Risk and Predictors of Mortality Associated With Chronic Hepatitis B Infection

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2007;5: Risk and Predictors of Mortality Associated With Chronic Hepatitis B Infection UCHENNA H. ILOEJE,* HWAI I YANG, CHIN LAN JEN, JUN SU,* LI YU WANG, SAN LIN YOU, and CHIEN JEN CHEN, FOR THE RISK EVALUATION OF VIRAL LOAD ELEVATION AND ASSOCIATED LIVER DISEASE/CANCER HEPATITIS B VIRUS STUDY GROUP *Global Epidemiology and Outcomes Research, Pharmaceutical Research Institute, Bristol-Myers Squibb Company Wallington, Connecticut; Genomic s Research Center, Academia Sinica and Graduate Institute of Epidemiology, College of Public Health, National Taiwan University, Taipei, Taiwan; and the Institute of Aboriginal Health, Tzu-Chi University, Hualien City, Taiwan Background & Aims: The study objective was to determine the risk of all-cause and cause-specific mortality as well as to examine the predictors of mortality in chronic hepatitis B infection. Methods: We performed a prospective cohort study of 23,820 persons (age, y) recruited between 1991 and 1992 and followed up through 2004 from 7 townships in Taiwan. The main outcomes were all-cause and liver-related mortality rates. Mortality analyses used time-to-events methods, and survival curves were derived by the Kaplan Meier method. Cox proportional hazard models were used to estimate multivariable-adjusted hazard ratios. Results: There were 1814 deaths during a mean follow-up period of 12.5 years (282,323.7 person-years of follow-up evaluation). Persons positive for hepatitis B surface antigen (HBsAg) had significantly (P <.01) higher adjusted hazard ratios for all causes of mortality (1.7; 95% confidence interval [CI], ), liver cancer mortality (22.4; 95% CI, ), and chronic liver disease and cirrhosis mortality (5.4; 95% CI, ). When compared with HBsAg-negative persons, hepatitis B virus (HBV)- infected persons with HBV DNA levels less than 10 4 had a high risk of hepatocellular carcinoma mortality (4.4; 95% CI, ). In HBsAg-positive persons, the mortality rate increased with cohort entry serum HBV DNA level. Liver cancer mortality ranged from 72.8 per 100,000 person-years for subjects with HBV DNA levels less than 300 copies/ml to per 100,000 person-years for those with HBV DNA levels of 1 million copies/ml or greater. Chronic liver disease and cirrhosis deaths ranged from 9.1 to per 100,000 person-years. Conclusions: Chronic HBV infection is associated with significant preventable excess mortality risk. This mortality risk is correlated strongly with the level of viral replication among other factors. Chronic hepatitis B (CHB) infection accounts for well over 1 million deaths each year and represents the 10th leading global cause of death. 1 Hepatitis B virus (HBV) infection accounts for the vast majority of chronic liver diseases in endemic areas such as Taiwan (75% 80%) and mainland China (73%). 2 Moreover, CHB infection accounts for 60% 80% of primary liver cancer globally, which is 1 of the 3 major causes of death in Asia, the Pacific Rim, and Africa. 1 Cirrhosis develops in approximately 20% of CHB-infected persons, and its complications, such as hepatic insufficiency and portal hypertension, contribute significantly to CHB-related morbidity and mortality. 3 Besides these well-known causes of morbidity and mortality, CHB infection also has been associated with an increased risk of lymphomas in different studies from Australia, the United Kingdom, and the United States of America. 4 6 Several studies have addressed the mortality associated with chronic HBV infection, and survival rates for patients with CHB-related liver disease have been established from these studies. 7 9 A question still remaining is which CHB-infected patients are at greatest risk of mortality as a result of liverrelated complications. There is also no information on the risk of mortality in CHB-infected patients with low viral load (HBV DNA level 10 4 copies/ml) in comparison with uninfected persons. The Risk Evaluation of Viral Load Elevation and Associated Liver Disease/Cancer (REVEAL)-HBV study is a large, prospective, community-based, cohort study of Taiwanese people as previously described. 10,11 For these mortality analyses, the 22,472 enrolled subjects who were known not to be hepatitis C virus (HCV) infected were studied. Data linkage to the Taiwanese national death certification system was conducted from January 1, 1991, through December 31, 2004, with a mean follow-up time of 12.5 years and 282,323.7 person-years of observation. In these analyses, we report on the following: (1) the common causes of mortality in this cohort, (2) the difference in mortality between the CHB-infected persons and those without evidence of CHB infection, and (3) the predictors of mortality in the CHB-infected subjects. Methods Cohort Recruitment and Follow-up Evaluation The REVEAL-HBV study was conducted in 7 townships in Taiwan as part of a community-based cancer screening program. Between 1991 and 1992, there were 89,293 ethnically Abbreviations used in this paper: ALT, alanine aminotransferase; CHB, chronic hepatitis B; CI, confidence interval; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; PYFU, personyears of follow-up; REVEAL, Risk Evaluation of Viral load Elevation and Associated Liver Disease/Cancer by the AGA Institute /07/$32.00 doi: /j.cgh

2 922 ILOEJE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 8 provided assistance in securing referrals for subjects requiring assistance. In addition, anyone with suspected hepatocellular carcinoma (HCC) was referred directly by the study team to their hospital of choice. In addition, on the availability of reimbursement for HBV therapy in October of 2003, subjects meeting the established treatment guidelines were referred to the National Taiwan University Hospital for evaluation. The study protocol was reviewed and approved by the Institutional Review Board of the College of Public Health at National Taiwan University. Table 1. Demographics of the Study Cohort HBsAg seronegative (n 18,541) N(%) HBsAg seropositive (n 3931) N(%) Total (n 22,472) N(%) Figure 1. Study design and flow chart. Chinese individuals aged years who were invited to participate, and 23,820 individuals agreed and gave written informed consent. Demographic information on 63,454 of the 65,463 subjects (97%) not participating in this study show that they were quite similar except in educational level; there were more people with low-level formal education among those enrolling in the study (21% vs 6%). The gender distribution was similar (49% vs 46% women), the age distribution was slightly different (there were 29% enrolled within the age group, vs 40% nonenrolled in the age group), and the distribution of subjects across the townships was quite similar in both groups. Subjects enrolled in the study were interviewed by trained public health nurses, received a health examination that included abdominal ultrasonography, and had blood collected for serologic, biochemical, and HBV DNA testing. Active follow-up evaluation in hepatitis B surface antigen (HBsAg)-positive persons included planned ultrasonography every 6 12 months, and serum was collected for storage as well as additional serologic and biochemical testing. Vital status information was obtained from death certificates filed with the local household registration systems and were cross-checked with the computerized records at the National Death Certification Profiles as described later. To ensure that subjects enrolled in the study were receiving the appropriate standard of care, subjects with evidence of a biochemical liver abnormality were sent a letter with the test results and instructed to see their physician. The study team Female 9481 (51.1) 1600 (40.7) 11,081 (49.3) Male 9060 (48.9) 2331 (59.3) 11,391 (50.7) Age, y (28.6) 1297 (33.0) 6594 (29.3) (26.2) 1091 (27.8) 5943 (26.5) (30.3) 1142 (29.1) 6766 (30.1) (14.9) 401 (10.2) 3169 (14.1) Residential area: township (county) Chutung (Hsinchu) 4324 (23.3) 858 (21.8) 5182 (23.1) Kaoshu (Pingtung) 3228 (17.4) 569 (14.5) 3797 (16.9) Potzu (Chiayi) 2715 (14.6) 633 (16.1) 3348 (14.9) Sanchih (Taipei) 1022 (5.5) 222 (5.7) 1244 (5.5) Huhsi (Penghu) 1491 (8.0) 278 (7.1) 1769 (7.9) Makung (Penghu) 5083 (27.4) 1089 (27.7) 6172 (27.5) Paisha (Penghu) 678 (3.7) 282 (7.2) 960 (4.3) Educational level a Uneducated 3967 (21.4) 693 (17.6) 4660 (20.8) Primary school 7712 (41.6) 1565 (39.8) 9277 (41.3) Junior high school 2521 (13.6) 615 (15.6) 3136 (14.0) Senior high school 2787 (15.0) 682 (17.4) 3469 (15.4) Vocational school 1019 (5.5) 259 (6.6) 1278 (5.7) College or graduate 525 (2.8) 117 (3.0) 642 (2.9) school Cigarette smoking b No 13,270 (71.7) 2647 (67.4) 15,917 (71.0) Yes 5236 (28.3) 1281 (32.6) 6517 (29.1) Alcohol consumption c No 16,544 (89.4) 3455 (88.1) 19,999 (89.2) Yes 1953 (10.6) 469 (12.0) 2422 (10.8) HBV DNA, copies/ml d (23.9) (10.2) (21.6) 10,000 99,999 NA 643 (17.6) NA 100, , (9.6) million 154 (4.2) million 100 (2.7) 100 million 373 (10.2) a There were 10 HBsAg-negative persons whose educational level was not available. b Data were not available for 38 participants. c Data were not available for 51 participants. d Data were not available for 278 participants.

3 August 2007 HEPATITIS B VIRUS INFECTION AND MORTALITY 923 Table 2. All-Cause and Cause-Specific Mortality (n 22,472) HBsAg seronegative (n 18,541; 233,645.4 HBsAg seropositive (n 3931; 48,678.2 Cause of death No. of deaths No. of deaths All causes Infectious and parasitic diseases Neoplasms Malignant neoplasms Neoplasms of uncertain behavior Endocrine, nutritional, and metabolic diseases, and immunity disorders Diabetes mellitus Disease of blood and blood-forming organs Mental disorders Diseases of the nervous system and sense organs Disease of the circulatory system Hypertensive disease Ischemic heart disease Other forms of heart disease Cerebrovascular disease Diseases of the respiratory system Pneumonia and influenza Chronic obstructive pulmonary disease and allied conditions Diseases of the digestive system Chronic liver disease and cirrhosis Diseases of the genitourinary system Disease of the skin and subcutaneous tissue Diseases of the musculoskeletal system and connective tissue Congenital anomalies Symptoms, signs, and ill-defined conditions Injury and poisoning Laboratory Examinations Serologic testing was performed as follows: HBsAg and hepatitis B e antigen (HBeAg) levels by radioimmunoassay with commercial kits (Abbott Laboratories, North Chicago, IL); anti- HCV levels by enzyme immunoassay using second-generation commercial kits (Abbott Laboratories); alanine aminotransferase (ALT) levels by serum chemistry autoanalyzer (Model 736; Hitachi Co., Tokyo, Japan) using commercial reagents (Biomerieux, Marcy L Etoile, France); and serum HBV DNA levels by polymerase chain reaction using commercial tests (Roche Diagnostics, Co., Indianapolis, IN). The HBV DNA level was measured using the COBAS AMPLICOR HBV MONITOR (Roche Diagnostics, Co) test kit, an in vitro nucleic acid amplification test for quantifying HBV DNA on the COBAS AMPLI- COR analyzer, with a certified lower limit of detection of 300 copies/ml. At this level of HBV DNA, the test has a 98.1% positivity rate and a 100% clinical specificity rate. The procedures for sample preparation, storage, and testing as outlined by the manufacturer were adhered to. Briefly, whole blood was collected with a vacuum blood collection tube without anticoagulant, separated by centrifugation at room temperature, and the serum was stored in a sterile tube at 70 C. This procedure was accomplished within 6 hours of sample collection. Each sample run for HBV DNA included a replicate of the negative, low-positive and high-positive controls, and each run was determined to be valid. Linkage Procedure for Mortality Data In Taiwan, all deaths are registered through a nationwide system of offices called the household registration offices. These records subsequently are double-checked annually by registration officers and, where available, autopsy data are used to verify the diagnoses. The rate of autopsies in this study cohort was very low at less than 5%. The death certificates then are coded by the National Department of Health and these records became computerized in To ascertain the cause of death of study participants, the death certificates were obtained from their local household registration offices. To ensure complete ascertainment of the mortality information, data linkage was performed with the National Death Certification Profiles from January 1, 1991, to December 31, The data linkage between the REVEAL-HBV study cohort and the National Death Certification Profiles was performed using the unique national ID number that is assigned to every Taiwanese at birth. The linked data was validated further by comparing the birthdays in both data sets. Data linkage was performed using SAS software version 8.01 (SAS Institute Inc., Cary, NC). The classification of cause of death was based on the International Statistical Classification of Diseases, Injuries and Causes of Death (Ninth Revision). Code 155 referred to death from malignant neoplasm of the liver and intrahepatic bile ducts, and code 571 indicated death from chronic liver disease and cirrhosis. As previously reported, subjects with known or suspected HCC at the time of recruitment were excluded from enrollment in this study.

4 924 ILOEJE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 8 Figure 2. Total mortality according to HBsAg serostatus (n 22,472). Statistical Analysis Overall mortality analyses used time-to-events methods and were based on 282,323.7 person-years of follow-up evaluation. The person-years of follow-up evaluation were calculated from the date of recruitment to death, or the last date of follow-up data (December 31, 2004). s were calculated by dividing the number of cases of death by the personyears of follow-up evaluation, and survival curves by follow-up year were derived by the Kaplan Meier method. The log-rank test was used to compare survival curves. Comparisons were made between volunteers who were HBsAg positive and those who were HBsAg negative by using the Cox proportional hazards models to estimate multivariableadjusted hazard ratios and their 95% confidence intervals (CIs). Adjustments were made for sex, age, cigarette smoking, and alcohol consumption. Statistical significance was determined by 2-tailed tests (P.05). The attributable risk percentage was calculated by the following formula: Mortality risk in exposed mortality risk in unexposed mortality risk in exposed 100 It is the proportion of disease occurrence in the group with risk factors that potentially would be eliminated if the risk factor was not present. The population attributable risk percentage was calculated by the following formula: Mortality risk for entire population mortality risk for unexposed mortality risk for entire population 100 Table 3. Risks of Total Mortality and Cause-Specific Mortalities Comparing the HBsAg-Positive With HBsAg-Negative Group Causes of death No. of deaths HBsAg negative (n 18,541; 233,645.4 HBsAg positive (n 3931; 48,678.2 No. of deaths Adjusted hazard ratio for HBsAg-positivity vs HBsAg negativity (95% CI) a Attributable risk percent Population attributable risk percent All causes ( ) b 41.2% 10.9% Liver cancer ( ) b 95.5% 78.8% Chronic liver disease and ( ) b 81.5% 43.4% cirrhosis Causes other than liver cancer, cirrhosis, and chronic liver disease ( ) 0.0% 0.0% a Adjusted for sex, age, cigarette smoking, and alcohol drinking. b P.01.

5 August 2007 HEPATITIS B VIRUS INFECTION AND MORTALITY 925 Table 4. Risks of Total Mortality and Cause-Specific Mortalities Comparing HBsAg-Positive Subjects With Low Viral Replication (HBV DNA 10 4 copies/ml) With HBsAg-Negative Group HBsAg negative (n 18,541; 233,645.4 HBsAg positive with HBV DNA 10,000 c/ml (n 2034; 25,633.3 Causes of death No. of deaths No. of deaths Adjusted hazard ratio for HBV DNA 10,000 c/ml vs HBsAg negativity (95% CI) a All causes ( ) Liver cancer ( ) b Chronic liver disease and cirrhosis ( ) Causes other than liver cancer, cirrhosis, and chronic liver disease ( ) NOTE. Subjects known to be anti-hcv antibody positive were removed. a Adjusted for sex, age, cigarette smoking, and alcohol drinking. b P.01. It is the proportion of disease occurrence in the general population that would be eliminated if the risk factor was not present. s for all causes, liver cancer, chronic liver disease and cirrhosis, and non liver-related causes in this subgroup of subjects were derived and in addition stratified by sex, age, cigarette smoking, alcohol consumption, HBeAg status, serum ALT level, cirrhosis at entry, and serum HBV DNA level. Cox proportional hazards models were used to estimate univariate and multivariable-adjusted hazard ratios. Survival curves by follow-up year with stratification by HBV DNA level were derived by the Kaplan Meier method. The assumption of proportionality was tested for the Cox regression analyses and no violation of this assumption was observed. Colinearity also was tested, especially between certain variables (cigarette smoking and alcohol consumption; HBV DNA and HBeAg status) and no significant changes in the observed hazard ratios were seen. Results Demographics Figure 1 shows the patient flow through the REVEAL- HBV study and identifies the different subpopulations that were used for the 3 different study objectives. Of the 23,820 subjects, anti-hcv antibody testing was completed in all except 35 subjects. All anti-hcv antibody positive subjects (n 1313) and the subjects in whom this was not tested (n 35) were removed from the analyses, leaving a cohort of 22,472 subjects. Of these, 18,541 (82.5%) were HBsAg negative and 3931 (17.5%) were HBsAg positive. Table 1 provides demographic information. The 2 groups (HBsAg positive and HBsAg negative) were fairly balanced for age, residential area, and education level. A higher proportion of the HBsAg-positive group was male (59.3%) compared with the HBsAg-negative group (48.9%). Minor differences existed in the age distribution with 14.9% of the HBsAg-negative group being older than 60 years compared with 10.2% in the HBsAg-positive group. For those with available data on cigarette smoking and alcohol consumption, 15,917 participants (71%) reported no smoking history and 19,999 (89.2%) reported never drinking alcohol. The HBV DNA viral load distribution in the subcohort of HBsAg-positive and anti-hcv antibody negative individuals (n 3653) also is shown in Table 1. Approximately 17% of subjects had a viral load of 1 million copies/ml or greater. All-Cause and Cause-Specific Mortality Rates All-cause and cause-specific mortality rates for the entire study cohort by HBsAg status (n 22,472) are presented in Table 2. There were 1814 deaths over 282,323.7 person-years of follow-up (PYFU) for an overall death rate of 642.5/100,000 PYFU. The all-cause mortality rate in the HBsAg-seropositive individuals was higher, as was mortality from neoplasm and diseases of the digestive system. The top 3 causes of mortality by HBsAg serostatus were as follows: neoplasm (439.6), diseases of the circulatory system (117.1), and diseases of the digestive system (117.1) in HBsAg-positive individuals; neoplasm (188.7), diseases of the circulatory system (133.5), and injury and poisoning (76.2) in HBsAg-negative individuals. Mortality in Hepatitis B Surface Antigen Positive Versus Hepatitis B Surface Antigen Negative Subjects As shown in Figure 2, the overall survival for HBsAgpositive subjects was significantly worse than that of HBsAgnegative subjects (P.001). The number of deaths and the mortality rates are shown in Table 3. The HBsAg-positive subjects had an adjusted all-cause mortality risk that was 1.7 times greater (95% CI, , P.01) than for HBsAgnegative subjects. The mortality difference between both groups is secondary to differences in liver-related mortality. HBsAg-positive individuals were at a significantly greater risk of mortality from liver cancer (multivariable-adjusted hazard ratio, 22.4; 95% CI, ; P.01) and chronic liver disease and cirrhosis (multivariable-adjusted hazard ratio, 5.4; 95% CI, ; P.01) than HBsAg-negative individuals. Excluding liver-related events, the risk of death was similar (multivariable-adjusted hazard ratio, 1.0; 95% CI,.9 1.2) between both groups.

6 926 ILOEJE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 8 Table 5. Mortality in HBsAg-Positive Subjects Without Evidence of HCV Infection (n 3653) No. of deaths per 100,000 person-years No. (%) of participants (N 3653) Person-years of follow-up evaluation All causes (N 388) Liver cancer (N 119) Chronic liver disease and cirrhosis (N 40) Others a (N 229) All causes Liver cancer Chronic liver disease and cirrhosis Others a Female 1393 (38) 17, Male 2260 (62) 27, Age, y (33) 15, (28) 12, (29) 12, (10) Cigarette smoking b No 2416 (66) 30, Yes 1234 (34) 14, Alcohol consumption c No 3195 (87) 39, Yes 451 (12) HBeAg Seronegative 3088 (85) 38, Seropositive 565 (15) Level of ALT, U/L (94) 42, (6) Cirrhosis d No 3584 (98) 44, Yes 69 (2) Level of HBV DNA, copies/ml 300 (undetectable) 873 (24) 10, (32) 14, ,000 99, (18) , , (9) million 627 (17) a Causes other than liver cancer, chronic liver disease, and cirrhosis. b Data were not available for 3 participants. c Data were not available for 7 participants. d Diagnosed with ultrasonography within 6 months of study entry. As much as 41.2% of the mortality in the HBsAg-positive group (n 3931) and almost 11% of the mortality in the entire study population (n 22,472) was attributable to chronic infection with HBV. As would be expected, the proportion of deaths from liver cancer (95.5%) and chronic liver disease and cirrhosis (81.5%) in the HBsAg-positive group attributable to CHB infection was much higher, as was the population attributable risk for these causes of death (78.8% and 43.4%, respectively, in Table 3). In addition, we evaluated the all-cause and cause-specific mortality risk in persons infected with HBV but with low viral replication and compared them with the risk in HBsAgnegative people. To account for HCV related HCC, all subjects known to be positive for anti-hcv antibody (n 1313), or for whom this test was not done (n 35) were removed from these analyses as shown in figure 1. The overall risk of mortality was not different between HBsAg-negative (anti- HCV persons) and the HBV-infected persons with low viral load, however, the risk of mortality from HCC was 4.4-fold greater in these HBV-infected persons. The risk of mortality from chronic liver disease and cirrhosis also was higher in the HBV-infected persons with low viral replication level, but did not reach statistical significance. Predictors of Mortality in Hepatitis B Surface Antigen Positive Persons (n 3653) Table 5 shows the all-cause, liver-related, and non liverrelated mortality stratified by various risk factors in HBsAgpositive individuals. Overall, males, older people, cigarette smokers, and alcohol drinkers had higher mortality rates (for all-cause, liver-related, and non liver-related causes). Also, the all-cause and liver-related mortality rates were higher for individuals with positive HBeAg, increased serum ALT level, cirrhosis at study entry, and with increased serum HBV DNA levels. There was no association between these CHB-specific parameters and non liver-related causes of mortality. We also evaluated the association between HBV DNA level and mortality from all causes and liver-related causes stratified by certain key demographic variables (Table 6). As can be seen, within every risk factor stratum, there is a correlation between increasing HBV DNA level and risk of all-cause and liver-related mortality. This is not true for other causes of mortality once liver-related causes are removed. Table 7 shows the regression models evaluating various risk factors and their association with mortality risk. The significant predictors of all-cause mortality in this population were the

7 August 2007 HEPATITIS B VIRUS INFECTION AND MORTALITY 927 Table 6. Association Between HBV DNA Level and Mortality Stratified by Several Variables Adjusted hazard ratio a (95% CI) for HBV DNA in copies/ml Variables 300 (undetectable) ,000 99, , ,999 1 million All causes (n 388) Female Reference 0.8 ( ) 0.9 ( ) 2.4 ( ) b 3.4 ( ) c Male 1.2 ( ) 1.3 ( ) 1.4 ( ) 3.1 ( ) c 4.7 ( ) c Age, y 50 Reference 1.0 ( ) 1.3 ( ) 3.1 ( ) c 3.0 ( ) c ( ) c 4.3 ( ) c 4.8 ( ) c 9.2 ( ) c 18.1 ( ) c Cigarette smoking No Reference 0.8 ( ) 0.8 ( ) 2.6 ( ) c 3.1 ( ) c Yes 1.4 ( ) 1.7 ( ) b 2.1 ( ) d 3.5 ( ) c 6.2 ( ) c Alcohol consumption No Reference 0.8 ( ) 1.0 ( ) 2.3 ( ) c 3.4 ( ) c Yes 0.9 ( ) 1.5 ( ) 1.3 ( ) 3.0 ( ) c 4.8 ( ) c Liver related e (n 159) Female Reference 1.4 ( ) 2.7 ( ) 37.0 ( ) c 33.0 ( ) c Male 3.3 ( ) 4.7 ( ) 11.2 ( ) b 29.0 ( ) d 59.1 ( ) c Age, y 50 Reference 5.0 ( ) 9.3 ( ) b 29.1 ( ) d 35.7 ( ) c ( ) b 9.3 ( ) b 29.7 ( ) d 88.1 ( ) c ( ) c Cigarette smoking No Reference 0.8 ( ) 1.9 ( ) 14.4 ( ) c 19.3 ( ) c Yes 1.5 ( ) 3.1 ( ) 7.3 ( ) c 13.4 ( ) c 30.3 ( ) c Alcohol consumption No Reference 0.7 ( ) 3.1 ( ) b 11.7 ( ) c 20.4 ( ) c Yes 1.3 ( ) 5.9 ( ) c 5.5 ( ) d 11.7 ( ) c 23.0 ( ) c Others f (n 229) Female Reference 0.7 ( ) 0.8 ( ) 0.3 ( ) 1.5 ( ) Male 1.1 ( ) 1.1 ( ) 0.8 ( ) 1.5 ( ) 1.5 ( ) Age, y 50 Reference 0.8 ( ) 0.8 ( ) 1.5 ( ) 0.9 ( ) ( ) c 3.9 ( ) c 3.3 ( ) c 4.2 ( ) c 6.8 ( ) c Cigarette smoking No Reference 0.8 ( ) 0.6 ( ) 1.0 ( ) 1.0 ( ) Yes 1.5 ( ) 1.6 ( ) 1.4 ( ) 2.2 ( ) b 2.9 ( ) c Alcohol consumption No Reference 0.9 ( ) 0.7 ( ) 1.1 ( ) 1.1 ( ) Yes 0.8 ( ) 1.0 ( ) 0.7 ( ) 1.8 ( ) 2.4 ( ) d a Adjusted for age, sex, cigarette smoking, and alcohol consumption except for the stratifying variable. b P.05. c P.001. d P.01. e Including liver cancer, chronic liver disease, and cirrhosis. f Causes other than liver cancer, chronic liver disease, and cirrhosis. presence of cirrhosis at cohort entry, increased HBV DNA level, cigarette smoking, presence of HBeAg, male sex, and increasing age. The only significant predictors of non liver-related mortality were increasing age and cigarette smoking. For liver cancer deaths, the presence of cirrhosis at cohort entry, increased HBV DNA level, HBeAg serostatus, male sex, and increasing age were found to be significant predictors. Similar risk factors in addition to serum ALT predicted cirrhosis deaths. Figure 3 shows the survival curves by baseline serum HBV DNA level for all-cause, liver-related, and non liver-related mortalities, respectively. Higher baseline serum HBV DNA level was associated with poorer survival. Discussion In this report, we have follow-up data for mortality information to the end of 2004 for 22,472 subjects over a mean follow-up period of 12.5 years, and 282,323.7 PYFU. In the HBsAg-positive population, the overall mortality rate was per 100,000 person-years and men had a much higher mortality rate than women ( vs per 100,000. Others have reported mortality rates in HBsAg-positive Chinese individuals from Haimen City with similar gender disparity. 12 The classic study linking chronic HBV to HCC also was from Taiwan. 13,14 In that study, a total of 22,707 male government

8 928 ILOEJE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 8 Table 7. Predictors of Mortality in HBsAg-Positive Subjects Without Evidence of HCV Infection (n 3653) Adjusted hazard ratio (95% CI) All causes (n 388) Liver cancer (n 119) Chronic liver disease and cirrhosis (n 40) Others a (n 229) Female Referent Referent Referent Referent Male 1.4 ( ) b 1.8 ( ) b 1.2 ( ) 1.2 ( ) Age in 1-year increments 1.09 ( ) c 1.10 ( ) c 1.06 ( ) c 1.10 ( ) c Cigarette smoking No Referent Referent Referent Referent Yes 1.8 ( ) c 1.4 ( ) 1.9 ( ) 2.1 ( ) c Alcohol consumption No Referent Referent Referent Referent Yes 1.3 ( ) 1.4 ( ) 1.6 ( ) 1.2 ( ) HBeAg Seronegative Referent Referent Referent Referent Seropositive 1.9 ( ) c 3.3 ( ) c 2.2 ( ) 1.0 ( ) Level of ALT, U/L 45 Referent Referent Referent Referent ( ) 1.3 ( ) 2.3 ( ) b 1.3 ( ) Cirrhosis d No Referent Referent Referent Referent Yes 3.7 ( ) c 8.9 ( ) c 6.7 ( ) c 0.7 ( ) Level of HBV DNA, copies/ml 300 (undetectable) Referent Referent Referent Referent ( ) 0.7 ( ) 5.3 ( ) 0.9 ( ) 10,000 99, ( ) 2.1 ( ) 7.6 ( ) 0.8 ( ) 100, , ( ) c 6.9 ( ) c 11.1 ( ) b 1.2 ( ) 1 million 2.1 ( ) c 5.7 ( ) c 15.6 ( ) b 1.3 ( ) a Causes other than liver cancer, chronic liver disease, and cirrhosis. b P.05. c P.001. d Diagnosed with ultrasonography within 6 months of study entry. employees were studied for the risk of HCC. In the original report in 1981, after a mean follow-up period of only 3.3 years, the hazard ratio of HCC mortality was 223 (95% CI, ) in the HBsAg-positive men. This was based on only 40 HCC deaths in the HBsAg-positive men and 1 HCC death in the control group, accounting for the extremely wide CI. In an interim report from this study with follow-up evaluation through December 31, 1983, there were 116 cases of HCC over 140,000 person-years of follow-up. 15 Almost all of the HCC patients were HBsAg-positive men (113 of 116), for a reported HCC incidence rate of 527/100,000. After 8.9 years of follow-up evaluation, there were 161 cases of HCC in the HBsAg-positive men (incidence rate, per 100,000 and 9 in the HBsAg-negative men (incidence rate, 5.3 per 100, To see how our cohort compared with the Beasley study, 14 we recalculated the HCC incidence in the men from our study cohort, which after more than 12 years of follow-up evaluation was /100,000 PYFU. This compared favorably with the reported incidence in the Beasley study 14 because it is expected that with longer follow-up times (12.5 y in our study vs 8.9 y in the Beasley study) 14 more HCC cases would have occurred because the population would be aging. Our study varied in 2 significant ways from the Beasley study, 14 which also might explain the differences in the reported HCC incidence rates. First, we restricted recruitment to age 30 or older as compared with age 20 or older in the Beasley study. 14 Second, we did not screen for evidence of past HBV infection in the uninfected persons, which was an important difference because all the HCC cases in the HBsAg-negative men in the Beasley study 14 occurred in men with evidence of past HBV infection. We also found a 22.4-fold increase in risk of HCC mortality as well as a 5.4-fold increase in risk of death from chronic liver disease/cirrhosis. This compared very well with other studies from Asia. 16,17 For example, recent data from Haimen City, China, showed the age-adjusted mortality risk from HCC to be 23.2 in males and 25.8 in females compared with noninfected matched controls. 17 In contrast to our study, the Haimen City study investigators found a small but significantly greater risk of nonliver deaths (adjusted relative risk, 1.4) and nonliver cancers (adjusted relative risk, 1.7) among HBsAg-positive persons. 17 In Western countries where the prevalence of HBV is low, mortality risk from chronic HBV is correspondingly low. In Canada, 855 HBV-related deaths were reported between 1979 and 1997, and the mortality rate per 100,000 persons was.03 in 1979, and.26 in In the study of blood donors in the United Kingdom, 5 all-cause mortality was not different between HBsAg-positive men and the general population (standardized mortality ratio, 1.10); however, the standard mortality rate for HCC was 26 times higher for infected men than the general population. Other smaller studies in Western countries have

9 August 2007 HEPATITIS B VIRUS INFECTION AND MORTALITY 929 Figure 3. (A) Total mortality according to baseline HBV DNA level in HBsAg-positive subjects without evidence of HCV infection (n 3653). (B) Liver-related mortality according to baseline HBV DNA level in HBsAgpositive subjects without evidence of HCV infection (n 3653). (C) Non liver-related mortality according to baseline HBV DNA level in HBsAgpositive subjects without evidence of HCV infection (n 3653). reported no significant differences in overall mortality between HBsAg-positive and HBsAg-negative cohorts. 19,20 We provide further insight into the potential benefits at a population level that might accompany a comprehensive hepatitis B management program, which should include effective antiviral therapy. In our calculation of the population attributable risk, we show that almost 11% of all deaths (78.8% of liver cancer deaths and 43.4% of chronic liver disease/cirrhosis deaths) in the entire population of 22,472 subjects were caused by HBV infection. This suggests that a significant reduction in mortality at a population level can be achieved by preventing the complications of CHB infection (liver cancer and cirrhosis),

10 930 ILOEJE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 8 Figure 3. Continued which account for the excess mortality risk. Focusing only on the HBsAg-positive people, the attributable mortality risk is even higher for HCC and chronic liver disease/cirrhosis, and the potential benefit of effectively preventing progression to these liver complications is even greater. These estimates refer to expected benefits of preventing CHB-related complications after infection already has taken place (secondary prevention). Primary prevention of CHB infection by universal childhood immunization including a birth dose, is projected by the US Centers for Disease Control and Prevention to have the potential to prevent 84% of a projected 1.4 million HBV-related deaths globally. 21 Childhood immunization has been shown to reduce both the incidence of acute HBV infection and the rate of HCC in areas of the world that have used vaccination programs of newborn babies, such as Taiwan. 22 Other countries such as Thailand also have instituted effective population-wide vaccination programs with measurable success in reducing HBsAg seroprevalence rates, but are yet to document an impact on HCC rates. 23 Unfortunately, not all countries (even in highly endemic regions of the world) have adopted universal vaccination policies nor experienced such success. In 2001, data on vaccination rates in children younger than 1 year old showed that 65% in the Western Pacific, 9% in southeast Asia, and 6% in Africa were vaccinated fully. 24 Moreover, approximately 50 million new cases of hepatitis B infection are diagnosed annually. 2 These factors mean that the burden of disease from CHB will remain a public health concern for the foreseeable future. An important contribution of this study that we did not address in our previous publications is the risk of long-term mortality in HBV-infected persons with low viral load compared with uninfected persons. HBV-infected persons with HBV DNA levels less than 10 4 copies/ml are considered to have inactive disease by international treatment guidelines 25 and therefore are not candidates for therapy. Our data, however, show that they have a highly significant 4.4-fold greater risk of dying from liver cancer compared with uninfected persons. The risk of mortality from other liver complications, although increased, did not reach significance in our cohort. This finding may yet provide some evidence to reconsider the classification of these subjects as inactive. The data we present were generated from people who most probably were infected in early childhood with genotype B and C strains of the hepatitis B virus, as such, the generalizability of these results to people infected later in life or with other strains of the virus should be performed cautiously. Among the HBsAg-positive cohort, we have reported only the relationship between baseline risk factors and eventual mortality. It is important to continue this line of investigation and evaluate the impact of changes in these risk factors over time on the risk of eventual mortality, as our group is currently doing. The data we present show that significant excess mortality risk exists in HBV-infected persons, and this is predictable based on the level of viral load. Second, persons currently considered to have inactive disease do indeed have a significantly high risk of dying from HCC when appropriately compared with uninfected persons. Finally, secondary prevention targeting the prevention of liver cancer and cirrhosis should result in a significant reduction in mortality, especially in highly endemic populations such as the REVEAL-HBV cohort. This is supported by the convincing evidence showing that effective antiviral therapy slows disease progression 26 and reverses liver injury The data we have presented in this current report

11 August 2007 HEPATITIS B VIRUS INFECTION AND MORTALITY 931 suggest that a plausible mechanism for these benefits may be through sustained suppression of viral replication. References 1. Lavanchy D. Hepatitis B virus epidemiology, disease burden, treatment, and current and emerging prevention and control measures. J Viral Hepat 2004;11: Merican I, Guan R, Amarapuka D, et al. Chronic hepatitis B virus infection in Asian countries. J Gastroenterol Hepatol 2000;15: Ganem D, Prince AM. Hepatitis B virus infection natural history and clinical consequences. N Engl J Med 2004;350: Amin J, Dore GJ, O Connell DL, et al. Cancer incidence in people with hepatitis B or C infection: a large community-based linkage study. J Hepatol 2006;45: Crook PD, Jones ME, Hall AJ. Mortality of hepatitis B surface antigen-positive blood donors in England and Wales. Int J Epidemiol 2003;32: Yood MU, Quesenberry CP, Guo D, et al. Incidence of non- Hodgkins lymphoma among individuals with chronic hepatitis B virus infection. Hepatology 2007;46: Xu B, Hu DC, Rosenberg DM, et al. Chronic hepatitis B: a longterm retrospective cohort study of disease progression in Shanghai, China. J Gastroenterol Hepatol 2003;18: Lin X, Robinson NJ, Thursz M, et al. Chronic hepatitis B virus infection in the Asia-Pacific region and Africa: review of disease progression. J Gastroenterol Hepatol 2005;20: Fattovich G, Brollo L, Guistina G, et al. Natural history and prognostic factors for chronic hepatitis type B. Gut 1991;32: Chen CJ, Yang HI, Su J, et al, for the REVEAL-HBV Study Group. Risk of hepatocellular carcinoma across a biological gradient of serum hepatitis B virus DNA level. JAMA 2006;295: Iloeje UH, Yang HI, Su J, et al, for the REVEAL-HBV Study Group. Predicting liver cirrhosis risk based on the level of circulating hepatitis B viral load. Gastroenterology 2006;130: Evans AA, Chen G, Ross EA, et al. Eight-year follow-up of the 90,000-person Haimen City cohort: I. Hepatocellular carcinoma mortality, risk factors, and gender differences. Cancer Epidemiol Biomarkers Prev 2002;11: Beasley RP, Hwang LY, Lin CC, et al. Hepatocellular carcinoma and hepatitis B virus. A prospective study of 22,707 men in Taiwan. Lancet 1981;2: Beasley RP. Hepatitis B virus the major etiology of hepatocellular carcinoma. Cancer 1988;61: Beasley RP, Hwang LY. Hepatocellular carcinoma and hepatitis B virus. Semin Liver Dis 1984;4: Sakuma K, Saitoh N, Kasai M, et al. Relative risks of death due to liver disease among Japanese male adults having various statuses for hepatitis B s and e antigen/antibody in serum: a prospective study. Hepatology 1988;8: Chen G, Lin W, Shen F, et al. Chronic hepatitis B virus infection and mortality from non-liver causes: results from the Haimen City cohort study. Int J Epidemiol 2005;34: Pohani G, Zou S, Tepper M. Trends of hepatitis B and hepatitis C mortality in Canada, Can J Public Health 2001;92: Manno M, Camma C, Schepis F, et al. Natural history of chronic HBV carriers in northern Italy: morbidity and mortality after 30 years. Gastroenterology 2004;127: Villeneuve P, Desrochers M, Infante-Rivard C, et al. A long-term follow-up study of asymptomatic hepatitis B surface antigen-positive carriers in Montreal. Gastroenterology 1994;106: Goldstein ST, Zhou F, Hadler SC, et al. A mathematical model to estimate global hepatitis B disease burden and vaccination impact. Int J Epidemiol 2005;34: Chang MH, Chen CJ, Lai MS, et al, the Taiwan Childhood Hepatoma Study Group. Universal hepatitis B vaccination in Taiwan and the incidence of hepatocellular carcinoma in children. N Engl J Med 1997;336: Chongsrisawat V, Yoocharoen P, Theamboonlers A, et al. Hepatitis B seroprevalence in Thailand: 12 years after hepatitis B vaccine integration into the national expanded programme on immunization. Trop Med Int Health 2006;11: Centers for Disease Control and Prevention Global progress towards universal childhood hepatitis B vaccination. Centers for Disease Control and Prevention. MMWR Morb Mortal Wkly Rep 2003;52: Lok ASF, McMahon BJ. Chronic hepatitis B. Hepatology 2007; 45: Liaw YF, Sung JJY, Chow WC, et al, the Cirrhosis Asian Lamivudine Multicentre Study Group. Lamivudine for patients with chronic hepatitis B and advanced liver disease. N Engl J Med 2004;351: Dienstag JL, Goldin RD, Heathcote EJ, et al. Histological outcome during long-term lamivudine therapy. Gastroenterology 2003; 124: Hadziyannis S, Tassopoulos NC, Chang TT, et al. Long-term adefovir dipivoxil treatment induces regression of liver fibrosis in patients with HBeAg-negative chronic hepatitis B: results after 5 years on therapy. Hepatology 2005;42:754A. 29. Hadziyannis SJ, Tassopoulos NC, Heathcote EJ, et al, the Adefovir Dipivoxil 438 Study Group. Long-term therapy with adefovir dipivoxil for HBeAg-negative chronic hepatitis B. N Engl J Med 2005;352: Address requests for reprints to: Chien-Jen Chen, ScD, Genomics Research Center, Academia Sinica, 128 Academia Road, Section 2, Nankang, Taipei, 115, Taiwan. cjchen@ntu.edu.tw; fax: (886) Supported by the Department of Health, Executive Yuan, Republic of China; and Bristol-Myers Squibb Company. U.H.I. and J.S. are employed by Bristol-Myers Squibb Company. A grant was provided by Bristol- Myers Squibb Company to C.-J.C. for conducting the laboratory tests for this study. The authors would like to thank the following institutions and investigators who participated in the Risk Evaluation of Viral load Elevation and Associated Liver Disease/Cancer (REVEAL)-HBV Study Group as a part of the Taiwan Community Based Cancer Screening Project: National Taiwan University Hospital: C. Y. Hsieh, H. S. Lee, P. M. Yang, C. H. Chen, J. D. Chen, S. P. Huang, C. F. Jan; College of Public Health, National Taiwan University: T. H. H. Chen; Department of Public Health, National Defense Medical Center: C. A. Sun; Taipei City Psychiatric Center: M. H. Wu; Department of Public Health, Tzu Chi University: S. Y. Chen; Shin Kong Wu Ho-Su Memorial Hospital: K. E. Chu; Huhsi Health Center, Penghu County: S. C. Ho, T. G. Lu; Provincial Penghu Hospital: W. P. Wu, T. Y. Ou; Sanchi Health Center, Taipei County: C. G. Lin; Provincial Chutung Hospital: K. C. Shih; Provincial Potzu Hospital: W. S. Chung, C. Li; Kaohsu Health Center, Pingtung County: C. C. Chen; and Paihsa Health Center, Penghu County: W. C. How.