. Walter; H. F. Gidding; J. Amin; M. G. Law; J. George; G. J. Dore
Authors and Disclosures
Posted: 08/08/2011; J Viral Hepat. 2011;18(7):e232-e241. © 2011 Blackwell Publishing
Abstract
Chronic hepatitis B virus (HBV) and hepatitis C virus (HCV) infections are the major risk factors for hepatocellular carcinoma (HCC). We examined trends in the incidence of HCC among a population-based cohort of people infected with HBV or HCV. HBV and HCV cases notified to the New South Wales Health Department between 1992 and 2007 were linked to the Central Cancer Registry, Registry of Births, Deaths and Marriages, and National HIV/AIDS Registries. Crude HCC incidence rates were estimated using person-time methodology.
Age-standardized incidence rates were calculated using the 2001 Australian population. Trends in incidence were examined using join point regression models. Between 1992 and 2007, 1201 people had a linked HCC record: 556 of those with HBV; 592 with HCV; 45 with HBV/HCV co-infection; and 8 with HIV co-infection. The overall age-standardized HCC incidence rates declined non-significantly from 148.0 (95% confidence intervals (CI) 63.7, 287.4) per 100 000 population in 1995 to 101.2 (95% CI 67.3, 144.6) in 2007 among the HBV monoinfected group and significantly from 151.8 (95% CI 62.4, 299.8) per 100 000 population to 75.3 (95% CI 50.8, 105.5) among the HCV monoinfected group.
However, incidence rates in the HCV monoinfected group progressively increased from the period 1992–1997 to 2004–2007 when adjusted for age, sex, and birth cohort, and the total number of cases per annum continued to increase. Despite declines in the age-adjusted incidence rates of HCC over time, the absolute number of cases increased likely due to the ageing cohort and an increasing prevalence of both hepatitis B and C in Australia.
Introduction
Hepatocellular carcinoma (HCC) represents an increasing burden of disease worldwide. It is the fifth most common cancer globally and is the third most common cause of cancer-related mortality.[1,2] Liver injury caused by hepatitis B virus (HBV) and hepatitis C virus (HCV) infections, and alcohol is a major risk factor for HCC, with the vast majority of cases developing in the presence of underlying cirrhosis.[3] HBV/HCV co-infection[4–6] and co-infection with HIV further increases the risk of HBV and HCV disease progression,[7–9] HCC[10,11] and mortality.[7–9] Several studies using national cancer registries have reported an increase in HCC incidence over the past two decades,[12–18] related to increasing HBV and HCV prevalence and ageing cohort effects for both infections.[19] However, estimates of the incidence of HCC have varied widely, and most studies only reported data up to early 2000.
This study was therefore undertaken to examine trends in the incidence of HCC among hepatitis B and C infected individuals in NSW over the period 1992–2007, including the contribution of HIV co-infection
Methods
We conducted a population-based retrospective study, linking cases of HBV and HCV infection recorded in the NSW notifiable diseases database (NDD), to the HCC cases (defined by ICD-O-3 topography code C22.0 and histology codes 81703-81753) recorded in the NSW Central Cancer Registry (CCR), deaths recorded in the NSW Registry of Births, Deaths and Marriages (RBDM) and the Australian Bureau of Statistics (ABS), and HIV and AIDS cases recorded in the National HIV database (NHD) and National AIDS Registry (NAR), respectively, between 1 January 1992 and 31 December 2007.
Data Sources
Notification of a new diagnosis of HBV and HCV infection to the NSW NDD is mandated by law for the diagnosing laboratory and/or clinician (NSW Public Health Act 1991). HBV diagnosis requires detection of HBV surface antigen (HBsAg) or HBV DNA, while HCV diagnosis requires detection of anti-HCV antibody or HCV RNA.[18] Notification of incident cancer to NSW CCR has been mandatory for hospitals, nursing homes, the RBDM, radiotherapists and pathologists since 1972 (NSW Public Health Act 1991). Registration of all deaths occurring in NSW in the RBDM is required based on the Medical Certificate of Cause of Death. Lastly, notification of a new diagnosis of HIV infection to the health authority is mandated by law for the diagnosing laboratory and AIDS by clinicians.[20] HIV infections have been recorded in the NHD since 1985, and the NAR contains data on all AIDS diagnoses since 1982.
Linkage Procedure
Record linkage involved two steps: (i) probabilistically linking NDD notifications to CCR liver cancer records and RBDM death records using full name, date of birth, sex, and place of residence using ChoiceMaker software (ChoiceMaker Technologies, Inc., New York, NY, USA). RBDM death records were then linked to their corresponding ABS cause of death records using a deterministic linkage process and (ii) the data were then matched deterministically to NHD and the NAR notifications using name code, date of birth, and sex. Record linkage was performed by the NSW Centre for Health Record Linkage (CHeReL).[21] HBV/HCV, HBV/HIV, and HCV/HIV co-infected cases were identified by matching internally.
Statistical Analysis
Where multiple HBV or HCV records were matched to one individual, the earliest record was used to define the date of diagnosis. When defining HBV/HCV co-infection, the date of diagnosis was defined by the date of the second infection. Descriptive analysis of viral hepatitis notifications and HCC cases linked to HBV and HCV in each viral hepatitis group was performed.
For the analysis of HCC incidence, as the hepatitis notification might have been a result of the cancer diagnosis introducing a bias, the cohort was restricted to people notified with HBV or HCV infection in whom HCC had not been reported prior to, or within 6 months from the date of HBV or HCV notification and for whom NDD information about sex was available (0.7% missing). This resulted in exclusion of 384 HCC cases. Country of birth information was only available for 52% of HBV and HCV notifications, but was available for 97% of HCC cases. Therefore, country of birth was only included where analyses pertained to HCC cases only.
Crude incidence rates of HCC stratified by age group, sex, and year of HBV or HCV notification in NSW were determined using person-time methodology (age and year of notification treated as time dependent). Person-time for each person was defined as the number of years from 6 months after the date of NDD notification to the first of either date of diagnosis of HCC, date of death, or end of study period, 31 December 2007. Age-period-cohort analysis of HCC incidence was performed using Poisson regression models. The risk for HCC associated with each factor (age, period or cohort) was evaluated after adjusting for age, period, cohort, and sex, with age treated as a continuous covariate. Incidence rate ratios (IRR) were calculated for four four-year time periods (1992–1995, 1996–1999, 2000–2003, 2004–2007), sex, and five 10-year birth cohorts (<1930, 1930–1939, 1940–1949, 1950–1959, 1960+). A cohort analysis of incidence rates for the HBV and HCV monoinfection groups was performed with age-specific rates calculated by birth decade.
We compared our age-standardized incidence rates to those without removal of HCC cases diagnosed within 6 months of hepatitis notification. Additionally, HCV infection may remain undetected for many years, and thus the notified date of infection needs to be adjusted for the delay between infection and diagnosis. Therefore, we compared our results to those obtained following our previous linkage study adjustment for this discrepancy assuming median age of HCV infection to be 25 years.[24]
Figure 2. Number of cases of hepatocellular carcinoma, 1992–2007: (a) HBV monoinfection; (b) HCV monoinfection. All/Excl: before/after excluding people notified with HBV or HCV infection in whom HCC had been reported prior to, or within 6 months from the date of HBV or HCV notification. HBV, hepatitis B virus; HCV, hepatitis C virus; HCC, hepatocellular carcinoma.
After excluding people notified with HBV or HCV infection in whom HCC had been reported prior to, or within 6 months from the date of HBV or HCV notification, the study cohort for HCC incidence analysis consisted of 811 linked HCC records; 329 of those with HBV, 446 with HCV, 30 with HBV/HCV co-infection, and six with HIV co-infection (Table 1).
Trends in Hepatocellular Carcinoma Incidence in the Hepatitis B Virus Monoinfected Cohort
The overall crude HCC incidence rate for the HBV monoinfected cohort was 94.0 (95% CI 84.3, 104.7) per 100 000 person-years. Incidence rates were significantly different between each age group (Poisson test for difference: P < 0.05) except between the two oldest age groups (P = 0.19) (Table 2). Crude incidence was also significantly higher among men than women (145.0 vs 36.5 per 100 000 person-years, P < 0.001), but there was no significant difference between HBV notification periods.
The overall age-standardized HCC incidence rate among the HBV monoinfected cohort was 128.4 (95% CI 112.8, 145.3), and the rate for men was almost three times as high as women: 180.7 (95% CI 156.1, 207.7) per 100 000 population vs 59.2 (95% CI 43.4, 78.4) per 100 000 population (Fig. 3a). There was a non-significant reduction in age-standardized HCC incidence from 1995 to 2007 for the HBV monoinfected cohort (APC −3.3%, 95% CI −8.2, 1.8%) (Fig. 4a).
Figure 3.
Trends in age-standardized incidence rates of hepatocellular carcinoma per 100 000 population by sex and type of viral hepatitis, 1995–2007: (a) HBV monoinfection; (b) HCV monoinfection; (c) HBV/HCV co-infection. After excluding, people notified with HBV or HCV infection in whom HCC had been reported prior to, or within 6 months from the date of HBV or HCV notification. HBV, hepatitis B virus; HCV, hepatitis C virus; HCC, hepatocellular carcinoma.
Figure 4.
Trends in age-standardized hepatocellular carcinoma incidence rates per 100 000 population by type of viral hepatitis, 1995–2007: (a) Excluding and (b) Including people notified with HBV or HCV infection in whom HCC had been reported prior to, or within 6 months from the date of HBV or HCV notification. HBV, hepatitis B virus; HCV, hepatitis C virus; HCC, hepatocellular carcinoma.
Trends in Hepatocellular Carcinoma Incidence in the Hepatitis C Virus Monoinfected Cohort
The overall crude incidence of HCC for the HCV monoinfected cohort was 67.1 (95% CI 61.1, 73.6) per 100 000 person-years (Table 2). Similar to the HBV monoinfected cohort, incidence rates were significantly different between each age group (Poisson test for difference: P < 0.001) except between the two oldest age groups (Table 2). Crude incidence was also significantly higher among men than women (81.6 vs 43.0 per 100 000 person-years, P < 0.001), but there was no significant difference between HCV notification periods.
The overall age-standardized HCC incidence rate among the HCV monoinfected cohort was 132.7 (95% CI 118.8, 147.6), and the rate for men was almost twice as high as women: 166.6 (95% CI 145.3, 189.6) per 100 000 population vs 87.6 (95% CI 71.0, 106.8) per 100 000 population (Fig. 3b). In contrast to the HBV monoinfected cohort, there was a significant reduction in the overall age-standardized HCC incidence from 1995 to 2007 (APC −3.3%, 95% CI −6.3, −0.2%) (Fig. 4a). This decline was mainly because of men with incidence rates decreasing from 225.4 (95% CI 79.8, 483.7) per 100 000 population in 1995 to 102.2 (95% CI 64.0, 149.9) per 100 000 population in 2007, with a significant APC of −3.2% (95% CI −6.2, −0.2%).
Trends in Hepatocellular Carcinoma Incidence in the Co-infected Cohorts
The overall crude HCC incidence rates for the HBV/HCV, HBV/HIV, and HCV/HIV co-infected cohorts were 124.3 (95% CI 86.9, 177.7) (Table 2), 213.3 (95% CI 80.0, 568.3), and 46.9 (95% CI 11.7, 187.3) per 100 000 person-years, respectively. The respective age-standardized HCC incidence rates were 216.4 (95% CI 125.2, 336.2), 169.3 (95% CI 40.1, 446.2), and 60.0 (95% CI −0.9, 240.9) per 100 000 population. Among the HBV/HCV co-infected cohort, the rate for men was 1.8 times greater than women: 244.8 (95% CI 132.4, 394.9) per 100 000 population vs 134.7 (95% CI 22.4, 406.7) per 100 000 population (Fig. 3c). Small numbers of HCC diagnoses made it difficult to examine demographic patterns in crude or standardized rates among those co-infected with HIV.
Age-period-cohort Effect
In both HBV and HCV monoinfected groups, there was a significant age, period and cohort effect, as well as a significant effect of sex on HCC incidence rates when adjusted for other factors (Table 3). Incidence rates progressively increased from the period 1992–1995 to 2004–2007 among the HCV monoinfected group. A strong cohort effect was observed in both HBV and HCV groups, with lower incidence rates among younger birth cohorts for fixed age (Table 3, Figure S1a,b).
Sensitivity Analysis
When all cases of HCC were included in the analysis, there was a significant reduction in the overall age-standardized HCC incidence from 1995 to 2007 in both HBV monoinfected (APC −7.0%, 95% CI −9.7, −4.2%) and HCV monoinfected (APC −4.1%, 95% CI −7.3, −0.8%) cohorts, and age-standardized rates declined in earlier years more so than in later years, especially for the HBV monoinfected group (Fig. 4b). In the cohort, 64.4% of HCV monoinfected people and 86.2% of HBV/HCV co-infected people were more than 25 years old. When the median age of HCV infection was adjusted to 25 years, although the crude incidence rate of HCV monoinfected cohort (52.6, 95% CI 48.5, 57.0 per 100 000 person-years) was significantly lower than in the primary analysis (P < 0.001), the age-standardized rate (113.6, 95% CI 103.2, 124.6 per 100 000 population) was not (P = 0.09). The crude incidence rate (99.6, 95% CI 74.4, 133.4 per 100 000 person-years) and age-standardized incidence rate (192.7, 95% CI 125.9, 276.2 per 100 000 population) of HBV/HCV co-infected cohort were not significantly different from the primary analysis (P = 0.35 and P = 0.79, respectively).
Discussion
The overall burden of HBV and HCV-related HCC remains considerable in the era of improved antiviral therapy, although there are differences in trends between total incidence, age-standardized incidence and rates adjusted for age, sex and birth cohort and between HBV and HCV. Over the period (2003–2007) since our previous NSW linkage study,[18] the number of HBV-related HCC cases has stabilized at around 45–50 per year. In contrast, cases of HCV-related HCC continued to increase to 60–70 cases per year in 2005–2006, with a decline only seen in 2007. Incidence rates, adjusted for age, sex and birth cohort, progressively increased from the period 1992–1997 to 2004–2007, in particular among the HCV monoinfected group.
Over the period 1992–2007, we found that age-specific incidence rates increased with age in both HBV and HCV cohorts. It is well established that like most other tumours, HCC is an extremely age-dependent cancer.[25,26] Thus, the changing age structure of a population may strongly influence the incidence of cancer. Many studies have used a directly age-standardized method to account for this effect on sequential trends within a population and have shown an increase in the age-adjusted incidence of HCC in both high- and low-incidence regions.[12,16,18,27] In the primary analysis, our study found declines in the age-standardized HCC incidence rates from 1995 to 2007, significantly in the HCV monoinfected group. Although age-standardized incidence of HCC has declined, crude incidence increased because of increasing prevalence of both HBV and HCV in Australia,[28,29] predominantly driven by high-level immigration from HBV-endemic countries in the Asia-Pacific region[29] and high incidence of HCV among people who inject drugs,[28] respectively. Importantly, it should be noted that there is some evidence of a slower increase or a decline in the HCC incidence trends in some parts of the world.[30]
Nevertheless, it has been suggested that the directly age-standardized method does not fully account for the ageing of a population.[26] When adjusted for age, sex and birth cohort, we found a significant increase in the incidence of HCC from the period 1992–1997 to 2004–2007. This trend is consistent with a study by El-Serag et al.,[15] where the authors found a significant increase in the incidence of HCC from the period 1975–1977 to 1996–1998 after adjusting for age, race or ethnicity, sex and geographic region. Consistent with other studies,[12,15] we also found that the age-standardized HCC incidence rates for men were 2–3 times as high as women over the period.
Our study has several limitations, and the results should be interpreted cautiously. First, our study is based on diagnosed and subsequently mandatory notification of cases largely through laboratories and/or clinicians in Australia. However, an estimated 60% of people with hepatitis B[31] and 70% of people with hepatitis C[32] in Australia have been screened and notified. Since the introduction of mandatory notification in the early 1990s, around 110 000 and 260 000 cases of hepatitis B and hepatitis C, respectively, have been reported through public health surveillance mechanisms in Australia at the end of 2008.[33] NSW accounted for approximately 50% and 40% of the notifications, respectively. Thus, our findings are generally representative of the Australian-wide hepatitis-infected population. Second, while HBV notifications are based on evidence of chronic infection, HCV notifications are largely based on the presence of anti-HCV antibodies. An estimated 25% of infections spontaneously clear[34] and remain HCV antibody positive. Thus, a considerable minority of HCV notifications will not have chronic HCV infection. These limitations are likely to have produced an underestimate of HCC incidence among people with chronic HCV infection. Third, our study lacks information on country of birth for the vast majority of HBV and HCV notifications and was unable to adjust for this in the incidence analysis. Finally, the natural history of viral hepatitis may have been affected by antiviral therapy. Our study lacks information about hepatitis B or C treatment, and therefore we were not able to determine its specific impact on HCC incidence.
In summary, our results suggest that although age-adjusted incidence rates of HCC were steady or slightly declining over time, the population-level incidence increased. Prolonged latency to HCC development in the context of expanding population-level prevalence, particularly with ageing cohorts, is estimated to contribute to further increases in HCC over coming years, unless antiviral therapy uptake can be increased for both HBV and HCV. Improved antiviral therapy responses and removal of some barriers to Government-subsidized treatment access have led to some increase in treatment uptake, but levels remain low particularly for hepatitis C. Incidence trends within at-risk populations when linked to treatment information, surveillance system and vaccination programmes could be used to monitor the success of prevention and treatment for HCC.
Abstract and Introduction Methods
Results
Discussion
References
No comments:
Post a Comment