|Home | About | Journals | Submit | Contact Us | Français|
There is global variation in the onset of hepatocellular carcinoma (HCC). We aimed to investigate the impact of country of birth on the age of HCC diagnosis in the US.
Incident HCC cases diagnosed between 2000 and 2012 in the Surveillance, Epidemiology, and End Results Program (SEER) 18 registry were included. Factors associated with very early onset (age of diagnosis <40) and early onset (age of diagnosis <50) were identified by logistic regression.
A total of 59,907 patients were included. The median age at HCC diagnosis was 62 years and 76% were male. Of the 75% of patients with birth country information, 29% were foreign born. In multivariate logistic regression, birth in West Africa (adjusted odds ratio [AOR]: 16.3, 95% confidence interval [CI]: 9.2–27.9, P<0.01), Central/South/Other Africa (AOR: 11.0, 95%CI: 4.5–23.7, P<0.01), Oceania (AOR: 4.9, 95%CI: 2.9–8.0, P<0.01) and East Africa (AOR: 3.5, 95%CI: 1.5–6.8, P<0.01) had the strongest associations with very early onset HCC after adjusting for gender and race/ethnicity. Birth in West Africa, Central/South/Other Africa, Oceania or East Africa was also strongly associated with early onset HCC.
Birth country was independently associated with age of HCC diagnosis in the US. Birth in Africa (except for North Africa), and Oceania was strongly associated with very early onset HCC. These finding have implications for the design of comprehensive HCC surveillance programs in the US.
Hepatocellular carcinoma (HCC) is the second leading cause of cancer related death in the world.1 Over the past decade, there were marked increases in the incidence rates of HCC in the US.2 Viral hepatitis is the leading cause of HCC, accounting for more than 90% of all HCC cases, particularly in developing countries.3 There are racial and regional variations in the risk factors for HCC, which result in different demographic and clinical characteristics of HCC across the US and world.4 HCV is the leading etiology of HCC in North America, Europe, Japan and South America while HBV is the major cause of HCC in most of Asia and Sub-Saharan Africa.5 It is well known that HBV-associated HCC tends to occur earlier than HCC from HCV or other etiologies, as HBV infection is acquired during birth or early childhood in HBV endemic areas. Aflatoxin exposure, which is quite common in some HBV endemic areas, can also contribute to genetic instability, causing early onset HCC.6 For this reason, the American Association For the Study of Liver Diseases (AASLD) recommends starting HCC surveillance after the age of 20 years for African/North American Black with HBV, 40 years for Asian men, and 50 years for Asian women with HBV.7 However data to support these recommendations are scarce in the literature as there are few data on HCC epidemiology in HCC endemic areas, particularly in Africa, where HCC is highly endemic. Indeed, in the recent GLOBOCAN 2012 report, cancer incidence data was not available for a third of the countries, most of which are developing/underdeveloped countries.8
Our recent publication showed that HCC tends to occur at a younger age in Africa, with a median age of diagnosis of HBV-associated HCC of only 42.9 This raises an important question about the optimal strategy for reducing the burden of morbidity and mortality from HCC among immigrants in the US, and by extension in regions where HCC is highly endemic but epidemiologic data are lacking. We hypothesized that the country of birth affects the onset of HCC in the US. To test our hypothesis, we aimed to investigate the impact of country of birth on the age of HCC diagnosis in the US general population using the most current Surveillance, Epidemiology, and End Results Program (SEER) 18 registry.
Incident HCC cases, diagnosed between January 1 2000 and December 31 2012, were extracted from the SEER 18 registry using the International Classification of Diseases for Oncology (third edition), topography codes (C22) and morphology codes (8170–8175).10 The SEER 18 registries include Metropolitan Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco-Oakland, Seattle-Puget Sound, Utah, Los Angeles, San Jose and Monterrey, Rural Georgia, the Alaska Native Tumor Registry, Greater California, Kentucky, Louisiana, New Jersey, and Greater Georgia and covers approximately 28% of the United States population.11
Age at the time of HCC diagnosis, gender and country of birth were extracted from the SEER registry. As country of birth data was not publicly available, the data was obtained after SEER scientific review of the study protocol. Reported country of birth was classified according to the World Population Prospects, 2015 Revision, from the United Nations.12
Europe includes Albania, Austria, Belarus, Belgium, Bulgaria, Caucasian Republics of former USSR NOS, Channel Island, Cyprus, Czech Republic, Czechoslovakia (former) NOS, Denmark, England, England, Isle of Man NOS, Europe NOS, Finland, France, Germanic Countries NOS, Germany, Gibraltar, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Malta, Netherlands, Norway, Poland, Portugal, Romania, Russian Federation, Scotland, Slavic Countries NOS, Spain, Sweden, Switzerland, Ukraine, Ukraine and Moldova NOS, United Kingdom, Wales and Yugoslavia (former) NOS.
Other North or Central America/Caribbean include Bahamas, Belize, Bermuda, Canada, Central America NOS, Costa Rica, Cuba, Dominican Republic, El Salvador, Guatemala, Haiti, Honduras, Jamaica, Mexico, Nicaragua, North America NOS, Other Caribbean Islands NOS, Panama, Puerto Rico, Virgin Islands, U.S. South America includes Argentina, Bolivarian Republic of Venezuela, Bolivia, Brazil, Chile, Colombia, Ecuador, Guyana, Peru, South America NOS and Uruguay.
East Asia includes China, China NOS, Hong Kong, Democratic Peoples Republic of Korea (North Korea), Japan, Mongolia, Province of China, Republic of Korea (South Korea) and Taiwan. South East Asia includes Brunei NOS, Cambodia, Indonesia, Lao Peoples Democratic Republic (Laos), Macao, Malaysia, Myanmar, Philippines, Singapore, Southeast Asia NOS, Thailand and Vietnam. Other Asia includes Afghanistan, Arabian Peninsula NOS, Armenia, Asia NOS, Bangladesh, India, Iran, Islamic Republic of Iraq, Israel and Palestine NOS, Jordan, Lebanon, Maldives, Nepal, Other Asian Republics of former USSR NOS, Pakistan, Sri Lanka, Syrian Arab Republic, Turkey and Yemen. Oceania includes American Samoa, Australia, Federated States of Micronesian Islands NOS, Guam, Kiribati, Marshall Islands, Melanesian Islands NOS, Micronesia, New Zealand, Northern Mariana Islands, Pacific NOS, Palau, Polynesian Islands NOS and Tonga.
West Africa includes Nigeria, West Africa NOS. East Africa includes Burundi, East Africa NOS, Ethiopia, Ethiopia and Eritrea NOS, Kenya, Mauritius, Rwanda, Somalia and Uganda. Central/South/Other Africa includes Africa NOS, African Islands NOS, Angola, Congo and Democratic Republic of the South (southern) Africa NOS. North Africa includes Egypt, Morocco, North Africa NOS, Sudanese Countries NOS and Tunisia.
The Chi-square test was used to compare categorical variables and the ANOVA test for continuous variables. Factors associated with very early onset HCC (age of HCC diagnosis<40), early onset HCC (age of HCC diagnosis <50) or age <60 were identified by multivariate logistic regression analysis. The main independent variables were gender, race/ethnicity (Non-Hispanic White, Non-Hispanic Black, Non-Hispanic Asian Pacific Islander [API] and Hispanic), and country of birth. Statistical analyses were performed using JMP statistical software v.10 (SAS Institute, Cary, NC). Joinpoint regression analysis for annual percent change was performed by SAS. All P values presented are 2-sided and differences were considered statistically significant when P<0.05. Two sensitivity analyses were performed. First, logistic analysis was performed after excluding patients with age less than 20 to minimize the potential misdiagnosis of HCC as HCC rarely develops before the age of 20. Second, logistic regression analysis was repeated after excluding HCC cases with recent diagnosis years (2010–2012), due to the higher rate of missing country of birth data.
A total of 59,907 patients diagnosed with HCC between 2000 and 2012 included in the SEER registry were analyzed in the study. Demographic characteristics of patients were summarized in Table 1. Country of birth information was available for 75% of patients and 29% of the patients with country of birth information available were foreign born. The proportion of foreign born persons was significantly different in race/ethnicity subgroups: 82% of APIs, 48% of Hispanics, 8% of Whites and 6% of Blacks were foreign born (P<0.01) (Figure 1). Overall, the proportion of foreign born patients remained stable over time in all race/ethnicity subgroups. Due to the variations in the proportions of foreign born persons in each race/ethnicity subgroup, US born patients with HCC were most commonly White (64%), while foreign born HCC patients were most commonly API (55%).
The median age at HCC diagnosis was 62 years and 76% were male. The median age at HCC diagnosis was younger among men than women (61 vs. 68, P<0.01) and varied by race/ethnicity (Black , White , API  and Hispanic ) (P<0.01). The median age at HCC diagnosis was higher in foreign born than US born White patients (P<0.01). The median age at HCC diagnosis was lower in foreign born than US born Black patients (P<0.01). The median age at HCC diagnosis was similar between foreign and US born API and Hispanic patients.
The median age of HCC diagnosis varied among persons born in different regions of the world (Figure 2). The cumulative proportion of HCCs with increasing age in different regions of the world is shown in Table 2 and the proportion of early onset HCCs per country or region is shown in Figure 3. The proportion of US born patients with very early onset HCC (age of HCC diagnosis <40) was less than 3% in each race/ethnicity subgroup. The proportion of persons with very early onset HCC was highest in those born in Africa, particularly West Africa (28%) followed by persons born in Oceania (11%). The proportion of persons with very early onset HCC was 3–4% in persons born in Asia while it was less than 3% in persons born in all other regions.
The proportion of US born patients with early onset HCC (age of HCC diagnosis <50) was 15% or less in each race/ethnicity subgroup. The proportion of patients with early onset HCC was again highest in patients born in West Africa (57%) followed by patients born in Oceania (24%). The proportion of early onset HCC was between 10 and 15% in patients born in Asia while it was lower than 5% in patients born in Europe. With regard to the proportion of patients with age of diagnosis <60, <70, and <80, the proportions remained highest in patients born in West Africa and lowest in patients born in Europe.
Factors associated with very early/early age onset of HCC in multivariate logistic regression are summarized in Table 3. Black race (AOR: 1.8, 95%CI: 1.5–2.2, P<0.01), API race (AOR: 1.9, 95%CI: 1.6–2.4, P<0.01) and country of birth were associated with very early onset of HCC, <40 years. Birth in West Africa (AOR: 16.3, 95%CI: 9.2–27.9, P<0.01), Central/South/Other Africa (AOR: 11.0, 95%CI: 4.5–23.7, P<0.01), Oceania (AOR: 4.9, 95%CI: 2.9–8.0, P<0.01) or East Africa (AOR: 3.5, 95%CI: 1.5–6.8, P<0.01) had the strongest associations with very early onset of HCC. Male sex appeared to be protective against very early onset of HCC (adjusted odds ratio [AOR]: 0.8, 95% confidence interval [CI]: 0.7–0.9, P<0.01).
For early onset HCC, independent predictors were male gender (AOR: 1.4, 95%CI: 1.3–1.5, P<0.01), minority race/ethnicity; including Black race (AOR: 1.4, 95%CI: 1.3–1.5, P<0.01), API race (AOR: 1.4, 95%CI: 1.3–1.6, P<0.01) and Hispanic ethnicity (AOR: 1.7, 95%CI: 1.6–1.8, P<0.01) and country of birth. Birth in West Africa (AOR: 10.5, 95%CI: 6.6–16.9, P<0.01), Central/South/Other Africa (AOR: 4.0, 95%CI: 2.0–7.5, P<0.01), Oceania (AOR: 2.5, 95%CI: 1.8–3.5, P<0.01) or East Africa (AOR: 2.1, 95%CI: 1.3–3.3, P<0.01) had the strongest associations with early onset HCC. Birth in Japan (AOR: 0.2, 95%CI: 0.1–0.4, P<0.01), Canada (AOR: 0.4, 95%CI: 0.1–0.9, P=0.04) or Europe (AOR: 0.5, 95%CI: 0.4–0.7, P<0.01) had the strongest inverse associations with early onset HCC.
For the age of HCC diagnosis <60, independent predictors were male gender (AOR: 2.0, 95%CI: 1.9–2.1, P<0.01), Black race (AOR: 1.7, 95%CI: 1.6–1.8, P<0.01), Hispanic ethnicity (AOR: 1.4, 95%CI: 1.3–1.5, P<0.01) and country of birth. Birth in West Africa (AOR: 3.3, 95%CI: 1.9–5.9, P<0.01) or Oceania (AOR: 2.2, 95%CI: 1.7–3.0, P<0.01) had the strongest associations with age of HCC diagnosis <60. Birth in Canada (AOR: 0.4, 95%CI: 0.3–0.6, P<0.01), Europe (AOR: 0.4, 95%CI: 0.4–0.5, P<0.01) or Japan (AOR: 0.5, 95%CI: 0.4–0.6, P<0.01), had the strongest inverse associations with age of HCC diagnosis <60.
The proportion of patients with missing country of birth information increased over time during the study period (Supplementary Figure 1). As the proportion of patients with missing country of birth increased steeply after 2010, we performed additional logistic regression analysis after excluding cases diagnosed between 2010 and 2012 (n=18,377). The main results did not change significantly (Supplementary Table 1). Next, the analysis was repeated after excluding patients aged less than 20 (n=168) in order to avoid potential misdiagnosis of HCC, considering that development of HCC before the age of 20 is extremely rare. Again the main findings remained essentially the same (Supplementary Table 2).
In a US population based study using the SEER registry, we found that country of birth was associated with the age of HCC diagnosis, after adjusting for gender and race/ethnicity. The median age of HCC diagnosis was 7 years younger in men compared to women (P<0.01). Non-White race and Hispanic ethnicity were associated with earlier age of HCC diagnosis. The vast majority of White and Black patients were US born. On the other hand, most Asian and 40% of Hispanic HCC patients were foreign born. The proportion of foreign born HCC patients remained largely stable between 2000 and 2012. Logistic regression analysis showed that country of birth had the largest impact on the age of HCC diagnosis. Birth in West Africa, Central/South/Other Africa, Oceania and East Africa were associated with a 16.3-fold, 11.0-fold, 4.9-fold and 3.5-fold increased risk of very early onset of HCC (age of HCC diagnosis <40), respectively. Again, birth in West Africa, Central/South/Other Africa, Oceania and East Africa were associated with a 10.5-fold, 4.0-fold, 2.5-fold and 2.1-fold increased risk of developing early onset HCC (age of HCC diagnosis <50), respectively. On the other hand, birth in Europe, Canada or Japan was inversely associated with early onset of HCC.
The data clearly showed that country of birth, rather than race/ethnicity, had the largest impact on the age of HCC diagnosis in the US general population. In particular, the high proportion of early onset HCC in persons born in Africa is quite striking. This finding closely mirrors the local epidemiology of HCC in that region and is consistent with previous study findings on the epidemiology of HCC in different regions of the world.9, 13 For example, we have developed a consortium of collaborating centers to establish a database of HCC cases in order to describe the clinical features of patients with HCC in Africa.9 In our interim analysis of 1552 HCC patients from 14 different institutions in 7 African countries, the median age at the time of HCC diagnosis was 44 in West Africa. Similar to the results of the SEER study reported here, about a third of West African HCC patients developed HCC before the age of 40. A nationwide Gambia Liver Cancer Study of 216 HCC cases revealed similar results where the median age of HCC onset were 47.14 The global cancer statistics database published by the International Agency for Research on Cancer, GLOBOCAN 2012, reported 11- and 6-fold higher age specific HCC incidence rates in West Africa and sub-Saharan Africa compared to the US in individuals aged between 15–39 years old, respectively.15 These represent significantly higher incidence rates at early ages, considering that the age standardized incidence rates of HCC were only 2- and 1.7-fold higher in West Africa and Sub-Saharan Africa compared to the US, respectively.15 On the other hand, the age specific HCC incidence rates in North Africa were the same as in the US in individuals aged between 15–39 years old, while the age standardized incidence rates of HCC were 2-fold higher in North Africa compared to US.15 Thus there are clear regional variations in HCC epidemiology in countries and regions within Africa. The SEER data are also consistent with previous studies of HCC outside Africa. For example, the global HCC BRIDGE (‘Bridge to Better Outcomes in HCC’) study was a large multiregional longitudinal cohort study designed to describe the global epidemiology, practice patterns and outcomes of 18,031 HCC patients in North America, Europe and Asia.13 The mean age of HCC diagnosis was significantly higher in Japan and Europe and lower in other Asian countries in comparison to the US.
The median age at diagnosis is dependent on the age structure of the population.16 As the age structure of the foreign born populations in the SEER populations sites is not available, we compared the median age of the US born population (35.0 years) with the median ages for African born (36.1), Oceania born (37.7), Asian born (38.7), Latin America born (34.2), Europe (50), and non-US North America born individuals (48.5) based on data from the US 2010 Census.17, 18 Since the median age of the US born population is not substantially different from the median age of the African born or Oceania born persons, it appears unlikely that difference in the age distribution of the population would completely account for the observed differences in age at HCC diagnosis. In addition, we were able to obtain data on the proportions of persons aged less than 45 years in the different groups from the US 2000 Census; this proportion was 66% for the US born population, while it was 75% for Africa born, 67% for Oceania born, 64% for Asia born, 74% for Latin America born, 42% for Europe born and 44% for other non-US North America born individuals. While a higher proportion of young people among African born may have affected the results to some extent, it is again highly unlikely that adjusting for population structure would substantially reduce the AORs of early onset HCC incidence in Africa born and Oceania born individuals compared to US born.
Regional variation in the age onset of HCC could be explained by different degree of risk factors exposure. For example, the age at which HCC develops in individuals infected with HBV or HCV is closely related to their age at acquisition of infection. HBV infection is acquired before the age of 5 years in the majority of cases in many Asian and African countries, which could lead to earlier onset HCC.5, 19 The synergistic interaction between HBV infection and environmental exposures such as aflatoxin, which is widely prevalent in these regions, may be a major contributor to the earlier onset of HCC in these regions.20–22 Co-infections with HBV/HDV and HBV/HCV are also more common in some of these endemic areas, potentially further contributing to earlier onset HCC.14, 23 Geographic variation in the HBV genotype and integration patterns may also contribute to early-onset HCC in specific HCC endemic regions.24 On the other hand, HCV is acquired in adulthood, leading to later onset of HCC in most European and American countries.25
Our study results have important clinical implications. First, they have implications for HCC surveillance strategies in the US. It is important to recognize that place of birth rather than race/ethnicity per se has a major impact on the age of onset of HCC. Identifying high-risk foreign-born individuals and enrolling them into HCC surveillance programs at appropriate earlier ages should be strongly considered in the development of national guidelines. On the other hand, the risk of early onset HCC does not appear to be as high in US born Blacks or APIs compared to their foreign born counterparts.
This study has several limitations. First, a quarter of patients did not have country of birth information, which may introduce bias. The main source of data for country of birth information is death certificates. Therefore, country of birth information was preferentially obtained from decedents. For this reason, a higher proportion of patients had missing country of birth information among persons diagnosed with HCC after 2010 (Supplementary Figure 1). Our sensitivity analysis showed a minimal impact of excluding HCC cases diagnosed after 2010 on the AOR. Further, the AORs of persons with unknown country of birth were between 1.07 and 1.25, suggesting a minimal impact of unknown country of birth on the age of HCC diagnosis. In addition, the SEER registry does not have information on the etiology of liver disease, severity of any underlying liver disease or aflatoxin exposure. Thus, our study was not able to objectively adjust for the impact of etiology and aflatoxin or other environmental exposures on the age of HCC diagnosis. Overall, compared to those born in other regions, the number of US HCC patients born in Africa was relatively small, approximately 1% of all US HCC cases of known country of birth. Further, the findings of the current study reflect attributes of cases in the US immigrant population and should not be viewed as a proxy for age of diagnosis in a specific country of origin. Due to the privacy protection rule of SEER, the numbers of cases in some regions (Non-West Africa, Other North America, Central and South America) were combined in Table 2. The age at the time of immigration or duration of stay in the US was not available and might also introduce a potential bias.
In conclusion, race/ethnicity, gender and country of birth were independently associated with the age of HCC diagnosis in the US general population. Birth in Africa (except for North) and Oceania was strongly associated with very early onset of HCC, mirroring the HCC epidemiology in these countries. These finding have implications for the design of comprehensive HCC surveillance programs in the US and potentially for strategies to minimize the burden of morbidity and mortality from HCC in regions such as Africa and Oceania where HCC is highly endemic but robust epidemiologic data are lacking.
We appreciate Jeremy Lyman for assistance with drawing the global maps.
Fundings: This publication was supported by Grant Number T32 DK07198 from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (to JY) and CA165076 from the National Cancer Institute (NCI)(to LRR). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.
Conflict of Interest: Nothing to disclose
Author Contributions:Contributed to study design, acquisition of data, analysis and interpretation of data, and drafting of the manuscript: Ju Dong Yang
Contributed to study concept and design, interpretation of data, and critical revision of the manuscript: Sean Altekruse; Mindie H. Nguyen; Gregory J. Gores; Lewis R. Roberts