Decreasing incidence of hepatocellular carcinoma among most racial groups: SEER‐22, 2000–2019

Abstract Background Hepatocellular carcinoma (HCC) incidence was rising in the United States. Previously, using data collected by the Surveillance, Epidemiology, and End Results (SEER) Program through 2017, we found that overall incidence had begun to decline, although not in Black and American Indian/Alaska Native (AI/AN) populations. Utilizing expanded SEER data encompassing ~50% of the population, we examined secular trends and demographic differences in HCC incidence through 2019. Methods We included cases of HCC diagnosed in adults aged ≥20 years residing in SEER‐22 registry areas. We examined case counts, incidence rates (per 100,000 person‐years), annual percent changes (APCs), and calendar years when APCs changed significantly. Results HCC incidence increased from 5.56 in 2000 to 8.89 in 2009 (APC, 5.17%), then rose more slowly during 2009–2015 (APC, 2.28%). After peaking at 10.03 in 2015, incidence fell to 9.20 in 2019 (APC, −2.26%). In Asian/Pacific Islanders (A/PI), the decline began in 2007 and accelerated in 2015 (APCs: 2007–2015, −1.84%; 2015–2019, −5.80%). In 2014, incidence began to fall in the White (APC: 2014–2019, −1.11%) and Hispanic populations (APC: 2014–2019, −1.72%). In 2016, rates began to fall in Black individuals (APC: 2016–2019, −6.05%). In the AI/AN population, incidence was highest in 2017, although the subsequent decline was not statistically significant. In 2019, population‐specific rates were: White, 6.94; Black, 10.74; A/PI, 12.11; AI/AN, 14.56; Hispanic, 15.48. Conclusion HCC incidence is now decreasing in most US racial/ethnic populations, including among Black individuals. The onset of decline differed among racial/ethnic groups and wide disparities in HCC rates remain.


| INTRODUCTION
Liver cancer is a highly lethal malignancy that is the sixth leading cause of cancer mortality in the United States. 1 The incidence of hepatocellular carcinoma (HCC), the most common form of liver cancer, rose sharply in the United States beginning in ~1980; 2 however, using data from 21 Surveillance, Epidemiology, and End Results (SEER) Program registries that was available through calendar year 2017, we previously found that overall HCC incidence peaked in 2014-2015. 3,4We also observed marked differences in secular trends among racial/ethnic groups with an early decline in HCC incidence among Asian/ Pacific Islander individuals, but no evidence of a decline in Black or American Indian/Alaska Native populations. 3,4ecently, the SEER Program was modified and expanded to encompass 22 registries (SEER-22), which together capture data for almost 50% of the US population.With this larger dataset, we have now examined temporal trends and demographic differences in HCC incidence through 2019 to provide additional information regarding racial disparities for this cancer.

| Data source
The National Cancer Institute funds population-based SEER Program registries around the United States.For all newly diagnosed cancers among residents of those catchment areas, information is collected on patient demographics, date of diagnosis, and tumor characteristics, but not potential etiologic factors (https://seer.cancer.gov/about/overv iew.html).For the present study, we included HCC cases diagnosed between 2000 and 2019 among adults aged ≥20 years residing in 22 SEER registries (Alaska Native Tumor Registry, Connecticut, Atlanta, Greater Georgia, Rural Georgia, San Francisco-Oakland, San Jose-Monterey, Los Angeles, Greater California, Hawaii, Iowa, Idaho, Illinois, Kentucky, Louisiana, Massachusetts, New Mexico, New Jersey, New York, Seattle-Puget Sound, Texas, and Utah; https://seer.cancer.gov/regis tries/ terms.htmlaccessed October 27, 2022).

| Case definitions and other variables
We selected cancers for analysis based on codes from the International Classification of Diseases for Oncology, Third Edition [ICD-O-3]).We defined HCC as cancers with ICD-O-3 primary site C220 (liver) and histology codes 8170-8175 (hepatocellular carcinoma).We examined incidence in five nonoverlapping categories for race/ethnicity (White; Black; Asian/Pacific Islander; American Indian/Alaska Native; Hispanic) All individuals of Hispanic ethnicity were classified as 'Hispanic' and excluded from other race/ethnicity categories.

| Statistical analysis
We examined the trends in overall HCC incidence and also stratified these analyses by sex, age at diagnosis (20-39; 40-64; 65-79; ≥80 years), race/ethnicity and SEER registry.Rates among American Indian/Alaska Native individuals were restricted to the Purchase/Referred Delivery Areas.
We calculated case counts and incidence rates (per 100,000 person-years) using SEER*Stat (version 8.3.6), age standardizing the rates to the 2000 US standard population by 5-year age groups.
To quantify trends in incidence, we used the Joinpoint Regression Analysis program (version 4.7.0.0) to calculate annual percent changes (APCs) and corresponding 95% confidence intervals (CIs).This program selects the best fitting log-linear regression model to identify calendar years during which the APCs changed significantly.We used 2-sided t-tests to calculate p-values.
Overall HCC rates rose and then fell during the 2000-2019 study period (Figure 1A; Table 2).Using output from the Joinpoint Regression Analysis program to identify calendar years when an APC changed significantly, we found that HCC Incidence increased from 5.56 in 2000 to 8.89 in 2009 for an annual percent change (APC) of 5.18% (95% CI: 4.78 to 5.57) and then rose more slowly to 10.03 in 2015 (APC: 2.27%; 95% CI: 1.55 to 3.01).After reaching that peak, HCC rates declined consistently and significantly, falling to 9.20 by 2019, (APC: −2.25%; 95% CI: −3.17 to −1.33).
From 2000 to 2019, rates rose by 88% in American Indian/Alaska Native, 73% in White individuals, 52% in Black individuals, and 49% in Hispanic individuals, while falling by 22% in Asian/Pacific Islander individuals (Table 3).Over the entire two decades, rates were consistently and notably lower among White individuals.In 2019, compared to the White population, rates were more than twice as high among the American Indian/Alaska Native and Hispanic populations (incidence rate ratio [IRR], 2.83 and 2.23, respectively) and > 50% higher in the API (IRR, 1.74) and Black (IRR, 1.55) populations (Table 3).With that important finding, there is evidence for a significant downward trend in all US racial ethnic groups except the American Indian/Alaska Native population, and even among American Indian/Alaska Native individuals there was some indication for a decline in rates beginning in 2017.While these declines are encouraging, there continued to be wide disparities in HCC incidence; in 2019, HCC was more than twice as frequent in American Indian/Alaska Natives and Hispanic populations and >50% more frequent in Black individuals and Asian/Pacific Islanders compared to White individuals.Several groups of investigators have examined recent data from SEER to determine the year of peak incidence for HCC in the United States.Previously, using data from the SEER-21 registries through 2017, we found that HCC incidence peaked in 2014. 3,4In contrast, based on data from 13 SEER registries for the years 1992-2017, Han et al. reported that HCC incidence began to decrease in 2011. 5lvarez et al. also used SEER-13 data, but extended the analysis to 2018, reporting that overall HCC incidence significantly declined between 2015 and 2018. 6The reason F I G U R E 1 Age-adjusted rates (per 100,000 persons) of hepatocellular carcinoma in the United States, overall (A) as well as by sex (B), age group (C), and race/ethnicity (D)-SEER 22, 2000-2019.Individuals of Hispanic ethnicity were classified as 'Hispanic' and excluded from other race/ethnicity categories.Open symbols indicate a joinpoint where the annual percent change differs between segments (p < 0.05).
for those seemingly discrepant results is not clear.Now, using data from the largest grouping of SEER registries to date (i.e., SEER-22), we also find that HCC incidence peaked in 2015.Some differences in estimates of the peak for HCC incidence in the United States may reflect use of different groupings of SEER registries, while other differences are more difficult to explain.Taken together, these reports provide evidence that HCC incidence in the United States began to decline in about 2015.
The most common causes of HCC in the United States are chronic hepatitis B (CHB), chronic hepatitis C (CHC), alcoholic liver disease (ALD), and nonalcoholic fatty liver disease (NAFLD).Of these factors, ALD and NAFLD are more common, but viral hepatitis infection carries a higher risk for HCC. 7SEER does not collect information on the presence of these etiologic factors or their treatment; however, data from other sources indicates that the prevalence of these conditions varies markedly by race/ethnicity.CHB is endemic in areas of Asia and Africa. 8 individuals (1.94%); prevalence was considerably lower in the US-born population but greater in Asian (0.79%) or Black (0.52%) individuals compared to White individuals (0.08%). 9NHANES underestimates the true prevalence of HCV infection in the United States because some groups who at increased risk of CHC are underrepresented or excluded from the survey (e.g., people who inject drugs and people who are incarcerated). 10onetheless, it is noteworthy that the prevalence of CHC in 2011-2016 NHANES differed markedly by race/ ethnicity: Black, 1.52%; Hispanic, 0.94%; White, 0.65%; Asian, 0.23%. 11These results from NHANES are consistent with our previous findings regarding the prevalence of viral hepatitis among older Americans with HCC, in which the Asian population had the highest incidence of HBV-associated HCC and the highest incidence of HCV-associated HCC was seen in the Black population. 12American Indian/Alaska Native are not well represented in NHANES, but other studies suggest a high prevalence of viral hepatitis in that group. 13,14The frequency of ALD and NAFLD also vary by race/ethnicity.In a 2014 national survey, the frequency of heavy drinking (i.e., consuming ≥5 drinks on ≥5 days in the past month) in US populations was: American Indian, 9.2%; White, 7.1%; Hispanic, 5.1%; Black, 4.5%; Native Hawaiian/Pacific Islander, 4.6%; Asian, 2.0%. 15The frequency of NAFLD in population-based cohorts was 22.9% in Hispanic individuals, 14.4% in the White population, and 13.0% in Black persons. 16The marked racial/ethnic differences in HCC incidence we observed likely reflect these differences in the prevalence of HCC risk factors.
Major improvements in the treatment of viral hepatitis occurred during 2000-2019.As neither alcohol use nor obesity decreased in the United States during that period, 17,18 more effective antiviral therapies likely explain the fall in HCC incidence.The effect of those advances in specific racial/ethnic populations depends on infection prevalence and treatment availability.Oral therapies that markedly reduce the risk of HCC in patients with CHB were introduced in 2005-2006, 19 the years just prior to peak HCC incidence in the Asian/Pacific Islander population.Through most of the study period, treatment for CHC consisted of poorly tolerated interferon-based regiments that resolved infection in ~40% of patients. 20In 2014, direct acting antiviral (DAA) therapies that raised the cure rate above 95% were introduced.Access to DAA regimens is lower among Medicaid recipients, 21 who are disproportionately Black or Hispanic. 22Major limitations in the availability of DAA therapies through the Indian Health Service, the agency that provides medical care to the American Indian and Alaska Native populations, have been noted. 23We speculate that the introduction of DAA therapies in 2014, possibly in conjunction with the 2012 CDC recommendation for the HCV screening of all individuals born in the United States between 1945 and 1965, 24 contributed to the peaking of HCC incidence the following year and that the later decline in HCC incidence in the Black and American Indian/Alaska Native populations reflects more limited access to DAA treatments.That scenario would point to the need for especially strong efforts to make treatment for hepatitis virus infections available to those populations.
A strength of this study is that SEER employs a rigorous protocol for diagnosis of cancer based on microscopic findings from pathology or cytology, radiography or clinical records and this paper presents results obtained from the largest collection of SEER registries that have been available to date.Our study also has important limitations.SEER does not collect information on etiologic factors that can cause HCC or on foreign nativity.We also lacked the information needed to examine subpopulations within the broadly defined Hispanic and Asian populations.
In summary, our study provides welcome evidence of a broad decrease in HCC incidence in the United States, but much remains to be done.Epidemiologically, it will be important to confirm that these trends continue in T A B L E 3 Incidence rate (per 100,000 persons), change in incidence, incidence rate ratio (IRR), and accompanying 95% confidence interval (95% CI) for hepatocellular carcinoma among those aged ≥20 years, by race/ethnicity- SEER 22, 2000- subsequent years.Efforts to identify and treat individuals with chronic viral hepatitis infections, especially in populations with a high prevalence of infection, as well as those with or at risk for ALD or NAFLD, are essential to ensure that trends in HCC incidence continue to improve.
Our analysis of SEER-22 data during the years 2000 through 2019 shows that overall US HCC incidence peaked in 2015.HCC incidence started to decline in the Black population in 2016, 2 years after White and Hispanic populations and 7 years later than the Asian and Pacific Islander population.
Incidence (per 100,000 persons) of hepatocellular carcinoma, by sex, age, and race/ethnicity-SEER 22, 2000-2019.Individuals of Hispanic ethnicity were classified as 'Hispanic' and excluded from other race/ethnicity categories.
T A B L E 1 In the 2011-2016 National Health and Nutrition Examination Survey (NHANES), CHB was most frequent in foreign-born Asian (3.85%) and Black 2019.Individuals of Hispanic ethnicity were classified as 'Hispanic' and excluded from other race/ethnicity categories.; **The White population is the referent for IRR calculations. *