Hepatocellular carcinoma (HCC) is the fourth leading. Obesity and Alcohol Synergize to Increase the Risk of Incident Hepatocellular Carcinoma in Men

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2010;8: Obesity and Alcohol Synergize to Increase the Risk of Incident Hepatocellular Carcinoma in Men ROHIT LOOMBA,*, HWAI-I YANG, JUN SU, DAVID BRENNER,* UCHENNA ILOEJE, and CHIEN-JEN CHEN, *Division of Gastroenterology and Division of Epidemiology, University of California at San Diego, La Jolla, California; Genomics Research Center, Academia Sinica, Taipei, Taiwan; Research and Development, Bristol-Myers Squibb Company, Wallingford, Connecticut; and Graduate Institute of Epidemiology, College of Public Health, National Taiwan University, Taipei, Taiwan BACKGROUND & AIMS: Body mass index (BMI) and alcohol use are risk factors for hepatocellular carcinoma (HCC). We performed a prospective study to determine if these factors have synergistic effects on HCC risk. METHODS: Over 14 years, we followed up 2260 Taiwanese men from the Risk Evaluation of Viral Load Elevation and Associated Liver Disease/Cancer Hepatitis B Virus (REVEAL HBV) Study Cohort who tested positive for the hepatitis B surface antigen (mean age, y; mean BMI, 24 3 kg/m 2 ); 20% reported alcohol use. Incident HCC cases were identified via linkage to the national cancer registry. Multivariate-adjusted hazard ratio (HR) and 95% confidence interval (CI) were estimated using Cox-proportional hazards models. RESULTS: In univariate analysis, the interaction between BMI and alcohol predicted incident HCC (P.029). Alcohol use and extreme obesity (BMI 30 kg/m 2 ) had synergistic effects on the risk of incident HCC in analyses adjusted for age (HR, 3.41; 95% CI, ; P.025) and multivariables (HR, 3.40; 95% CI, ; P.025). The relative risk estimate for the interaction and the attributable proportion from the interaction and synergy index were 1.59, 0.52, and 4.40, respectively; these indicate a multiplicative interaction between alcohol use and extreme obesity. In an analysis stratified into 4 World Health Organization categories of BMI and alcohol use, the risk of incident HCC increased in overweight (HR, 2.4; 95% CI, ), obese (HR, 2.0; 95% CI, ), and extremely obese (HR, 2.9; 95% CI, ) users of alcohol (P for trend.046). CONCLUSIONS: Obesity and alcohol have synergistic effects to increase the risk of incident HCC in hepatitis B surface antigen positive men. Lifestyle interventions might reduce the incidence of HCC. Keywords: Hepatocellular Carcinoma; Epidemiology; Cirrhosis; Nutrition. Hepatocellular carcinoma (HCC) is the fourth leading cause of cancer-related mortality worldwide. 1 The World Health Organization (WHO) estimates that HCC leads to 610,000 deaths each year globally. 1 Based on the WHO database, the burden of HCC is increasing and by the year 2030 it will become the third leading cause of cancer mortality, replacing colon cancer. 1 The Surveillance, Epidemiology, and End Results database of the National Cancer Institute also suggests that both incidence and mortality related to HCC is increasing 2,3 and will continue to increase until 2030 in the United States. 4 Despite improvement in treatment modalities, from 2000 to 2005, HCC incidence and mortality rates continue to increase. 5 Preventive measures in addition to treatment advances are needed to effectively control this global epidemic of HCC. Viral hepatitis B is the leading cause of HCC worldwide. Body mass index (BMI) and alcohol use are modifiable and independent risk factors for HCC and further modify the risk of individuals with chronic hepatitis B and C. 6 Studies suggest that obesity increases the risk of HCC by approximately 2-fold. 7,8 There is conflicting evidence whether alcohol use modifies the risk of HCC independent of or via interaction with smoking. 6,9 12 Here, we examine the joint effect of alcohol and obesity on incident HCC risk in a prospective cohort study. Obesity and alcohol both lead to fatty liver disease. Nonalcoholic fatty liver disease (NAFLD) has become the leading cause of liver disease in the United States and may lead to cirrhosis and HCC in a subset of individuals. 13 Several groups have reported that fatty liver and/or nonalcoholic steatohepatitis (NASH) increase the risk of HCC. 14 Up to 50% of HCC cases in the United States are not related to hepatitis B or C. 14 Bugianesi et al 15 showed that metabolic syndrome traits are seen more commonly in cryptogenic cirrhosis patients with HCC versus matched controls. Marrero et al 13 conducted a large, single-center study of 135 HCC cases and reported that NAFLD accounted for 47% of the HCC cases in patients with cryptogenic cirrhosis and 11% of all cases of HCC. Hashimoto et al reported a prospective study including 137 NASH patients, and showed that the 5-year cumulative incidence of HCC in NASH was 7.6%. 16 Therefore, NAFLD and NASH are emerging as major risk factors for HCC. Previous studies have shown that alcohol and obesity synergistically increase the risk of subclinical and clinical liver injury. 17 However, limited data exist on whether alcohol and obesity are additive or multiplicative in increasing the risk of HCC in patients with hepatitis B virus (HBV) infection. We hypothesize that obesity and alcohol are synergistic (multiplicative, not just additive) in increasing the risk of future Abbreviations used in this paper: ALT, alanine aminotransferase; AP, attributable risk proportion due to interaction; BMI, body mass index; CI, confidence interval; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HR, hazards ratio; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; RERI, relative excess risk due to interaction; REVEAL HBV, Risk Evaluation of Viral Load Elevation and Associated Liver Disease/Cancer Hepatitis B Virus; SI, synergy index; WHO, World Health Organization by the AGA Institute /$36.00 doi: /j.cgh

2 892 LOOMBA ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 10 development of HCC. To test this hypothesis, we conducted a prospective study in a well-characterized (Risk Evaluation of Viral Load Elevation and Associated Liver Disease/Cancer Hepatitis B Virus [REVEAL HBV]) cohort to examine the joint effect of obesity and alcohol on the future risk of HCC in men with chronic HBV infection over a 14-year follow-up period after adjusting for confounding variables using Cox-proportional hazards analysis. Methods Study Population A detailed description of the recruitment and blood sample collection and assay procedures has been reported previously and well characterized 6,18,19 : 2260 hepatitis B positive men who were part of the REVEAL HBV cohort 6,18,19 were included in this analysis. Information on demographic and lifestyle factors was collected via a structured personal interview by well-trained public health nurses. Overnight fasting blood samples were collected at baseline, and tests on serum hepatitis B surface antigen (HBsAg), hepatitis B e antigen (HBeAg), and HBV-DNA levels were performed using commercial kits and detailed laboratory procedures as described previously. 19 None of them had co-infection with hepatitis C. Because more than 99% of women in this cohort do not drink, they were not included in the analysis because of a lack of power to test our hypothesis. This was decided before conducting data analyses. All participants gave informed consent to participate in the REVEAL HBV study, and the data collection procedures were reviewed and approved by the Institutional Review Board of the College of Public Health, National Taiwan University. Duration of Follow-Up Evaluation Participants were followed up until June 30, 2004, with more than 14 years of follow-up evaluation. Assessment of Exposure BMI was defined as weight (kg) divided by the square of height (in meters). Weight and height were measured by a nurse at the baseline visit and was categorized into normal ( 23 kg/m 2 ), overweight (23 to 25 kg/m 2 ), obese (25 to 30 kg/ m 2 ), and extremely obese ( 30 kg/m 2 ) based on the guidelines for the adult Asian population published by the WHO. 20 Alcohol Alcohol use was ascertained at the baseline visit with a questionnaire as previously described. 21 Individuals were asked about habit of alcohol consumption (yes/no), starting age, and present alcohol use (yes/no). Duration of alcohol use (in years) was also documented. Alcohol drinkers were defined as having alcohol consumption at least 4 days per week for at least 1 year. Identification of Incident Hepatocellular Carcinoma Baseline screening for HCC was performed with an abdominal ultrasonographic examination in all participants. Cirrhosis was diagnosed using abdominal ultrasonography with a systematic and quantitative scoring algorithm, which was determined by the appearance of the liver surface and the liver parenchyma as well as by the size of the hepatic vessel and the spleen. 22,23 Unique national identification numbers were used to link with the computerized national cancer registry profiles in Taiwan to identify newly diagnosed HCC cases between January 1, 1991, and June 30, Subjects who had HCC identified through ultrasound screening at the baseline visit, through linkage to the national cancer registry, or from personal interview before or at enrollment were excluded from the analysis. The Taiwan nationwide cancer registry system was started in 1978, which has updated, accurate, and complete information on HCC. Because not all cases of HCC reported to the national cancer registry had histologic confirmation, chart reviews were performed on 164 HCC cases in HBsAg-positive carriers to further validate the primary HCC diagnosis. Therefore, all HCC cases included in this study met at least one of the following criteria: histopathologic confirmation, detection by at least 2 imaging tools (ultrasonography, angiography, or computed tomography), or 1 imaging diagnosis with a serum- fetoprotein level of 400 ng/ml or greater. Statistical Analysis Baseline characteristics of the population were described as mean (standard deviation) for continuous variables and proportions for categoric variables. Cox-proportional hazards analysis was conducted to determine the hazards of incident HCC over 14 years of follow-up evaluation and reported as a relative risk estimate. Univariate, followed by multivariable models (adjusted for age, BMI, alcohol use, serum HBV-DNA level when applicable, smoking, serum alanine aminotransferase [ALT] level, HBeAg status [yes/no], and cirrhosis at baseline visit) were examined. Joint effects of extreme obesity (yes/no) and alcohol use (yes/no) (and duration and dose of alcohol use) were examined using Cox-proportional hazards models. Hazard ratios (HRs) and their respective 95% confidence intervals (CIs) were reported. Incidence curves across 4 BMI-alcohol categories also were examined. Estimate of interaction on a multiplicative scale. Joint effects of alcohol and obesity were estimated using relative excess risk resulting from interaction (RERI), attributable risk proportion resulting from interaction (AP), and synergy index (SI), and their respective 95% CIs as previously described. 24,25 RERI is an estimate of excess risk that is directly attributable to the interaction between 2 exposures. AP is defined as the proportion of risk that is attributable to the interaction between the 2 exposure variables. SI is a ratio that estimates whether a synergistic (SI, 1) or antagonistic (SI, 1) interaction exists between 2 exposures. A multiplicative interaction is suggested by the following scores: an RERI greater than 1.5, an AP greater than 0.25, and an SI greater than Joint effects of body mass index and alcohol. Joint effects between WHO BMI categories and alcohol use (yes/no) were performed on hazards of HCC. Furthermore, joint effects of BMI WHO categories and duration of alcohol use ( 10, 10 19, and 20 y) were examined. Sensitivity analysis. Joint effects between BMI quartiles (1, lowest; through 4, highest), and alcohol use (yes/no) were analyzed using proportional hazards models with incident HCC as the outcome. In addition, joint effects of BMI quartiles and duration of alcohol use (no alcohol, 1 19 y, 20 y) was examined. A 2-tailed P value of.05 or less was considered statistically significant. All analyses were conducted using SAS statistical software (version 8.01; SAS Institute Inc, Cary, NC).

3 October 2010 OBESITY, ALCOHOL, AND HCC RISK 893 Results Baseline Characteristics A total of 2260 Taiwanese men who were HBsAg-positive were included in this analysis. The mean ( standard deviation) age and BMI were 45.8 ( 10) years and 23.9 ( 3) kg/m 2, respectively. A fifth of participants reported alcohol use and 53.8% of the population were classified as smokers. A majority (83.9%) of individuals were HBeAg-negative, the prevalence of increased ALT levels ( 45 U/L) was 7.6%, and only 2.7% of all participants had cirrhosis at baseline. Prevalence of cirrhosis was similar in normal weight, overweight, obese, and extremely obese participants. Table 1 shows the baseline characteristics of the patients included in the study. Risk Factors for Future Development of Hepatocellular Carcinoma There were a total of 135 cases of HCC over 41,778 person-years of follow-up evaluation. Median age (range) at Table 1. Baseline Characteristics of the Participants Included in the Study Characteristic 2260 men Age, y Mean standard deviation Median (25th percentile 75th percentile) 45 (37 54) (33.5) (27.4) (28.5) (10.6) BMI, kg/m 2 Mean standard deviation Median (25th percentile 75th percentile) 23.6 ( ) 23 (normal weight) 931 (41.3) 23 to 25 (overweight) 554 (24.6) 25 to 30 (obese) 684 (30.3) 30 (extremely obese) 88 (3.9) (96.1) (3.9) Cigarette smoking No 1044 (46.2) Yes 1214 (53.8) Alcohol consumption No 1808 (80.2) Yes 447 (19.8) HBeAg status Positive 363 (16.1) Negative 1897 (83.9) ALT level, U/L (55.7) (36.7) (7.6) HBV DNA, copies/ml Median (25th percentile 75th percentile) 10 3 ( ) 10, (69.9) 10, (30.1) Diabetes No 2200 (97.4) Yes 58 (2.6) Cirrhosis No 2198 (97.3) Yes 62 (2.7) diagnosis of HCC was 59 (36 76) years. We then conducted Cox-proportional hazards analysis to identify predictors of incident HCC in this cohort. The univariate analysis (Table 2) showed that age, alcohol use, HBeAg status, HBV-DNA levels, increased ALT levels, cirrhosis at baseline, and interaction between alcohol and BMI (P.029) were statistically significant predictors of incident HCC over 14 years of follow-up evaluation. Multivariate-adjusted analysis showed that age, alcohol use, HBeAg status, HBV-DNA levels, and cirrhosis at baseline were statistically significant independent predictors of incident HCC over 14 years of follow-up evaluation. Although the hazards of incident HCC were increased with BMI, smoking, increased ALT level, and interaction between BMI and alcohol, the results were not significant. The risk of incident HCC was highest in alcohol users with a BMI of 30 kg/m 2 or greater as shown in Figure 1. A total of 19.8% of the participants reported alcohol use. The incidence rate of HCC among alcohol users versus nonusers was 852 cases of HCC versus 253 cases of HCC per 100,000 personyears of follow-up evaluation, respectively (Table 2). Alcohol use was associated with increased risk of incident HCC both in unadjusted and multivariate adjusted models with an HR of 1.91 (95% CI, ; P.001) and an HR of 1.54 (95% CI, ; P.031), respectively (Table 2). Joint Effects of Obesity and Alcohol on Incident Hepatocellular Carcinoma Joint effects of alcohol use and extreme obesity (BMI, 30) showed that the risk of HCC was increased synergistically in alcohol users who had extreme obesity compared with those without extreme obesity and nonusers of alcohol (HR, 3.1; 95% CI, ; P.028) as shown in Figure 2. Results remained consistent in both age-adjusted (HR, 3.41; 95% CI, ; P.025) and multivariable-adjusted (HR, 3.40; 95% CI, ; P.025) analyses as shown in Table 3. The Wald test of interaction between BMI and alcohol use was statistically significant (P.029). RERI, AP, and SI were 1.59 (95% CI, 1.6 to 4.79), 0.52 (95% CI, 0.09 to 1.13), and 4.40 (95% CI, ), respectively, indicating the presence of a multiplicative interaction between obesity and alcohol use. A multiplicative interaction is suggested by the following scores: an RERI greater than 1.5, an AP greater than 0.25, and an SI greater than ,26 Figure 2 shows the cumulative incidence of HCC over 14 years of follow-up evaluation across the 4 BMI-alcohol categories. Incidence rates were higher in alcohol users with a BMI less than 30 kg/m 2 than nonusers of alcohol with a BMI of 30 kg/m 2 or greater. The lowest incidence was in nonusers who had a BMI of less than 30 kg/m 2. The HCC rate among nonusers of alcohol with a BMI of less than 30 kg/m 2 and a BMI of 30 kg/m 2 or greater were 5.2% and 3.1%, respectively, and the HCC rate among users of alcohol with a BMI of less than 30 kg/m 2 and a BMI of 30 kg/m 2 or greater were 9.1% and 14.9%, respectively. The incidence rate of HCC was significantly higher in alcohol users with a BMI of 30 kg/m 2 or greater as compared with the other 3 categories (P.002). Joint Effects of Body Mass Index Categories and Alcohol Use on Incident Hepatocellular Carcinoma We further explored at which level of BMI the hazards of incident HCC increase in alcohol users. Risk of HCC was

4 894 LOOMBA ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 10 Table 2. Unadjusted and Multivariate-Adjusted Risk of Incident HCC in Men Over 14 Years of Follow-Up Evaluation Variable No. of participants Person-years of follow-up evaluation No. of HCC cases Incidence rate per 100,000 person-years Crude HR (95% CI) P value Multivariable-adjusted HR (CI) P value Age in 1-year increments 1.07 ( ) ( ).001 BMI in 1-kg/m 2 increments 1.02 ( ) ( ).90 Cigarette smoking No , Referent Referent Yes , ( ) ( ).81 Alcohol consumption No , Referent Referent Yes ( ) ( ).031 HBeAg status Negative , Referent Referent Positive ( ) ( ).001 ALT level, U/L , Referent Referent ( ) ( ).76 HBV DNA, copies/ml 10, , Referent Referent 10, , ( ) ( ).001 Cirrhosis No , Referent Referent Yes ( ) ( ).001 Alcohol/BMI (interaction) 1.12 ( ) ( ).29 higher in users of alcohol than nonusers across all BMI categories but was not increased significantly in the normal weight category (Figure 3). However, the risk synergistically increased in overweight (HR, 2.4; 95% CI, ), obese (HR, 2.0; 95% CI, ), and extremely obese (HR, 2.9; 95% CI, ) users of alcohol (P value for trend.046). We also analyzed the joint association between BMI categories and duration of alcohol use and found that the risk of HCC was highest in extremely obese individuals who had 20 years or more of alcohol use (HR, 8.2; 95% CI, ) as shown in Supplementary Figure 1. The risk of HCC was higher in overweight or obese individuals with a history of 20 or more years of alcohol use and synergistically increased with higher BMI (P value for trend.006). The results remained statistically significant for the joint effect of extreme obesity and alcohol use in age-adjusted (HR, 3.07; 95% CI, ; P.04) and multivariate-adjusted models (HR, 3.21; 95% CI, ; P.03) as shown in Table 3. Similarly, the joint effect of extreme obesity and 20 years or more of alcohol use in age-adjusted (HR, 5.93; 95% CI, ; P.001) and multivariate-adjusted models (HR, 5.17; 95% CI, ; P.003) remained statistically significant as shown in Table 3. Sensitivity Analyses Joint effects of body mass index (quartiles) and alcohol use on incident hepatocellular carcinoma. Risk of HCC was higher in users of alcohol than nonusers across quartiles 2 to 4 (Supplementary Figure 2). Participants in the highest BMI quartile who consumed alcohol had a statistically significantly increased risk of incident HCC (HR, 2.5; 95% CI, Figure 1. Risk of incident HCC is 3.1 (95% CI, ) times higher in alcohol users who have a BMI of 30 kg/m 2 or greater as compared with nonusers of alcohol with a BMI of less than 30 kg/m 2, which is suggestive of multiplicative effect. This figure shows that obesity significantly augments the risk of incident HCC in alcohol users but not in nonusers. Figure 2. This figure shows the incidence curves of development of HCC in the 4 BMI-alcohol categories. The y-axis depicts the proportion of individuals in each group who are developing HCC (0.1 representing 10% and 0.2 representing 20% of participants). The risk of incident HCC was highest in alcohol users with a BMI of 30 kg/m 2 or greater.

5 October 2010 OBESITY, ALCOHOL, AND HCC RISK 895 Table 3. Age-Adjusted and Multivariable-Adjusted Models Showing the Joint Association Between Alcohol Use and BMI With Incident HCC Over 14 Years of Follow-Up Evaluation Variable Age-adjusted HR (95% CI) P value Multivariate-adjusted HR (95% CI) a P value BMI/alcohol use 30 kg/m 2 /no 30 kg/m 2 /no 0.74 ( ) ( ) kg/m 2 /yes 1.80 ( ) ( ) kg/m 2 /yes 3.41 ( ) ( ).0176 WHO BMI categories/alcohol use b Normal weight/no Overweight/no 0.76 ( ) ( ).3713 Obese/no 0.89 ( ) ( ).8585 Extremely obese/no 0.67 ( ) ( ).5167 Normal weight/yes 1.05 ( ) ( ).6593 Overweight/yes 2.38 ( ) ( ).003 Obese/yes 1.75 ( ) ( ).3106 Extremely obese/yes 3.07 ( ) ( ).0275 WHO BMI categories/year of alcohol consumption Normal weight/0 y Overweight/0 y 0.76 ( ) ( ).394 Obese/0 y 0.89 ( ) ( ).8515 Extremely obese/0 y 0.67 ( ) ( ).5027 Normal weight/1 19 y 1.57 ( ) ( ).1882 Overweight/1 19 y 2.35 ( ) ( ).2899 Obese/1 19 y 1.61 ( ) ( ).3514 Extremely obese/1 19 y Normal weight/ 20 y 0.64 ( ) ( ).6143 Overweight/ 20 y 1.81 ( ) ( ).0811 Obese/ 20 y 1.88 ( ) ( ).4858 Extremely obese/ 20 y 5.93 ( ) ( ).0023 BMI categories/drinker c Quartile 1/no Quartile 2/no 0.93 ( ) ( ).4352 Quartile 3/no 1.09 ( ) ( ).5666 Quartile 4/no 0.72 ( ) ( ).5084 Quartile 1/yes 1.15 ( ) ( ).4477 Quartile 2/yes 1.81 ( ) ( ).1275 Quartile 3/yes 1.66 ( ) ( ).6981 Quartile 4/yes 2.39 ( ) ( ).0075 BMI categories/year of alcohol consumption Quartile 1/0 y Quartile 2/0 y 0.93 ( ) ( ).4319 Quartile 3/0 y 1.09 ( ) ( ).5542 Quartile 4/0 y 0.72 ( ) ( ).5127 Quartile 1/1 19 y 1.87 ( ) ( ).2204 Quartile 2/1 19 y 1.53 ( ) ( ).3209 Quartile 3/1 19 y 2.23 ( ) ( ).3579 Quartile 4/1 19 y 1.30 ( ) ( ).7334 Quartile 1/ 20 y 0.76 ( ) ( ).8435 Quartile 2/ 20 y 1.47 ( ) ( ).6912 Quartile 3/ 20 y 1.30 ( ) ( ).6244 Quartile 4/ 20 y 3.03 ( ) ( ).0012 NOTE. A 2-tailed P value of less than.05 was considered statistically significant (bolded). a Adjustment for age, serum ALT level, serum HBV DNA level, and cirrhosis at entry. b WHO BMI categories: normal weight is BMI less than 23 kg/m 2, overweight is BMI of 23 to less than 25 kg/m 2, obese includes BMI of 25 to less than 30 kg/m 2, and extremely obese includes BMI of 30 kg/m 2 or greater. c Participants are classified into quartiles of BMI: quartile 1 (lowest BMI: referent group) and quartile 4 (highest BMI category) ; P.0042). We also analyzed the joint association between BMI (classified into quartiles) and duration of alcohol use and found that the risk of HCC was highest in the patients in the highest BMI quartile (quartile 4) who drank more than 20 years (HR, 4.5; 95% CI, ) as shown in Supplementary Figure 3. The risk of HCC was higher in quartiles 2, 3, and 4 with a history of 20 or more years of alcohol use and synergistically increased with a higher BMI (P for trend.027).

6 896 LOOMBA ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 10 Figure 3. Participants are classified into weight categories suggested by WHO for Asian populations. Normal weight is BMI less than 23 kg/m 2, overweight is BMI of 23 to less than 25 kg/m 2, obese includes BMI of 25 to less than 30 kg/m 2, and extremely obese includes BMI of 30 kg/m 2 or greater. This figure shows that risk of incident HCC increases as BMI dose increases only in alcohol users (P value for trend.046). The joint association between highest BMI quartile and 20 or more years of alcohol use remained statistically significant in both age-adjusted (HR, 3.03; 95% CI, , P.0025) and multivariable-adjusted analyses (HR, 3.29; 95% CI, ; P.0015) as shown in Table 3. Discussion Main Findings In this prospective population-based cohort of HBsAgpositive Asian men residing in Taiwan, we showed an interaction between BMI and alcohol use with the risk of incident HCC. We found that extreme obesity (BMI, 30 kg/m 2 ) and alcohol consumption synergistically increase the risk of incident HCC over 14 years of follow-up evaluation. This study showed that the interaction between obesity and alcohol with risk of incident HCC is multiplicative. Our findings show that the risk of HCC among alcohol users is highest among the extremely obese, and this risk decreases as BMI approaches the normal weight category, suggesting that prevention of obesity and minimizing alcohol use among patients with hepatitis B may reduce the burden of HCC worldwide. In Context With Published Literature Previous epidemiologic studies provide a significant association between male sex, Asian ethnicity, obesity, higher alcohol use, increased ALT, HBeAg status, HBV-DNA levels, and presence of cirrhosis as important risk factors for future development of HCC in patients with or without viral hepatitis. 10,11,27,28 These risk factors were also significant predictors of HCC in this cohort. Hepatitis B infection is the leading cause of HCC worldwide. Both obesity and alcohol use are independent and modifiable risk factors for fatty liver disease as well as future development of HCC. 29 Because of the global obesity and fatty liver epidemic, the joint effect of obesity and alcohol on future risk of HCC is of important public health significance. El-Serag 30 has shown that the incidence and mortality from HCC are increasing and will continue to increase in the next 2 decades. This increase in the burden of HCC in the United States parallels the increasing rates of obesity. 4,31 It is possible that increasing rates of obesity may further fuel the increase of increasing HCC rates. Mechanism It is possible that obesity and alcohol synergistically augment cytokine balance within the liver and increase the risk of HCC in predisposed individuals. Both obesity and alcohol lead to oxidative stress and progressive liver injury via multiple inter-related metabolic pathways such as cytochrome P450 2E1 induction, free radical generation, lipid peroxidation, nuclear factor- B activation, and increased transcription of proinflammatory mediators, including tumor necrosis factor- 32,33 Dietinduced obesity is known to accentuate the progression of alcoholic liver injury via free radical damage, increasing tumor necrosis factor- production, and worsening insulin resistance. Some of the earlier-mentioned pathways such as nuclear factor- B, tumor necrosis factor-, and interleukin-6 have been directly implicated in tumorigenesis and perhaps may be involved in the pathogenesis of HCC. 34 Our study provides epidemiologic evidence to pursue these hypotheses. Obesity and alcohol may increase the risk of both alcoholic liver disease and NAFLD. Emerging data suggest that NAFLD increases the risk of HCC. Marrero et al 13 have shown that NAFLD is present in 11% of HCC patients. Previous studies have suggested that HCC cases in the setting of cryptogenic cirrhosis may in fact be related to NAFLD because cryptogenic cirrhosis may represent burnt out NASH in a subset of patients. 15 A Danish cohort study including 7326 patients discharged from the hospital with a diagnosis of fatty liver showed that patients with fatty liver had a significantly increased standardized incidence ratio of A recent Japanese study estimated the risk of HCC in NASH patients over 5 years of follow-up evaluation to be 7.6%. 16 Therefore, the epidemiologic link between NAFLD and HCC is becoming stronger as newer data are emerging. Strengths and Limitations Strengths of our study included prospective follow-up evaluation over 14 years, large sample size, body mass index recorded by a trained investigator and not by self-report, availability of known risk factors of HCC at baseline for adjustment in multivariate analyses, and a well-characterized cohort. In addition, previous findings derived from this cohort have been generalizable to other populations, and have been reproduced in independent cohorts, and well accepted by independent investigators. However, we acknowledge the following limitations of our study, which include that alcohol use was based on self-report. Lifetime alcohol use and binge drinking could not be quantified accurately. This is likely to underestimate the quantity of alcohol use and potentially could lead to misclassification bias. However, underreporting and misclassification (heavier alcohol users classified as nonusers) is likely to favor a null finding (no effect of alcohol use on HCC risk) rather than a synergistic effect of the alcohol exposure and obesity. Therefore, it is possible that our study may underestimate the synergism between obesity and alcohol use. However, it can be safely concluded based on data from a previous case-control study by Marrero et al 36 and our prospective cohort study that obesity and alcohol use are synergistic. The prevalence of obesity is lower in this population and the relative increase in HCC risk in obese versus nonobese (especially in nonusers of alcohol) is not statistically significant among Taiwanese men with hepatitis B. However, alcohol appears to be particularly harmful in obese versus nonobese individuals, suggesting a double hit in

7 October 2010 OBESITY, ALCOHOL, AND HCC RISK 897 increasing the susceptibility to develop steatohepatitis that may multiply the risk of HCC in obese drinkers. Because women were not included in this analysis, further studies are needed to address whether the synergism between alcohol and obesity is sex-specific or independent. We were not able to examine the effect of obesity and alcohol in women because both the prevalence and quantity of alcohol use among Taiwanese HBsAgpositive women were extremely low. Therefore, women were excluded from these analyses. It remains to be determined whether the observed interaction between obesity and alcohol also is seen in patients with chronic hepatitis C because our study population did not include patients with hepatitis C virus infection. Future studies are needed in a large multi-ethnic cohort to accurately quantify the effects of obesity and alcohol on HCC risk in participants with or without viral hepatitis. Conclusions This study provides epidemiologic evidence of a multiplicative interaction between alcohol use and obesity in increasing HCC risk in HBsAg-positive men. These findings suggest that lifestyle intervention programs that encourage weight loss in obese or overweight in high risk individuals and endorse abstaining from alcohol use can be expected to lower the risk of future development of HCC. Given the expected increase in HCC incidence over the next 2 decades, public health interventions are needed to reduce the burden of HCC in endemic areas. Supplementary Material Note: to access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at and at doi: / j.cgh References 1. WHO. The global burden of disease: 2004 update. In: Geneva S, ed. Health statistics and informatics department WHO. Geneva: WHO Press, Available at: bod. Accessed August 23, El-Serag HB, Mason AC. Rising incidence of hepatocellular carcinoma in the United States. N Engl J Med 1999;340: El-Serag HB, Davila JA, Petersen NJ, et al. The continuing increase in the incidence of hepatocellular carcinoma in the United States: an update. Ann Intern Med 2003;139: Everhart JE, Ruhl CE. Burden of digestive diseases in the United States Part III: liver, biliary tract, and pancreas. Gastroenterology 2009;136: Altekruse SF, McGlynn KA, Reichman ME. Hepatocellular carcinoma incidence, mortality, and survival trends in the United States from 1975 to J Clin Oncol 2009;27: Chen CL, Yang HI, Yang WS, et al. Metabolic factors and risk of hepatocellular carcinoma by chronic hepatitis B/C infection: a follow-up study in Taiwan. Gastroenterology 2008;135: Calle EE, Rodriguez C, Walker-Thurmond K, et al. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med 2003;348: Larsson SC, Wolk A. Overweight, obesity and risk of hepatocellular carcinoma: a meta-analysis of cohort studies. Br J Cancer 2007;97: Kuper H, Tzonou A, Kaklamani E, et al. Tobacco smoking, alcohol consumption and their interaction in the causation of hepatocellular carcinoma. Int J Cancer 2000;85: Jee SH, Ohrr H, Sull JW, et al. Cigarette smoking, alcohol drinking, hepatitis B, and risk for hepatocellular carcinoma in Korea. J Natl Cancer Inst 2004;96: Ikeda K, Marusawa H, Osaki Y, et al. Antibody to hepatitis B core antigen and risk for hepatitis C-related hepatocellular carcinoma: a prospective study. Ann Intern Med 2007;146: Yu MW, Hsu FC, Sheen IS, et al. Prospective study of hepatocellular carcinoma and liver cirrhosis in asymptomatic chronic hepatitis B virus carriers. Am J Epidemiol 1997;145: Marrero JA, Fontana RJ, Su GL, et al. NAFLD may be a common underlying liver disease in patients with hepatocellular carcinoma in the United States. Hepatology 2002;36: El-Serag HB, Rudolph KL. Hepatocellular carcinoma: epidemiology and molecular carcinogenesis. Gastroenterology 2007;132: Bugianesi E, Leone N, Vanni E, et al. Expanding the natural history of nonalcoholic steatohepatitis: from cryptogenic cirrhosis to hepatocellular carcinoma. Gastroenterology 2002;123: Hashimoto E, Yatsuji S, Tobari M, et al. Hepatocellular carcinoma in patients with nonalcoholic steatohepatitis. J Gastroenterol 2009;44(Suppl 19): Loomba R, Bettencourt R, Barrett-Connor E. Synergistic association between alcohol intake and body mass index with serum alanine and aspartate aminotransferase levels in older adults: the Rancho Bernardo Study. Aliment Pharmacol Ther 2009;30: Yang HI, Lu SN, Liaw YF, et al. Hepatitis B e antigen and the risk of hepatocellular carcinoma. N Engl J Med 2002;347: Chen CJ, Yang HI, Su J, et al. Risk of hepatocellular carcinoma across a biological gradient of serum hepatitis B virus DNA level. JAMA 2006;295: WHO Expert Consultation. Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet 2004;363: Wang LY, You SL, Lu SN, et al. Risk of hepatocellular carcinoma and habits of alcohol drinking, betel quid chewing and cigarette smoking: a cohort of 2416 HBsAg-seropositive and 9421 HBsAgseronegative male residents in Taiwan. Cancer Causes Control 2003;14: Lin DY, Sheen IS, Chiu CT, et al. Ultrasonographic changes of early liver cirrhosis in chronic hepatitis B: a longitudinal study. J Clin Ultrasound 1993;21: Chen CJ, Iloeje UH, Yang HI. Long-term outcomes in hepatitis B: the REVEAL-HBV study. Clin Liver Dis 2007;11: , viii. 24. Rothman KJ, Greenland S, Walker AM. Concepts of interaction. Am J Epidemiol 1980;112: Richardson DB, Kaufman JS. Estimation of the relative excess risk due to interaction and associated confidence bounds. Am J Epidemiol 2009;169: de Mutsert R, Jager KJ, Zoccali C, et al. The effect of joint exposures: examining the presence of interaction. Kidney Int 2009;75: Di Bisceglie AM, Lyra AC, Schwartz M, et al. Hepatitis C-related hepatocellular carcinoma in the United States: influence of ethnic status. Am J Gastroenterol 2003;98: El-Serag HB, Mason AC. Risk factors for the rising rates of primary hepatocellular carcinoma in the United States. Arch Intern Med 2000;160: Donato F, Tagger A, Gelatti U, et al. Alcohol and hepatocellular carcinoma: the effect of lifetime intake and hepatitis virus infections in men and women. Am J Epidemiol 2002;155: El-Serag HB. Hepatocellular carcinoma: recent trends in the United States. Gastroenterology 2004;127:S27 S Ogden CL, Carroll MD, Curtin LR, et al. Prevalence of overweight and obesity in the United States, JAMA 2006;295: Purohit V, Brenner DA. Mechanisms of alcohol-induced hepatic fibrosis: a summary of the Ron Thurman Symposium. Hepatology 2006;43:

8 898 LOOMBA ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No Siegmund SV, Brenner DA. Molecular pathogenesis of alcoholinduced hepatic fibrosis. Alcohol Clin Exp Res 2005;29:102S 109S. 34. Naugler WE, Karin M. The wolf in sheep s clothing: the role of interleukin-6 in immunity, inflammation and cancer. Trends Mol Med 2008;14: Sorensen HT, Mellemkjaer L, Jepsen P, et al. Risk of cancer in patients hospitalized with fatty liver: a Danish cohort study. J Clin Gastroenterol 2003;36: Marrero JA, Fontana RJ, Fu S, et al. Alcohol, tobacco and obesity are synergistic risk factors for hepatocellular carcinoma. J Hepatol 2005;42: Reprint requests Address requests for reprints to: Rohit Loomba, MD, MHSc, Division of Gastroenterology and Division of Epidemiology, University of California at San Diego, 9500 Gilman Drive, MC 0063, La Jolla, California roloomba@ucsd.edu; fax: (858) ; or Chien-Jen Chen, ScD, Genomics Research Center, Academia Sinica, 128 Academia Road, Section 2, Nankang, Taipei 115, Taiwan. fax: (886) Conflicts of interest The authors disclose no conflicts. Funding This work is supported in part by the American Gastroenterological Association Foundation Sucampo Association of Specialty Professors Designated Research Award in Geriatric Gastroenterology and by a T. Franklin Williams Scholarship Award. Funding also provided by Atlantic Philanthropies, Inc, the John A. Hartford Foundation, the Association of Specialty Professors, and the American Gastroenterological Association (R.L.). The REVEAL-HBV study was supported by a grant from the Department of Health, Academia Sinica, National Health Research Institutes in Taiwan, and Bristol-Myers Squibb Company, Wallingford, CT.

9 898.e1 LOOMBA ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 10 Supplementary Figure 1. Joint effect of BMI categories and duration of alcohol use with risk of HCC. Participants are classified into weight categories suggested by WHO for Asian populations. Normal weight is BMI less than 23 kg/m 2, overweight is BMI of 23 to less than 25 kg/m 2, obese includes BMI of 25 to less than 30 kg/m 2, and extremely obese includes BMI of 30 kg/m 2 or greater. This figure shows that risk of incident HCC increases as BMI dose increases in individuals with 20 or more years of alcohol use (P value for trend.006). Within each weight category, the risk of HCC increased with greater duration of alcohol use as well. Supplementary Figure 2. Joint effect of BMI quartiles and alcohol use with risk of HCC in men. Participants are classified into quartiles of BMI: quartile 1 (lowest BMI: referent group) and quartile 4 (highest BMI category). This figure shows that risk of incident HCC increases as BMI dose increases only in alcohol users (P value for trend.08).

10 October 2010 OBESITY, ALCOHOL, AND HCC RISK 898.e2 Supplementary Figure 3. Joint effect of BMI quartiles and duration of alcohol use with risk of HCC in men. Participants are classified into quartiles of BMI: quartile 1 (lowest BMI: referent group) and quartile 4 (highest BMI category). This figure shows that risk of incident HCC increases as BMI dose increases in individuals with 20 or more years of alcohol use (P value for trend.027). Within each weight category, the risk of HCC increased with greater duration of alcohol use except in the lowest quartile in which the effect of modification was clinically not significant.