Incidence of hepatocellular carcinoma among older Americans attributable to hepatitis C and hepatitis B: 2001 through 2013

In the United States, incidence and mortality rates of hepatocellular carcinoma (HCC) are increasing in older individuals. Chronic infection with hepatitis C virus (HCV) and hepatitis B virus (HBV) are important causes of HCC; however, the contribution of viral hepatitis to recent trends in HCC incidence among older Americans is unclear.


INTRODUCTION
In the United States, liver cancer incidence and mortality rates increased substantially during 2003 through 2012. 1 Nearly 75% of all liver cancer diagnoses are hepatocellular carcinomas (HCCs). 2 During 2000 through 2010, temporal trends in HCC rates diverged by age group, with stable rates noted among individuals aged 35 to 49 years, but strong increases observed among those aged 50 to 64 years and ≥65 years. 2 If current trends continue, it has been projected that HCC rates among individuals aged ≥65 years in the United States will increase 5.9% per year through 2030. 3 Chronic infection with hepatitis C virus (HCV) is an important driver of these rising liver cancer rates. 1,2 HCV is associated with a 60-fold increase in HCC risk, 4 and caused an estimated 31% of US liver cancer cases in 2015. 5 Although available data do not support a precise determination of HCV prevalence in the United States, a recent estimate is that 2.5 to 4.7 million US residents have chronic HCV. 6 Because HCV infection rates were high in the 1960s through the 1980s, 7 and HCC risk rises with an increasing duration of HCV infection, rates of liver cancer due to HCV are rising over time. 1,3,8 Chronic infection with hepatitis B virus (HBV) is another important cause of liver cancer worldwide, increasing the risk 20-fold and causing an estimated 9% of cases in the United States. 4,5 In 2012, an estimated 850,000 individuals were living with chronic HBV in the United States, with the highest prevalence reported among foreign-born Asian individuals. 9 In addition to HCV and HBV, diabetes, obesity, metabolic syndrome, and excessive alcohol consumption also are associated with liver cancer, and also may have contributed to rising rates over time. 10 Cancer August 1, 2019 In the current study, we assessed HCC rates with data from the Surveillance, Epidemiology, and End Results (SEER)-Medicare database 11 among Americans aged ≥66 years, an age range that included 44% of US HCC cases in 2015. 12 We assessed trends in HCC rates, estimated HCV and HBV prevalence among HCC cases, and disaggregated HCC time trends by viral hepatitis status, sex, age group, and race/ethnicity during 2001 through 2013.

Study Population
The current analysis was performed using data from the SEER-Medicare data set, a linkage of 2 populationbased data sources: the SEER cancer registries and a database of US Medicare enrollees. 11 In the United States, Medicare is the federally funded health insurance program that primarily covers individuals aged ≥65 years. SEER is a set of cancer registries funded by the National Cancer Institute, and includes information regarding tumor characteristics, patient demographics, and survival. The SEER-Medicare database includes Medicare claims during 1991 through 2014 for all cancer cases diagnosed between 1991 and 2013 as well as a 5% random sample of all Medicare enrollees from SEER areas.

HCC Cases
Invasive HCC cases that were diagnosed as a first primary cancer during 2001 through 2013 were identified by cancer registries and defined based on International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3) site (C220) and histology (8170-8175) codes. The analysis was restricted to cases diagnosed among individuals aged 66 to 99 years at the time of diagnosis who had ≥12 months of Medicare coverage, including both Part A (inpatient) and Part B (health care provider) coverage outside of a health maintenance organization, prior to cancer diagnosis and at least 1 Medicare claim.
We ascertained infection with HCV or HBV from Medicare files for the period from 60 months before until 12 months after cancer diagnosis based on these International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes: HCV (including codes for acute HCV given the high probability of progression to chronic HCV 13

Five Percent Random Population Sample
A 5% random sample of all Medicare recipients (those with and without a cancer diagnosis) was ascertained for each calendar year, restricted to those enrollees who on July 1 of that year were aged 66 to 99 years; had ≥12 months of Medicare coverage, including both Part A and Part B coverage outside of a health maintenance organization; and had at least 1 claim.

Statistical Analysis
The percentages of HCC cases that were HCV-attributable, HBV-attributable, and both HCV-attributable and HBVattributable were assessed by calendar period of HCC diagnosis (2001-2004, 2005-2007, 2008-2010, and 2011-2013), sex, age at diagnosis (66-75 years, 76-85 years, and 86-99 years), and race/ethnicity (non-Hispanic white [ie, white], non-Hispanic black [ie, black], Asian, Hispanic, or other/unknown). We considered cases with evidence of HBV or HCV infection as attributable to those infections based on prior analyses demonstrating that the prevalence of these viruses among HCC cases is nearly identical to the percentage of HCC cases caused by these viruses. 4,14 HCC incidence rates were estimated by dividing the number of HCC diagnoses by person-time from the 5% random sample, which was multiplied by 20 to represent the full Medicare population. Age-standardized rates were based on the 2000 US population age distribution. Temporal trends were assessed separately for HCV-attributable, HBV-attributable, and HCV/HBVunrelated HCC cases, and stratified by sex, age group, and race/ethnicity during 2001 through 2013. Joinpoint regression was used to identify statistically significant changes in the slope of time trends in incidence rates (based on single-year data) and to estimate annual Cancer August 1, 2019 percent changes (APCs) in rates in each group. 15 For display purposes, we aggregated rates across 2-year or 3-year periods in figures.
To assess the extent to which SEER-Medicare rates represent rates from the full SEER population, HCC rates overall were compared with those for the full SEER population (individuals aged ≥66 years) of included registries (ie, not restricted to eligible Medicare recipients). 12 Furthermore, trends by HCV and HBV status were estimated requiring only one physician or outpatient claim for a diagnosis of HCV or HBV infection.

RESULTS
Between 2001 and 2013, a total of 15,300 cases of invasive HCC were diagnosed among eligible Medicare recipients (Table 1). Patients with HCC were more likely to be male than female, although a larger percentage of HCV-attributable cases occurred among women (39%) compared with HBV-attributable cases (29%) and HCV/HBV-unrelated cases (31%). The age distribution of HCV-attributable and HBV-attributable HCC cases was similar; in contrast, a smaller percentage of patients with HCV/HBV-unrelated HCC were aged 66 to 75 years, and a larger percentage of patients were aged ≥86 years. The largest percentage of HCV-attributable (54%) and HCV/HBV-unrelated (76%) HCC cases occurred among white individuals, whereas the largest percentage of HBV-attributable HCC cases occurred among Asian individuals (53%). Claims for diabetes were most common among HCV/HBV-unrelated HCC cases (59%), and claims for alcohol-related liver disorders were most common among HCV-attributable cases (31%). HCCrelated genetic disorders (2.7%) were uncommon across case groups.
The overall HCV prevalence among HCC cases was 29.5% (4518 cases) and that of HBV was 8.2% (1253 cases): 542 of these cases (3.5%) occurred among individuals who were infected with both HCV and HBV. HCV prevalence in HCC cases increased from 25.7% in 2001 through 2004 to 32.3% in 2011 through 2013, whereas HBV prevalence remained stable at 8% (Fig. 1A). These prevalence estimates are not mutually exclusive: individuals infected with both HBV and HCV were included in the prevalence estimates for both viruses. The prevalence of coinfection was 3.9% in 2001 through 2004 and 3.5% in 2011 through 2013. HCV prevalence among patients    (Fig. 2) (Table 2). Although the annual increase in incidence rates was most rapid for HCV-attributable HCC (APC, 5.62% per year), rates of HBV-attributable HCC (APC, 3.17% per year) and HCV/HBV-unrelated HCC (APC, 2.35% per year) also increased significantly over time. The absolute increase in age-standardized HCC rates between 2001 and 2013 was largest for HCV-attributable cases (4.01 per 100,000 population) and HCV/HBV-unrelated cases (2.80 per 100,000 population), and smaller for HBV-related cases (0.46 per 100,000 population).
HCC rates increased 3.16% per year among men and 3.14% per year among women; however, HCC rates were much higher among men compared with women in 2013 (36.6 vs 13.4 per 100,000 population in 2013) ( Table 2) (Fig. 2). HCC rates increased across age groups (ages 66-75 years: 3.20% per year; ages 76-85 years: 3.86% per year; and age ≥86 years: 2.76% per year), and were highest among individuals aged 66 to 75 years. Rates of HCV/HBV-unrelated HCC increased significantly in each age group. HCV-attributable HCC rates increased significantly among those aged 66 to 75 years and 76 to 85 years, and HBV-attributable HCC increased significantly over time among individuals aged 76 to 85 years, although few cases occurred in individuals aged ≥86 years.
Age-standardized HCC rates varied substantially by race/ethnicity: in 2013, rates were nearly 3 times higher among Asian individuals (54.8 per 100,000 population) than among white individuals (19.7 per 100,000 population) (Table 2) (Fig. 2). Overall HCC rates increased significantly over time among white (APC, 3.69% per year), black (APC, 4.21% per year), and Hispanic (APC, 3.62% per year) individuals, but not among Asian individuals or those of other or unknown race/ ethnicities. In 2013, Asians also had the highest rates of HCV-attributable (23.7 per 100,000 population) and HBV-attributable (19.5 per 100,000 population) HCC, whereas rates of HCV/HBV-unrelated HCC were highest among Hispanic individuals (24.9 per 100,000 population). HCV-attributable HCC rates increased significantly among white, black, and Hispanic individuals, whereas HBV-attributable HCC rates and HCV/HBVunrelated HCC rates increased significantly only among white individuals. Rates of HCV-attributable and HBVattributable HCC were stable among Asian individuals.
In the full SEER population, approximately 41% of all patients with HCC were aged ≥66 years during 2001 through 2013 (see Supporting Fig. 1). Comparing HCC rates in the SEER-Medicare database with those in the full SEER population, the SEER-Medicare rates were 10% to 14% lower, although the APCs were consistent (SEER-Medicare: 3.40% per year; and full SEER population: 3.29% per year) (see Supporting Fig. 2).
In a sensitivity analysis requiring only 1 inpatient, physician, or outpatient claim for a diagnosis of HCV or HBV, the HCV prevalence was 35.6% (5440 claims) and the HBV prevalence was 11.4% (1744 claims). Increases in prevalence were driven largely by an increase in HCV/HBV-coinfected HCC cases (1102 cases). In addition, APCs were consistent with those estimated in the primary analysis (HCV-attributable: 5.54% per year; HBV-attributable: 3.48% per year; and HCV/HBVunrelated: 2.15% per year).

DISCUSSION
During 2001 through 2013, overall HCC rates among Americans of Medicare age increased by 3.4% per year, with the percentage of HCC cases attributable to HCV infection increasing from 26% to 32%. The rates of HCV-attributable HCC increased strongly overall (5.6% per year) and in the majority of demographic subgroups, contributing substantially to the increase in total HCC rates over time. Although rates of HBV-attributable HCC increased significantly over time (3.2% per year), absolute rates remained low. Rates of HCC unrelated to viral hepatitis also increased over time (2.4% per year), indicating that HCC driven by other causes unrelated to viral hepatitis also has contributed substantially to increasing overall trends.
Chronic infection with either HCV or HBV can lead to inflammation of the liver with resulting development of fibrosis and cirrhosis in some infected individuals. 16 This process results in liver cells that are abnormal and at substantial risk of progressing to cancer. 8 From 2011 through 2013, we estimated that approximately  32% of HCC cases were HCV-attributable and 8% were HBV-attributable, similar to estimates published by the Global Burden of Disease group for the United States in 2015 (31% and 9%, respectively). 5 In the United States, HCV incidence was high in the 1960s through 1980s, most likely spread through injection drug use and the receipt of infected blood products prior to HCV screening of the blood supply. 16 As a result, Americans born between 1945 and 1965 (ie, "baby boomers") have a higher prevalence of chronic HCV infection compared with other birth cohorts. 7 Prolonged chronic HCV infection in this aging birth cohort has resulted in particularly high rates of liver cancer. Only the oldest baby boomers (ie, 1945-1947 birth cohorts) are included in the current study; therefore, rising HCC rates presented herein largely reflect earlier birth cohorts.
Given the higher HCV prevalence noted among baby boomers, we might anticipate that rates of HCVassociated HCC in the Medicare population will continue to increase in coming years. However, that trend might be blunted by the availability of more effective therapy for chronic HCV. 17 Until recently, treatment consisted of pegylated interferon-α plus ribavirin, which produced a virological cure ("sustained virological response") in only approximately 50% of patients overall and an even lower percentage of individuals with advanced fibrosis or cirrhosis (ie, those at highest risk of HCC). In 2011, first-generation direct antiviral agents for HCV were introduced and, in 2014, highly effective direct antiviral agent regimens pushed sustained virological response rates to >90%. The successful treatment of chronic HCV markedly reduces the risk of developing HCC 18 ; however, many infected individuals are unaware that they have chronic HCV. 19,20 Future rates of HCVassociated HCC in the Medicare population will be affected by both the higher HCV prevalence among baby boomers and success in identifying and treating chronic HCV in this population.
The prevalence of chronic HBV in the United States was low (0.3% in 2007-2012) and did not change significantly from 1999 through 2012. 21 Although current therapies for patients with chronic HBV generally are not curative, in observational studies, antiviral therapy in patients with immune-active chronic HBV has been found to reduce the risk of HCC by approximately 50%. [22][23][24] Current initial therapy recommendations for patients with chronic HBV include pegylated interferon-α, entecavir, tenofovir disoproxil fumarate, or tenofovir alafenamide. 23 However, even with effective treatment, HCC still may develop. 25 Efforts currently are underway to develop new antiviral agents for HBV infection. 22,24 It is interesting to note that although universal vaccination against HBV infection is recommended in the United States, 26,27 it cannot impact HBV-attributable HCC occurring among individuals who already are chronic HBV carriers.
Although prior studies have reported on rising liver cancer rates, 1,2 the current study has presented rates disaggregated based on the HBV and HCV status of cases, thereby providing insight into the etiologic drivers of the rising rates. The rate of HCV-attributable HCC nearly doubled during 2001 through 2013, increasing by 4% to 9% per year among men and women; those aged 66 to 85 years; and white, black, and Hispanic individuals. In contrast, HCV-attributable HCC rates were stable among Asian individuals and those of other or unknown race/ethnicity. Although rates of HBV-attributable HCC also increased significantly, the APCs generally were lower than those for HCV-attributable HCC, and HCV-attributable HCC rates were 3.5 times higher than HBV-attributable HCC rates in 2013. It is interesting to note that HBV-attributable HCC rates among Asians, the group with the highest rates, remained stable over time.
HCC rates unrelated to HCV or HBV infection increased overall, among both men and women and in all age groups. These increases appear to be more notable among white and Hispanic populations. Nonviral causes of HCC that may be contributing to rising HCC rates include alcohol consumption, obesity, diabetes, and nonalcoholic fatty liver disease (NAFLD), which is the hepatic manifestation of metabolic syndrome. 10 The prevalence of alcohol abuse in the United States increased among all age groups between 2001 and 2002 and 2012 and 2013 28 and there have been steady increases noted in the prevalence of obesity (37.0% of adults aged ≥60 years in 2011-2014), type II diabetes (25.2% of individuals aged ≥65 years in 2015), and NAFLD (31% of adults aged 20-74 years in 2011-2012). [29][30][31][32] Associations between these risk factors and liver cancer are much weaker than those noted for HCV and HBV, and therefore the presence of one of these conditions among HCC cases cannot be concluded as causal. 33 Nonetheless, alcohol abuse, obesity, diabetes, and NAFLD are far more common in the United States than chronic viral hepatitis. It should be noted that these nonviral risk factors may act synergistically with HCV and HBV infection, 16 thereby further contributing to HCC cases attributable to viral hepatitis in this analysis.
Cancer August 1, 2019 The main strength of the current study was the use of linked data sets to distinguish between HCV-attributable, HBV-attributable, and HCV/HBV-unrelated HCC cases. With the exception of a single older study that focused on HCC rates in the 1990s, 34 the current study is the only one to our knowledge that has estimated HCC incidence trends by viral hepatitis status. Although a recent study reported HCC mortality trends by viral status, this study relied on hepatitis information reported on death certificates, 35 which is known to be vastly underreported: a prior study estimated that only 31% of all liver cancer deaths reported among individuals with chronic HCV infection listed HCV on the death certificate. 36 In addition, the SEER-Medicare database is a large, nationally representative resource, and we have shown that HCC trends based on Medicare recipients closely represent rates from the same SEER cancer registries.
The main limitation of the current study was the use of medical claims, which do not capture HBV or HCV status with complete sensitivity or specificity and do not capture infection duration. To enhance specificity, we classified HCC cases as positive for HBV or HCV based on 1 inpatient claim or 2 physician or outpatient claims filed at least 30 days apart. An additional limitation of the current study was the lack of information regarding body mass index, cigarette smoking, and average alcohol consumption. We had limited power to detect significant changes in rates in some groups due to the small number of HCC cases. Finally, the age range of this analysis was restricted to Americans aged ≥66 years, and thus it did not capture the birth cohort with the highest HCV prevalence in the United States. The prevalence of HCV among patients with HCC is likely higher among Americans who were aged in their 40s and 50s during the study period, and HCV-attributable HCC rates for this birth cohort likely increased more rapidly than reported herein. However, it is important to note that approximately 41% of patients with HCC diagnosed in SEER during 2001 through 2013 were aged ≥66 years and therefore the Medicare-aged population captures a substantial percentage of HCC cases in the United States.
Among Americans aged ≥66 years, HCC rates increased rapidly between 2001 and 2013. HCVattributable HCCs contributed substantially to this observed increase, with a smaller contribution observed by HBV-attributable HCCs. Interventions to treat and prevent viral hepatitis infections will affect future HCC rates among Medicare-aged Americans. Nonviral-related HCC cases will likely continue to increase in the United States, largely due to sustained increases in obesity and metabolic syndrome-related conditions. Interventions focused on modifiable risk factors for HCC, such as promoting a healthy body weight and reducing alcohol intake and cigarette smoking, also could have an impact on future HCC rates.