Potential conflict of interest: Nothing to report.
In cohort studies of atomic bomb survivors and Mayak nuclear facility workers, radiation-associated increases in liver cancer risk were observed, but hepatitis B virus (HBV) and hepatitis C virus (HCV) infections were not taken strictly into account. We identified 359 hepatocellular carcinoma (HCC) cases between 1970 and 2002 in the cohort of atomic bomb survivors and estimated cumulative incidence of HCC by radiation dose. To investigate contributions of radiation exposure and hepatitis virus infection to HCC risk, we conducted a nested case-control study using sera stored before HCC diagnosis in the longitudinal cohort of atomic bomb survivors. The study included 224 HCC cases and 644 controls that were matched to the cases on gender, age, city, and time and method of serum storage, and countermatched on radiation dose. The cumulative incidence of HCC by follow-up time and age increased significantly with radiation dose. The relative risk (RR) of HCC for radiation at 1 Gy was 1.67 (95% confidence interval: 1.22-2.35) with adjustment for alcohol consumption, body mass index (BMI), and smoking habit, whereas the RRs for HBV or HCV infection alone were 63 (20-241) and 83 (36-231) with such adjustment, respectively. Those estimates changed little when radiation and hepatitis virus infection were fit simultaneously. The RR of non-B, non-C HCC at 1 Gy was 1.90 (1.02-3.92) without adjustment for alcohol consumption, BMI, or smoking habit and 2.74 (1.26-7.04) with such adjustment. Conclusion: These results indicate that radiation exposure and HBV and HCV infection are associated independently with increased HCC risk. In particular, radiation exposure was a significant risk factor for non-B, non-C HCC with no apparent confounding by alcohol consumption, BMI, or smoking habit. (HEPATOLOGY 2011;53:-)
Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide, and chronic infections with hepatitis B virus (HBV) or hepatitis C virus (HCV) are recognized as critically important risk factors for HCC. Our previous study actually showed that about 63% of HCC in atomic bomb survivors is related to HCV infection, 14% to HBV infection, and 2% to both HBV and HCV infections.1 However, an increase of non-B, non-C HCC without HBV and HCV infection has been noted recently in Japan.2, 3 The etiology of non-B, non-C HCC has been poorly understood, although alcoholic hepatitis, nonalcoholic fatty liver disease (NAFLD) including nonalcoholic steatohepatitis (NASH), and hemochromatosis4, 5 are known as risk factors. In Japan, NAFLD has increased along with Westernization of lifestyle, and most NASH cases have developed due to such lifestyle-related diseases such as obesity, diabetes mellitus, and hyperlipidemia.6 Obesity and diabetes mellitus, as well as NAFLD, have also recently received increased attention as risk factors for HCC.1, 7-12
An increased risk of liver cancer with radiation dose among atomic bomb survivors has been reported based on tumor registries, mortality studies, and pathology review,13-16 but hepatitis virus infection status was not taken into account. In three previous HBV studies at the Radiation Effects Research Foundation (RERF), the HBV surface antigen (HBsAg)-positive proportion increased with radiation dose.17-19 Previous research at RERF demonstrated no increase in the prevalence of anti-HCV antibody (anti-HCV Ab) with radiation dose,20 but reported supermultiplicative effects between radiation exposure and chronic HCV infection in the etiology of HCC without cirrhosis.21
On the other hand, the cohort study in workers at the Mayak nuclear facility demonstrated that the risk of liver cancer mortality was significantly associated with plutonium exposure,22 and that the incidence of HCC was marginally significantly associated with plutonium exposure.23 In the latest analysis, a significant plutonium dose-response relationship was observed for liver cancer mortality, with risk reasonably described by a linear function.24 However, liver cancer in those analyses included hepatoblastoma and intrahepatic cholangiocarcinoma as well as HCC. In addition, hepatitis virus infection status was not taken into account in a strict and in-depth manner, although HCC accounted for most of the liver cancer.
A lifespan study using B6C3F1 mice exposed to continuous low-dose-rate γ rays demonstrated that the incidence of HCC was significantly increased in male mice exposed to total doses equivalent to 8,000, 400, and 20 mGy and in females exposed to 8,000 mGy. However, the incidence of other liver tumors did not significantly increase except for that of hepatoblastoma in males exposed to 400 mGy.25
With the aim of determining whether radiation exposure is an independent risk factor for HCC, even after adjusting for hepatitis virus infection, alcohol consumption, body mass index (BMI), and smoking habit, we conducted a nested case-control study among atomic bomb survivors using stored sera. We also evaluated whether radiation, alcohol consumption, increase of BMI, and smoking habit contribute to increased risk for non-B, non-C HCC.
AHS, Adult Health Study; BMI, body mass index; CI, confidence interval; ERR: excess relative risk; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; RERF, Radiation Effects Research Foundation; RR, relative risk.
Patients and Methods
The Atomic Bomb Casualty Commission (ABCC) and its successor, the RERF, established the Adult Health Study (AHS) longitudinal cohort in 1958, in which more than 20,000 gender-, age-, and city-matched proximal and distal atomic bomb survivors and persons not present in the cities at the time of bombings are examined biennially in outpatient clinics in Hiroshima and Nagasaki.
Cases and Controls.
Incident cancer cases were identified through the Hiroshima Tumor and Tissue Registry and Nagasaki Cancer Registry, supplemented by additional cases detected by way of pathological review of related diseases.26 As described in our previous study,1 359 primary HCC cases were diagnosed among 18,660 AHS participants between 1970 and 2002 who visited our outpatient clinics before their diagnosis. Of these, 229 cases had serum samples obtained within 6 years before HCC diagnosis. After excluding five cases with inadequate stored serum, 224 cases remained for our study. There were no important differences in characteristics such as gender, age at HCC diagnosis, city, alcohol consumption, BMI, or radiation dose to the liver (among exposed persons) between HCC cases excluded due to nonavailability of stored serum and those included in the present study.
Three control sera per case were selected from the at-risk cohort members matched on gender, age, city, and time and method of serum storage, and countermatched on radiation dose in nested case-control fashion.27 Countermatching (to increase statistical efficiency for studying joint effects of radiation and other factors) was performed using four strata based on whole-body (skin) dose: zero dose (<0.0005 Gy), <0.05 Gy, <0.75 Gy, and ≥0.75 Gy (nonzero categories correspond roughly to tertiles of skin dose among all eligible exposed cases). At the time of each case diagnosis, one control serum was selected for each of the three dose strata not occupied by the case. Although the total number of potential matched control serum samples is 672, due to occasional lack of subjects with stored sera who met the matching and countermatching criteria, the total number of control serum samples actually selected was 644, which comprised 488 sera from unique noncase subjects and 156 sera from subjects sampled on repeated occasions.
Virological assays were performed on 211 case and 640 control sera, because 13 case samples and four control samples had insufficient stored sera for these assays. HBsAg and antibody to hepatitis B core antigen (anti-HBc Ab) were measured by enzyme immunoassay (EIA), and anti-HCV Ab was measured by second-generation EIA as described.28, 29 Qualitative detection of HCV RNA among anti-HCV-positive samples was performed using a thermocycler (Whatman Biometra, Goettingen, Germany) based on the nested polymerase chain reaction (PCR) method, as described.29 HBV infection (HBV+) status was defined as positive for HBsAg or having a high titer of anti-HBc Ab. HCV infection (HCV+) status was defined as positive for HCV RNA. Non-B, non-C status was defined as negative for HBsAg and not having a high titer of anti-HBc Ab (HBV−) as well as negative for HCV RNA (HCV−).
Radiation dose to the liver was estimated for each subject according to Dosimetry System DS02.30 A weighted sum of the gamma dose in gray plus 10 times the neutron dose in gray was used. Because of the countermatched selection of cases, direct comparison of doses between cases and controls in the study requires that control doses be weighted by the inverses of their selection probabilities.
Information on Alcohol Consumption, BMI, and Smoking Habit.
Information on alcohol consumption was obtained from the 1965 AHS questionnaire when available, with missing data complemented using the 1978 mail survey. Alcohol consumption was quantified as volume of each type of alcoholic beverage; mean ethanol amounts were calculated as grams per day as described.31 BMI (kg/m2) was calculated from height and weight measured at the AHS examination. Subjects were classified based on BMI quintiles with cutpoints of 19.5, 21.2, 22.9, and 25.0. Following the recommendations for Asian people by the World Health Organization (WHO), the International Association for the Study of Obesity, and the International obesity Task Force,32 21.3 to 22.9 kg/m2 was considered normal, 23.0 to 25.0 kg/m2 as overweight, and >25.0 kg/m2 as obese. We used information on BMI obtained 10 years before the time of HCC diagnosis or control matching because this condition is subject to change due to disease progression in the later stages before development of HCC. Information on smoking habit was obtained from the 1965 questionnaire; subjects were categorized as never, current (at time of survey), or former smoker.
This study (RERF Research Protocol 1-04) was reviewed and approved by the Research Protocol Review Committee and the Human Investigation Committee of RERF.
The nested case-control design was analyzed using a partial likelihood method analogous to that used for cohort follow-up studies,33 which is in practice the same as the conditional binary data likelihood for matched case-control studies34 except that the subjects (cases and “controls”) in the study are not completely independent due to repeated selection. Cumulative incidence of HCC by follow-up time (year) and age was derived according to the method of Nelson and Aalen, using Cox regression to adjust for age at start of follow-up. Cumulative incidence by radiation dose groups (0-0.0009, 0.001-0.999, and 1.0+ Gray) was compared using the Gehan/Breslow generalized Wilcoxon test. All factors other than radiation were analyzed using relative risks (RRs) estimated by a log-linear model. Although radiation exposure could have been adjusted by matching on radiation dose as an additional matching factor in the control selection,35 in addition to assessing effects of lifestyle factors and viral hepatitis, another purpose of the present study was to examine the effects of radiation exposure after adjustment for possible confounding and interaction by these factors, so matching on radiation—which precludes analysis of radiation risk—was not desirable; rather, we countermatched on radiation.27, 33, 36 Radiation risk was analyzed using an excess relative risk (ERR) model (ERR = RR-1) as done previously.37 The cumulative hazard estimator and comparisons by radiation dose groups were computed using Stata (StataCorp, College Station, TX; v. 11.1); all other analyses were conducted using Epicure (HiroSoft International, Seattle, WA; v. 1.81).
Characteristics of Cases and Controls.
Characteristics of the 224 HCC cases and 644 matched controls are shown in Table 1. HCC cases and controls were comparable with respect to gender, age, city, and time and method of serum storage by design. Prevalence of HBV and/or HCV infection status in HCC cases is higher than those in controls. Higher proportions of HCC cases had a history of alcohol consumption of more than 40 g of ethanol per day, were obese (BMI >25.0 kg/m2), and were current smokers, compared with the controls. HCC cases also received on average higher radiation doses to the liver, compared with the controls.
Table 1. Characteristics of HCC Cases and Controls
Number with Complete Data
Number with Complete Data
Weighted mean radiation dose (among controls), calculated by weighting according to their countermatching selection probabilities.
Figure 1A,B shows the cumulative incidence of HCC by radiation dose using either follow-up time (adjusted for age at start of follow-up) or age. Of 359 HCC cases diagnosed among 18,660 AHS subjects between 1970 and 2002, the analysis was performed using 322 HCC cases, based on 16,766 subjects with known radiation dose. A significant increase with radiation dose was seen with cumulative incidence both by follow-up time (P = 0.028) (Fig. 1A) and by age (P = 0.0003) (Fig. 1B). The effect of radiation was especially evident at age 60 years or later.
Risk of HCC for Radiation and Hepatitis Virus Infection.
Table 2 shows risk of HCC with and without adjustment for categorical alcohol consumption, BMI, and smoking habit based on all cases of HCC. The analysis was performed using 186 HCC cases and 600 controls, both separately (radiation only or hepatitis virus infection only) and jointly (radiation and hepatitis virus infection were fit simultaneously), based on subjects with known radiation dose and known HBV and HCV infection status. In analyses where effects of radiation and hepatitis virus infection were fitted separately, unadjusted RR at 1 Gy of HCC for radiation was 1.40 (95% confidence interval [CI], 1.07-1.89, P = 0.013), whereas unadjusted RRs of HCC for HBV+/HCV− status and HBV−/HCV+ status were 34 (95% CI, 13-106, P < 0.001) and 57 (95% CI, 27-140, P < 0.001), respectively. After adjustment for categorical alcohol consumption, BMI, and smoking habit, significant association was found between HCC and radiation dose or hepatitis virus infection, resulting in an RR at 1 Gy of 1.67 (95% CI, 1.22-2.35, P < 0.001) for radiation and RRs of 63 (95% CI, 20-241, P < 0.001) for HBV+/HCV− status and 83 (95% CI, 36-231, P < 0.001) for HBV−/HCV+ status. The above estimates changed little when radiation and hepatitis virus infection were fit simultaneously.
Table 2. Risk of HCC for Radiation and HBV or HCV Infection Status
Adjusted for categorical alcohol consumption, BMI 10 yrs before diagnosis, and smoking habit.
Radiation dose to the liver and hepatitis virus infection status were fit separately.
Radiation dose to the liver and hepatitis virus infection status were fit simultaneously.
Radiation (at 1Gy)
Risk of HCC for Radiation After Excluding Persons with Either or Both Hepatitis Virus Infections.
After excluding subjects with either or both hepatitis virus infections, the RRs at 1 Gy of HCC for radiation were estimated as shown in Table 3. There were 161 cases including 119 HCV-infected individuals and 452 matched controls including 29 HCV-infected individuals without HBV infection only. There were 66 cases including 24 HBV-infected individuals and 176 matched controls including 5 HBV-infected individuals without HCV infection only. The adjusted analyses indicated that radiation exposure was significantly associated with increased risks for HCC, even after excluding HBV- or HCV-infected individuals. Furthermore, significant association was found between non-B, non-C HCC and radiation dose, resulting in an RR at 1 Gy of 1.90 (95% CI, 1.02-3.92, P = 0.041) for radiation without adjustment for categorical alcohol consumption, BMI, and smoking habit and 2.74 (95% CI, 1.26-7.04, P = 0.007) with such adjustment.
Table 3. Risk of HCC for Radiation After Excluding Persons Infected with HBV and/or HCV
Effects of alcohol consumption, BMI, and smoking habit on non-B, non-C HCC risk with or without adjustment for radiation dose were estimated using continuous and categorical covariates as shown in Table 4. RRs for continuous covariates are for a one-unit difference in the factor. Risk of non-B, non-C HCC for alcohol consumption per 20 g of ethanol per day was significant with a log-linear model (adjusted RR 1.64, 95% CI, 1.05-2.81, P = 0.029), but was limited to the category ≥40 g of ethanol per day (adjusted RR 5.49, 95% CI, 0.98-39.2, P = 0.052). Significant log-linear association was not found with continuous BMI, and even the category BMI >25.0 kg/m2 (obese) 10 years before diagnosis did not evidence significant risk despite a rather large estimate of RR (adjusted RR 3.17, 95% CI, 0.92-12.3, P = 0.068). Current smoking evidenced significant risk (adjusted RR 5.95, 95% CI, 1.34-33.2, P = 0.018), but there were no continuous data on amount smoked. These results indicate that alcohol consumption per 20 g of ethanol per day, current smoking, and perhaps BMI of >25.0 kg/m2 10 years before diagnosis are associated independently with increased risk for non-B, non-C HCC.
Table 4. Risk of Non-B, Non-C HCC for Alcohol Consumption, BMI, and Smoking Habit
Alcohol consumption, BMI, and smoking habit were fit simultaneously, either as continuous (alcohol and BMI only) or categorical factors.
Alcohol consumption (per 20 g ethanol per day)
BMI 10 yrs before diagnosis (per +1 kg/m2 difference)
Alcohol consumption (g ethanol per day)
0 < < 20
20 ≤ < 40
BMI (kg/m2) 10 yrs before diagnosis
19.6 - 21.2
21.3 - 22.9
23.0 - 25.0
The present study confirmed that radiation is associated with increased incidence of HCC among atomic bomb survivors. Additionally, the nested case-control study indicates that radiation and HBV and HCV infection are associated with increased risk for HCC, and that radiation remains an independent risk factor for HCC after taking into account hepatitis virus infection, alcohol consumption, BMI 10 years before HCC diagnosis, and smoking habit. Furthermore, significant association was observed between non-B, non-C HCC and radiation dose, alcohol consumption, and smoking, whereas obesity 10 years before diagnosis was marginally significantly associated with increased risk for non-B, non-C HCC.
In the analysis (Table 2) in which radiation dose and hepatitis virus infection were fitted separately, radiation was significantly associated with increased risk for HCC with or without adjustment for alcohol consumption, BMI, and smoking habit. Although this finding is in agreement with our previous understanding that liver cancer risk is significantly associated with radiation without adjustment for hepatitis virus infection among atomic bomb survivors, it is difficult to compare the HCC risk estimates between the previous and current study results.13-16 The difficulty is caused by the inclusion of hepatoblastoma and intrahepatic cholangiocarcinoma in addition to HCC as liver cancer cases in analyses of tumor registry-based liver cancer risk (ERR at 1 Sv = 0.49),13 mortality study- and tumor registry-based15, 16 liver cancer mortality risk (male: ERR per Sv = 0.39, female: ERR per Sv = 0.35), and liver cancer risk (male: ERR per Gy = 0.32, female: ERR per Gy = 0.28), despite the fact that the majority of liver cancer cases were HCC. Because a relatively large fraction of liver cancer cases was included that were diagnosed only on the basis of death certificates,13, 16 complete exclusion of metastatic liver tumor cases from such cases may not have been possible. Metastatic liver tumor cases were excluded in an analysis of pathological review-based liver cancer risk (ERR per Gy = 0.81), but hepatoblastoma and intrahepatic cholangiocarcinoma were included with HCC.14
In the current analyses adjusted for alcohol consumption, BMI, and smoking habit, the RR estimates for radiation increased slightly and showed statistical significance with adjustment for HBV and HCV infection status. HBV infection may be considered an intermediate risk factor for HCC, because three of four previous HBV screenings demonstrated that HBsAg prevalence increases with radiation dose17-19, 38; therefore, adjustment for HBV infection status might be expected to result in a decreased radiation risk estimate. However, such interpretation is difficult because the risk estimate was also adjusted for HCV infection status, although anti-HCV Ab prevalence is not significantly associated with radiation dose.20 We therefore examined HBV and HCV infection status and concomitant radiation effects separately, excluding persons with one or the other viral infection.
RRs of HCC for radiation after excluding persons infected with HBV or HCV were generally higher than with the full data, but differed little depending on which virus was used for exclusion (Table 3). As with the full data, adjustment for HBV or HCV infection status reduced the statistical significance of the radiation effect but had little impact on the RR estimates themselves. The RR of HCC for radiation after excluding persons infected with HBV and HCV (i.e., the RR of non-B, non-C HCC for radiation) was significant with or without adjustment for alcohol consumption, BMI, and smoking habit. As there can be no viral mediation of the radiation risk in noninfected individuals, lower radiation risks estimated in infected individuals might be considered evidence of mediation, but mediation would imply that risk decreases with adjustment for viral infection status, which did not occur. The reduction in statistical significance with adjustment for HBV and HCV infection status might be due to loss of power when further parameters for the risks of HCC for hepatitis virus infection are estimated or the number of subjects is reduced by exclusion.
As with the results reported previously,1 there is evidence that alcohol consumption of ≥40 g/day ethanol and BMI >25.0 kg/m2 10 years before diagnosis are associated with non-B, non-C HCC risk (Table 4). However, the evidence is not as strong given the small amount of data after excluding persons infected with HBV and HCV. The current study demonstrates that smoking is significantly associated with non-B, non-C HCC risk, although lack of continuous data precluded estimation of the relationship to amount smoked. This finding is consistent with recent assessments by the International Agency for Research on Cancer (IARC) where HCC has been positioned as a smoking-related malignant disease.39 Some studies have shown effects of smoking on risk of HCC, but few studies have incorporated, in a strict and in-depth manner, HBV and HCV infections.11, 40
Cohort studies of atomic bomb survivors13-16 and Mayak nuclear facility workers22-24 have indicated beyond a doubt that radiation increases liver cancer risk, even though hepatitis virus infection was not taken into account. It is also well known that persistent long-term internal exposure to α particles from Thorotrast, a radioactive contrast agent, can induce hemangiosarcoma, cholangiocarcinoma, and HCC in humans.41-43 Because a significant radiation effect is observed in a high proportion of HCC cases having a p53 mutation, it has been suggested that p53 is one of the intracellular targets of atomic bomb radiation and thus a cause of the increased HCC incidence among atomic bomb survivors.44 A lifespan study in mice exposed to continuous low-dose-rate γ rays demonstrated that the incidence of HCC was significantly increased, especially in male mice.25 Liver weights of irradiated mice were significantly greater than those of nonirradiated controls, and the lipid content was significantly increased in irradiated mouse livers.45 It is considered that hepatic steatosis itself is a state conferring risk for high carcinogenicity, and that in steatohepatitis, oxidative stress due to fatty acid oxidation in hepatocytes may cause DNA injury and eventually lead to carcinogenesis.46 There is a significant association of radiation dose with prevalence of fatty liver among Nagasaki AHS participants, although a significant association has not been found between obesity (BMI ≥26.0 kg/m2) and radiation dose.47 These findings may explain part of the mechanism of increased risks of HCC with radiation exposure.
The main strengths of our study include its prospective cohort-based, nested case-control design, which minimizes selection bias, the use of stored sera, and a wealth of epidemiological information obtained prior to HCC diagnosis. It is difficult and expensive to perform full cohort serum analyses, whereas the nested case-control design utilized here can provide substantial reductions in cost and effort with little loss of statistical efficiency.36 Another major strength of our study is that it incorporated, in a strict and in-depth manner, hepatitis virus infection status and HCC cases were identified through the Hiroshima Tumor and Tissue Registry and Nagasaki Cancer Registry, supplemented by additional cases detected by way of pathological review of related diseases.26
A limitation of our study is that the joint effects of radiation and hepatitis virus infection could not be estimated from the standpoint of causality. As discussed previously, HBV and possibly HCV infection may act as intermediate risk factors in radiation-associated HCC. Previous studies have consistently demonstrated that prevalence of HBsAg increases with radiation dose within the AHS,17-19 although no dose response for anti-HCV Ab has been detected.20 Therefore, when the risk of HCC for radiation is estimated while controlling for HBV infection, some of the radiation risk may be absorbed in the coefficient for HBV infection. In other words, the radiation risk coefficient does not represent the radiation effect independent of mediation by HBV infection and the HCC risk for HBV infection itself is not correctly estimated, because the actual causal pathway is not explicitly modeled. In addition, we cannot easily disentangle the joint effects of radiation and HBV infection using standard regression models, because HBV infection is not a true confounding risk factor but an intermediate risk factor. Nevertheless, that the radiation risk did not decrease with concomitant adjustment for viral infection suggests that the practical extent of mediation may be small. We are currently developing methods of statistical analysis that jointly consider the dose response for the intermediate viral factor as well as the joint risk of HCC for both hepatitis virus infection and radiation in the countermatched, nested case-control design.
In conclusion, radiation exposure was associated with increased risk of HCC, even after adjusting for HBV or HCV infection, alcohol consumption, BMI, and smoking habit. Moreover, radiation exposure was an independent risk factor for non-B, non-C HCC with no apparent confounding by alcohol consumption, BMI, or smoking habit. The mechanistic form of joint effects of radiation and HBV or HCV infection on HCC risk could not be estimated, but the development of new statistical methods that jointly consider the dose response for the intermediate viral factor will make such an analysis possible in the future. In particular, in-depth understanding of the mechanisms by which radiation exposure as well as obesity, alcohol drinking, and smoking contribute to development of non-B, non-C HCC may lead to prevention, early detection, and better therapeutic strategies.
We thank Naomi Masunari and Sachiyo Funamoto for the collection and processing of the data and all members of the Division of Clinical Laboratories for excellent assistance. The RERF, Hiroshima and Nagasaki, Japan is a private, nonprofit foundation funded by the Japanese Ministry of Health, Labour and Welfare (MHLW) and the U.S. Department of Energy (DOE), the latter in part through the National Academy of Sciences. This publication was supported by RERF Research Protocol(s) 2-75 and 1-04.