Cofactors associated with liver disease mortality in an HBsAg-positive Mediterranean cohort: 20 years of follow-up
Article first published online: 22 FEB 2006
Copyright © 2006 Wiley-Liss, Inc.
International Journal of Cancer
Volume 119, Issue 3, pages 687–694, 1 August 2006
How to Cite
Ribes, J., Clèries, R., Rubió, A., Hernández, J. M., Mazzara, R., Madoz, P., Casanovas, T., Casanova, A., Gallen, M., Rodríguez, C., Moreno, V. and Bosch, F. X. (2006), Cofactors associated with liver disease mortality in an HBsAg-positive Mediterranean cohort: 20 years of follow-up. Int. J. Cancer, 119: 687–694. doi: 10.1002/ijc.21882
- Issue published online: 8 MAY 2006
- Article first published online: 22 FEB 2006
- Manuscript Accepted: 27 DEC 2005
- Manuscript Received: 29 JUL 2005
- Instituto de Salud Carlos III, Spanish Government. Grant Numbers: FIS 94/1635, RTICCC C03/09, RTICESP C03/10
- hepatocellular carcinoma;
- liver cirrhosis;
- risk factors;
- cohort study
The risk of developing liver cancer in hepatitis B virus (HBV) carriers differs across geographical areas, suggesting that exposure to other risk factors may contribute to HBV-linked cancer risk. Our study estimates the mortality due to liver disease and the role of other risk factors in a Spanish HBV cohort. 2,352 hepatitis B surface antigen (HBsAg)-positive and 15,504 HBsAg-negative subjects were identified among blood donors during 1972–1985 and were followed until December 2000 through the Mortality Registry. Clinical examination and an epidemiological questionnaire were performed on 1,000 HBsAg-positive survivors during 1994–1996. In subjects deceased from liver disease, medical records were revised and relatives were interviewed. A nested case-control analysis was conducted comparing both groups. In HBsAg-positive men, an excess mortality from liver cancer [standardized mortality ratio (SMR): 14.1; 7.7–23.6], cirrhosis (SMR: 10.5; 7.0–15.1), haematological neoplasms (SMR: 3.2; 1.2–6.9) and AIDS was detected (SMR: 5.5; 2.2–11.4). In women, an excess was found for cirrhosis (SMR: 7.2; 1.4–21.1). Progression factors to liver disease were alcohol intake [odds ratio (OR): 6.3; 3.1–12.8], diabetes (OR: 3.6; 1.3–9.6), HBV replication (OR: 50.0; 14.9–167.3) and hepatitis C virus (HCV) infection (OR: 27.4; 7.1–107.7). In conclusion, in Spain after 20 years of follow-up, chronic HBV exposure appears as a major risk factor for liver cancer among men and for cirrhosis in both sexes. The risk of death from liver disease among HBV carriers with the presence of HBV replication, HCV, alcohol consumption and diabetes was significantly increased and suggests synergism among these exposures and HBV. Mortality from haematological neoplasms was detected and could be associated to HIV coinfection. These results support screening and adequate follow-up among HBsAg-positive subjects at high risk to develop liver disease, particularly when these risk cofactors are present. © 2006 Wiley-Liss, Inc.
It has been estimated that, in the year 2000, 564,000 new cases of malignant liver tumours were diagnosed worldwide; 457,000 in developing countries and 107,000 in developed countries.1 High risk areas for hepatocellular carcinoma (HCC) include South East Asia and some regions of Africa, while low risk areas for HCC include Northern Europe, Australia, New Zealand and the Caucasian population of America. In Europe, Mediterranean countries are considered as intermediate risk for HCC.1 The different exposure to hepatitis B virus (HBV) and hepatitis C virus (HCV) is thought to explain part of this variability.2 However, it is unclear to what extent the coexistence of other risk factors may contribute to the risk of HCC in HBV carriers.
Several longitudinal studies based on cohorts of HBV carriers, published between 1981 and 2004, have estimated the risk of death from liver disease.3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 Most of these studies have been carried out in Asian countries (4 in China,3, 4, 5, 6 2 in Taiwan7, 8, 9 and 6 in Japan10, 11, 12, 13, 14, 15); 4 in Western countries (2 in the US,16, 17 1 in the UK18, 19 and 1 in Italy20). Only 3 studies included women17, 18, 19, 20 and only 5 evaluated the role of other risk factors in mortality risk.4, 5, 6, 9, 15, 20 Estimates of the risk of death due to HCC in HBV carriers range from 5.6 to 103 times higher than that in hepatitis B surface antigen (HBsAg)-negative subjects. Reasons for the heterogeneity in these estimates are unclear. Differences in the population sampled may play a role contributing to differences in baseline risk. Alternatively, variability in the coexistence of other risk factors may contribute to differences in risk among HBsAg-positive subjects.
The purpose of our study is to determine the risk of mortality from liver disease in Mediterranean HBsAg-positive subjects with a mean of 20 years of follow-up, and to identify the prevalence of other risk factors among deaths from liver disease.
Material and methods
2,352 HBsAg-positive subjects (1,656 men and 696 women) were identified among all subjects who went to donate blood voluntarily in 4 blood banks (Bellvitge, Clínic, Vall d'Hebron and Sant Pau Hospitals) in the area of Barcelona (Catalonia, North East of Spain) during the period 1972–1985. In the same period, 15,504 HBsAg-negative subjects (8,783 men, 6,721 women) accepted as blood donors at the Bellvitge Hospital blood bank were also included as a comparison group in the cohort.
Subjects were accepted as blood donors if temperature, arterial pressure and haemoglobin were within normal limits and body weight was over 50 kg. Subjects were excluded if they had a previous history of hepatitis, contact with patients who probably had hepatitis in the previous 6 months or if the HBsAg was positive. Subjects with insulin-dependent diabetes were also ineligible to donate blood.
The cohort was retrospectively recruited in 1987 by identifying individuals from the records of the blood banks. For every subject included in the cohort, the following data were collected: family and first names, sex, birth date, town and province of residence, address and number and dates of blood donations.
The detection of HBsAg was performed during routine screening in the blood banks, using immunoelectrophoresis through 1975 and using passive haemagglutination since 1976. Data about hepatitis B ‘e’ antigen (HBeAg) status and other HBV seromarkers were not available. HCV and human immunodeficiency virus (HIV) infections were not screened for among donors at the time of recruitment of the present cohort.
Information sources of mortality
The information source to detect mortality in the retrospective cohort (from 1972 until December 2000) was the Catalan Mortality Registry (CMR). CMR collects data for all deaths in Catalonia, and their completeness and their quality have been long evaluated through different studies.21, 22 Deaths were identified in the CMR through computer cross referencing of this registry with the cohort database. The computer application, based on a probabilistic method, identified possible matches from the combination of name, sex, birth date, town and province of residence. The primary cause of death registered, coded with the International Classification for Diseases, Ninth Edition (ICD-9),23 was collected.
For all HBsAg-positive subjects who died from liver disease in the cohort, medical records were reviewed, in order to contrast the clinical information with the CMR information and to obtain exposures to other risk factors associated to HCC. For the HBsAg-positive subjects who died from diseases not clearly related to the liver, medical records were reviewed if their cause of death was: malignant neoplasm without specification of site, cancer of pancreas, AIDS, diabetes, respiratory and digestive diseases and unspecified causes of death.
Nested case-control in HBsAg-positive subjects
We conducted a nested case-control study in the cohort to determine the role of other risk factors in the mortality from liver disease in HBsAg-positive subjects.
In the period 1972–1993, 109 subjects died (100 men, 9 women). All HBsAg-positive survivors (n = 2,243) were invited to a clinical examination at the Catalan Institute of Oncology, during the period 1994–1996. Thousand individuals agreed to participate (708 men, 292 women), 497 refused (335 men, 162 women) and 746 (513 men, 233 women) could not be contacted. No differences were detected in gender, age or the hospital of the blood donation between those who participated and those lost for follow-up.
Information sources of risk factor exposures
To obtain data on lifetime tobacco and alcohol exposures and history of previous diseases, all subjects agreeing to participate were interviewed using a validated epidemiological questionnaire administered by well-trained interviewers. Lifetime alcohol consumption was collected according to the different phases in the interviewees' life so that it was easier for them to recall exactly what type of alcohol (beer, wine or spirits) they consumed and the quantity. Lifetime alcohol intake was expressed in grams per day average and was categorized as ≤60 g/day and >60 g/day. Lifetime tobacco exposure was obtained the same way as alcohol consumption, by type of tobacco (cigarette, cigar and pipe). Lifetime exposure tobacco was expressed as packs per year and was categorized as ≤20 packs/year and >20 packs/year. Personal history was collected taking into account the presence of diabetes, hemochromatosis and autoimmune diseases.
In addition, subjects provided a 10-ml sample of blood by venipuncture. Serological HBV markers [HBsAg, hepatitis B surface antibody (HBsAb), HBeAg, hepatitis B ‘e’ antibody (HBeAb), hepatitis B ‘core’ antibody (HBcAb) IgM, HBcAb IgG] and those for hepatitis delta virus antibody (HDVAb) were tested using commercial enzyme immunoassays (Abbott Laboratories, North Chicago, IL). HBV DNA in sera was detected through the Hybrid Capture test, with a sensitivity level of 5 pg/ml (Digene Hybrid Capture System, Beltsville). HCVAb was tested through second generation enzyme immunoassays and was confirmed by immunoblotting (ORTHO EIA2 and RIBA2, Ortho Diagnostic, Raritan, NH). HIVAb was detected using commercial enzyme immunoassays (Abbott Laboratories) and was confirmed by Western blot assays (BioRad Laboratories, Hercules, CA). HIV infection diagnosis was conducted in a 228 HBV infected cohort sample in accordance with the ethical considerations implemented in the study period.
Among HBsAg-positive subjects deceased from liver disease detected in the cohort study, relatives were contacted by phone to complete the epidemiological questionnaire similar to that used among living HBsAg-positive subjects who took part in the clinical examination. A relative's phone questionnaire was completed in 65% of the dead subjects.
Diagnosis of liver disease
Hepatic function was determined by liver enzymes (alanine and aspartate aminotransferases, γ-glutamic transpeptidase), alkaline phosphates, bilirubin, albumin, prothrombin time tests and α-fetoprotein. When liver function was abnormal, diagnosis was conducted via imaging analysis, using ultrasound, magnetic resonance or computed tomography according to the physician's criterion. Liver disease pathologic confirmation was not conducted among all subjects because of ethical considerations.
In the clinical examination we identified 132 subjects (73 men, 59 women) with all HBV seromarkers negative (HBsAg, HBsAb, HBcAb IgM, HBcAb IgG, HBeAg and HBeAb). These subjects were excluded from further studies. Remaining cohort members were classified as follows : (i) HBsAg-positive subjects without liver disease when their liver function and clinical examinations were normal; (ii) HCC patients when focal liver lesions were detected through imaging tests and/or increases of α-phetoprotein; (iii) liver cirrhosis (LC) when clinical, laboratory or imaging tests showed hepatic decompensation and/or portal hypertension and (iv) subjects with unspecified hepatopathy when clinical, laboratory or imaging tests showed liver inflammation without signs or symptoms of LC. In men, 512 controls, 102 unspecific hepatopathies, 11 LC and 2 HCC were detected. In women, 202 controls and 28 unspecific hepatopathies were diagnosed. Hepatic biopsy was conducted in our hospital on 1 subject with mild chronic hepatitis, 5 subjects with active chronic hepatitis and 3 patients with LC. These patients were treated in the same centre. Remaining subjects diagnosed with liver disease were treated in other hospitals of Catalonia.
Mortality in the cohort
In the analysis of the cohort study we excluded 146 HBsAg-positive subjects: 14 (8 men, 6 women) for whom the birth date was unknown and 132 with all HBV seromarkers negative. Mortality analysis finally included 2,206 HBsAg-positive subjects (1,575 men, 631 women) and as comparison group, 15,504 HBsAg-negative blood donors.
The age-specific mortality rates were calculated for each cause of death observed in the HBsAg-negative subjects, by sex and for the follow-up period. The age distribution was stratified into 5-year age groups. Age-group mortality rates for each cause of death were calculated as the ratio between the numbers of deaths divided by the number of person-years at risk. Expected deaths by cause in the HBsAg-positive subjects were calculated multiplying the age-, sex- and year-specific person-years at risk by the corresponding HBsAg-negative subjects' cause-specific mortality rates. Standardized mortality ratios (SMRs) were then calculated by dividing the number of observed deaths (O) with the expected number (E) in the HBsAg-positive subjects. Confidence intervals were calculated on the basis of the Poisson distribution, by means of Byar's approximation to the Poisson exact test.24, 25
Nested case-control study
Statistical analysis was conducted only in men because of the lower number of deaths from liver disease among women. All HBsAg-positive men of the cohort who died because of liver disease were considered as cases (n = 51). Two groups were defined among HBsAg-positive men who accepted to take part in the clinical examination. The first was constituted by HBsAg-positive men without liver disease (n = 512), and the second was constituted by HBsAg-positive men who developed liver disease but did not die (n = 107). The first group was considered as control group in the nested case-control.
To assess the magnitude of association between liver disease mortality and risk factors, age-adjusted odds ratios (ORs) and 95% confidence interval (95% CI) were computed by logistic regression using the maximum likelihood point estimate.26 The significance of alcohol and tobacco consumptions, HCV, HIV and hepatitis delta virus (HDV) infections and metabolic diseases was evaluated through the comparison between 51 HBsAg-positive men who died from liver disease (cases) and 512 HBsAg-positive men living without liver disease (controls).
Ethical approval was obtained from the Bellvitge Hospital Ethics Committee. Permission to use the CMR was obtained from the Assessment Committee for the Management of the Confidential Information of the Catalan Government.
In the HBsAg-positive subjects, the mean age at entry was 33.1 years (SD, ±10.7) among men and 34.4 years (SD, ±11.7) among women. The mean follow-up was 20.5 years for men (SD, ±4.6) and 20.7 years for women (SD, ±3.6). In the HBsAg-negative subjects, mean age at entry was 35.3 years (SD, ±10.5) for men and 35.6 years (SD, ±11.1) for women. The mean follow-up was 18.7 years among men (SD, ±4.1) and 18.1 years among women (SD, ±3.4).
Specific causes of death in the HBsAg-negative subjects
The total person-years at risk were 164,212 among men and 121,609 among women. The main causes of death were cardiovascular disease (n = 132; 23.2%), lung cancer (n = 83; 14.6%) and external causes (n = 72; 12.7%). Liver disease was the fifth cause of death (n = 28; 5%), after colorectal cancer (n = 37; 6.5%). Among women, the main causes of death were cardiovascular disease (n = 37; 18.0%), colorectal and breast cancers (n = 29; 14.1% each) and external causes (n = 16; 7.8%). The fourth cause was endometrial and ovary cancers (n = 13; 6.3% each) and haematological neoplasms, liver and respiratory diseases (n = 11; 5.4% each) ranked fifth. AIDS ranked tenth as death cause for both sexes (men: 11; 1.9% and women: 1; 0.5%).
Specific causes of death in HBsAg-positive subjects
The total number of person-years at risk was 32,264 among men and 13,104 among women. In men, the main causes of death were liver disease (24.7%), cardiovascular disease (17.2%) and lung cancer (9.7%). In women, the main causes of death were liver disease, colorectal cancer and cardiovascular disease (16.0% each), breast cancer (12.0%) and endometrial and ovary cancers (8% each). Among men, the fifth cause of death was AIDS together with respiratory diseases (4.3% each). No deaths were caused by AIDS among women (Table I).
|Cause of death||Men||Women|
|O||E||SMR||95% CI||O||E||SMR||95% CI|
|Infective & parasitic disease||4||0.57||7.05||1.89–18.06||0||0.13||0.00||0.00–28.08|
|Oral cavity & pharynx||3||1.31||2.29||0.46–6.69||0||0.00||–||–|
|Bone, connective tissue||2||0.13||15.52||1.74–56.03||0||0.06||0.00||0.00–57.03|
|Body of uterus, ovary||–||2||1.04||1.91||0.22–6.91|
|Unspecified cause of death||5||0.29||16.88||5.44–39.40||1||0.08||12.51||0.16–69.54|
Estimated risk of death in the HBsAg-positive subjects
During the study period, global mortality among HBsAg-positive women was similar to that among HBsAg-negative women (SMR, 1.18; 95% CI, 0.76–1.74). In the mortality for specific causes, significant differences were only detected in the LC mortality (SMR, 7.23; 95% CI: 1.45–21.12). Among HBsAg-positive men, global mortality was 2-fold higher when compared to that among the HBsAg-negative men (SMR, 2.12; 95% CI, 1.83–2.45). An 11-fold higher risk of death from LC was observed (SMR, 10.48; 95% CI, 6.96–15.14), as well as a 14-fold higher risk of death from liver tumour (SMR, 14.09; 95% CI, 7.69–23.64) when compared to that from the HBsAg-negative group. An excess mortality among HBsAg-positive men was also confirmed by the following causes: malignant bone tumours, infectious diseases, respiratory disease, AIDS, cardiovascular disease and deaths codified as unknown cause (Table I).
Mortality analysis for specific causes conducted during 2 periods of follow-up (≤10 years and >10 years) showed that significant differences in women were only detected in the LC mortality during the first period (SMR, 14.63; 95% CI, 1.64–52.84) (Table II). In men, in the first period, a significant excess mortality was detected in LC (SMR, 7.87; 95% CI, 3.15–16.21) and HCC mortality (SMR, 9.09; 95% CI, 1.83–26.56). In the second period, a significant excess mortality was observed in infective diseases (SMR, 9.36; 95% CI, 2.52–23.96), HCC (SMR, 16.67; 95% CI, 8.31–29.82), LC (SMR, 11.81; 95% CI, 7.30–18.04), respiratory system (SMR, 10.0; 95% CI, 3.11–18.51), haematological neoplasms (SMR, 3.15; 95% CI, 1.15–6.86) and AIDS (SMR, 5.53; 95% CI: 2.22–11.41) (Table III).
|Cause of death||≤10 years of follow-up||>10 years of follow-up|
|O||E||SMR||95% CI||O||E||SMR||95% CI|
|Colon & rectum||0||1.08||0.00||0.00–3.41||4||2.41||1.67||0.45–4.27|
|Body of uterus & ovary||0||0.32||0.00||0.00–11.46||2||0.72||2.78||0.31–10.03|
|Mental disorders||0||0.00||–||–||1||0.09||11.11||0.15 – 61.81|
|Unspecified cause of death||0||0.00||–||–||1||0.08||12.51||0.16–69.54|
|Cause of Death||≤10 years of follow-up||>10 years of follow-up|
|O||E||SMR||95% CI||O||E||SMR||95% CI|
|Infective & parasitic disease||0||0.14||0||0.00–25.74||4||0.43||9.36||2.52–23.96|
|Oral cavity & pharynx||0||0.49||0||0.00–7.47||3||0.82||3.66||0.74–10.71|
|Colon & rectum||3||1.89||1.58||0.32–4.62||2||3.50||0.57||0.06–2.06|
|Bone & connective tissue||1||0.13||7.69||0.11–42.81||1||0.00||–|
|Unspecified cause of death||5||0.00||–||0||0.29||0.00||0.00–12.65|
Nested case-control study in HBsAg-positive men
Clinical records review confirmed that the cause of death was liver disease in 92% of the cases. Among liver tumours, 10 out of 12 HCCs (ICD-9: 1550) reported in the CMR were confirmed. Among 2 liver tumours coded as unspecified primary or secondary (ICD-9: 1552), 1 medical record was found, confirming that this liver tumour corresponded to a malignant gallbladder neoplasm with liver metastases. On the other hand, the intrahepatic bile duct tumour (ICD-9: 1551) detected in the CMR was confirmed.
Of the 31 deaths from LC, 30 were reviewed. Of these, 4 cases had a HCC that was not recorded in the CMR. The medical records review revealed that the excess mortality from infectious diseases among the HBsAg-positive men was attributable to HBV infection. Of the 4 deaths from infectious diseases, 3 were coded as acute hepatitis B (ICD-9: 070) and 1 as actinomycosis sepsis (ICD-9: 039) in the CMR. In this last case, the infection was a complication in the postoperative period of a liver transplant in a HCC patient. Two of three deaths from acute hepatitis B were reviewed. One case was a 50-year-old man HIV positive and with HBV replication (DNA HBV positive by PCR) who died due to a fulminant hepatitis during HBV viral reactivation. The other case was a 51-year-old man with LC associated to HBV who developed HCC. In the third case, a 46-year-old man, it was impossible to review the medical record.
Among the 161 deaths in the HBsAg-postive men, originally coded as being from nonliver diseases, 21 medical records (13%) were reviewed based on the criteria described earlier. Among these, 2 HCC and 5 LC were detected. Two HCC cases were coded as actinomycotic infection and diabetes in the CMR. Five deaths due to LC were coded in the CMR as acute kidney failure, malignant neoplasms without specification of site, 2 bronchopneumonias and 1 duodenal ulcer. The number of deaths from liver disease in the HBV carriers after adding these cases reached 57 (51 men: 8 unspecified liver diseases, 28 LC and 15 HCC; 6 women: 4 LC and 2 HCC).
Risk factors related to liver disease among HBsAg-positive men
Regression analysis controlling for age showed that the presence of HCV and HDV infections, alcohol consumption and diabetes were risk factors for death from liver disease in HBsAg-positive men (Table IV). The risk of death from liver disease increased 27-fold with HCV infection, 111-fold with HDV infection, 6-fold with lifetime alcohol consumption (>60 g/day) and 4-fold with diabetes. The number of missing values are greater in cases deceased from liver disease than that in the subjects living without liver disease. Missing values in these variables were associated to death due to liver disease (HCV: 121.3, 95% CI, 45.9–320.8; HDV: 134.2, 95% CI, 53.5–336.8; Alcohol: 72.0, 95% CI, 22.4–230.7 and diabetes: 55.9, 95% CI, 19.0–164.8). Although tobacco consumption was not associated to liver disease, missing values in this variable were related to death by this cause (OR: 61.6, 95% CI, 19.7–192.7).
|Risk factors||Living without liver disease1 (N = 512)||Living with liver disease (N = 107)||Deceased from liver disease (N = 51)||OR (95% CI)|
|HCV||−||500 (97.6)||98 (91.5)||16 (31.4)||1|
|+||5 (1.0)||7 (6.5)||5 (9.8)||27.4 (7.1–107.7)|
|Unknown||7 (1.4)||2 (2.0)||30 (58.8)||121.3 (45.9–320.8)|
|HIV2||−||155 (30.3)||16 (14.9)||7 (13.7)||1|
|+||0 (0.0)||1 (0.9)||3 (5.9)||NC|
|Unknown||357 (69.7)||90 (84.1)||41(80.4)||2.89 (1.3–6.6)|
|HDV||−||499 (97.4)||100 (93.5)||12 (23.5)||1|
|+||1 (0.2)||1 (1.0)||2 (3.9)||111.4 (8.46–1,001.2)|
|Unknown||12 (2.4)||6 (5.5)||37 (72.6)||134.2 (53.5–336.8)|
|Tobacco1||−||293 (57.2)||60 (56.1)||15 (29.4)||1|
|+||213 (41.6)||45 (42.1)||18 (35.3)||1.4 (0.7–2.9)|
|Unknown||6 (1.2)||2 (1.8)||18 (35.3)||61.6 (19.7–192.7)|
|Alcohol3||No||420 (82.0)||76 (71.0)||15 (29.4)||1|
|Yes||86 (16.8)||29 (27.1)||22 (43.1)||6.3 (3.1–12.8)|
|Unknown||6 (1.2)||2 (1.9)||14 (27.5)||72.0 (22.4–230.7)|
|Diabetes||No||482 (94.1)||95 (88.8)||27 (52.9)||1|
|Yes||24 (4.7)||9 (8.4)||6 (11.8)||3.6 (1.3–9.6)|
|Unknown||6 (1.2)||3 (2.8)||18 (35.3)||55.9 (19.0–164.8)|
|Hemochromatosis||No||512 (100.0)||107 (100.0)||50 (98.0)||1|
|Yes||0 (0.0)||0 (0.0)||1 (2.0)||NC|
Association between HIV infection and death due to liver disease was not possible to asses because no infected subjects were detected in the control group. In fact, missing values in the HIV variable were associated to death due to liver disease (OR: 2.89, 95%CI, 1.3–6.6) (Table IV).
Table V shows the risk of death from liver disease associated with the presence of HBV replication seromarkers. The risk of death from liver disease increased 61-fold with the presence of HBeAg and 29-fold with the presence of HBcAb IgM. DNA HBV risk was not possible to be estimated because of absence of positive controls. The risk of death from liver disease increased 50-fold when any of the replication seromarkers were positive. Missing values in these variables were also associated to death due to liver disease (HBeAg: 278, 95% CI, 84.2–919.1; HBcIgM: 115.3, 95% CI, 46.4–286.6; HBV replication: 477.9, 95% CI, 99.5–837.2).
|Risk factors||Living without liver disease (N = 512)||Living with liver disease (N = 107)||Deceased from liver disease (N = 51)||OR (95% CI)|
|HBeAg||−||505 (98.6)||98 (91.6)||14 (27.4)||1|
|+||3 (0.6)||8 (7.5)||5 (9.8)||60.8 (12.5–295.1)|
|Unknown||4 (0.8)||1 (0.9)||32 (62.8)||278.2 (84.2–919.1)|
|−||498 (97.3)||95 (88.8)||14 (27.5)||1|
|+||3 (0.6)||8 (7.5)||3 (5.8)||28.5 (4.9–163.9)|
|Unknown||11 (2.2)||4 (3.7)||34 (66.7)||115.3 (46.4–286.6)|
|−||90 (17.6)||29 (27.1)||5 (9.8)||1|
|+||0 (0.0)||16 (15.0)||5 (9.8)||NC|
|Unknown||422 (82.4)||62 (57.9)||41 (80.4)||1.87 (0.7–4.9)|
|HBV replication2||−||504 (98.4)||87 (81.3)||15 (29.5)||1|
|+||6 (1.2)||20 (18.7)||9 (17.6)||50.0 (14.9–167.3)|
|Unknown||2 (0.4)||–||27 (52.9)||477.9 (99.5–837.2)|
Among women, 2 HCC and 4 LC cases were detected. One woman who died from LC was HIV positive and another had HBV replication seromarkers. One woman who died from HCC was diabetic. In the third and fourth deaths from LC and the second death from HCC it was not possible to review the medical records.
This cohort study, with a mean follow-up of 20.5 years, shows a strong and significant association between chronic HBV infection and HCC among men and with LC in both sexes in a Mediterranean country. Alcohol, HCV infection, the presence of HBV replication seromarkers and diabetes were risk factors to death from liver disease in HBsAg-positive men.
The excess mortality from liver disease found in our study is consistent with that found in the previous Western studies. The study carried out in the UK confirmed a higher risk of death from HCC and from LC among HBsAg-positive men. Among women, only a significantly higher risk of death from LC was confirmed.19 In the US, 2 studies found an excess risk of death from HCC among men.16, 17 In contrast, in a study conducted in Italy after 30 years of follow-up, the cohort of a 296 HBsAg-positive blood donors was not associated with a higher mortality due to liver disease.20 In that study, HCC incidence among HBsAg-positive subjects was 33.80 per 105 person-years (data not reported) similar to that obtained in the cohort of UK (33.30)19 and in our cohort (34.10). However, the Italian study shows the highest HCC incidence rates in HBsAg-negative subjects (28.9 per 105; based on 1 case with a daily alcohol intake greater than 60 g/day on 120 controls), being very similar to the HCC incidence of HBsAg-positive subjects. The similarity of the Italian HCC incidence rate in HBV carriers with those found in other Western cohort studies suggests that the lack of excess of HCC risk could be due to a selected small group control. Most prospective studies published among HBV carriers have been carried out in Asian countries. In Japan, where HCC has been related mainly to HCV infection, an excess mortality due to HCC and LC in HBV carriers was also found.10, 11, 12, 13, 14, 15 In China and Taiwan, where HCC has been related mostly to HBV, the estimated risks for HCC among men vary between 7 and 103,3, 4, 5, 6, 7, 8, 9 showing higher risks than those in other geographical areas. The higher risk of death from liver disease found in South East Asia has also been reported in Western studies. The UK study found a risk of death from HCC among immigrant men higher (SMR, 52) than that among natives (SMR, 23),19 and in the American studies, 70% of deaths from HCC were immigrants, mostly African and Asian.16, 17 In our study, immigrant population was not included; therefore, risk of death from liver disease was estimated among Mediterranean population. The heterogeneity of the estimated death risks from HCC observed in these studies may be attributed to differences in intrinsic HVB and environmental factors.
HBV transmission in early age and aflatoxin exposure could explain the higher liver disease risk detected in the South East Asian studies and among immigrant subjects in the Western studies.27, 28, 29, 30 HBV genotypes show a characteristic geographical distribution and, although its clinical meaning is not yet clear, some studies have proposed that certain genotypes could be related to a greater progression to liver damage.31, 32 In our study, it was not possible to obtain data about HBV genotypes among deceased HBV carriers in order to evaluate their role in the HCC progression in a Mediterranean country.
Some studies conducted worldwide have suggested that exposure to other risk factors related to HCC could modify the risk of developing liver disease among HBV carriers.33, 34, 35 In our study, the HBV replication seromarkers were associated with a 50-fold increase in risk of death from liver disease among HBsAg-positive men. These results are consistent with those found in Taiwan, where HBV carriers also seropositive to HBeAg and DNA HBV presented a higher risk of HCC.9, 10 As reported in other studies, HCV coinfection was also related to a higher risk of death from liver disease in HBsAg-positive men. HDV infection seems to be also related to a higher hepatic mortality in our cohort, although we cannot establish HDV as a risk factor to liver progression because of the low number of subjects tested for this infection. Coinfection of these viruses could stimulate the liver cell necroinflammatory injury carried out by cellular immunology; thus, this coinfection could play a role in the carcinogenesis of HBV-infected subjects.36, 37, 38, 39, 40, 41
Lifetime alcohol consumption was also associated with a 6-fold increase in risk of death from HCC and LC; therefore, it supports the hypothesis of a possible synergism between HBV infection and alcohol consumption in the aetiology of liver disease, which has also been reported in other studies.42, 43, 44, 45 Another factor that could contribute to the progression of liver disease is the presence of diabetes. Some population studies have suggested that diabetes is related to higher HCC incidence and mortality.46, 47, 48 Later, cohort and case-control studies have confirmed this association49, 50 and showed possible interaction between diabetes, alcohol intake and viral infections.51, 52 In our study, a 3.6-fold higher mortality from liver disease was found among HBsAg-positive men with diabetes, when compared to those without.
It has not been possible to determine the risk of death by liver disease associated with HIV infection in the cohort because there were no infected subjects in the control group. Likewise, the higher HIV prevalence observed among HBV carriers deceased (5.9%) when compared to the subjects living with liver disease (0.9%) and the controls (0.0%), and also the risk of death associated to HIV missing values, suggest that HIV may play a role in the liver disease development among HBV carriers.
There are some potential limitations in our study. First, 55% of living HBsAg-positive subjects did not take part in the study, making it impossible to determine the liver morbidity and risk factor exposures in those subjects. Odds ratios could be overestimated if subjects who did not take part in the clinical examination had highest risk factor prevalences than those who did. The detection of deceased cases in the cohort was not affected for this selection bias, because mortality was determined from RMC. During the period 1994–2000 there has not been observed significant differences in mortality rates of HCC and LC among subjects who took part in the clinical examination and those who did not (HCC mortality rates: 22.2 and 28.0, p > 0.05; LC mortality rates: 44.5 and 33.6, p > 0.05, respectively; data not shown). In fact, this suggests that subjects who did not take part in the study did not have higher risk of death. Second, in subjects who died before 1990 and 1985, HCV and HIV could not be detected. Also, information about exposure to environmental factors among deceased cases was collected from the medical records review and the relatives' questionnaires. In contrast, the control group answered the questionnaire and a medical examination was conducted, making the percentage of missing values more important among those deceased than in the living subjects. This information bias could originate an underestimation of the odd ratios, and these could be conservative. Third, data of HBeAg and liver enzymes were not available at the entry of the cohort. A study carried out in Italy showed that subjects with chronic hepatitis B had a global mortality 5.2-fold higher than the general population, and in cirrhotic patients the excess mortality reached 18.6.53 Other European and US studies based upon HBV carriers have confirmed this results.54, 55, 56, 57, 58 In our study, it is reasonable to expect a low number of subjects with liver disease at entry of the cohort, because all subjects were screened volunteer blood donors. In fact, the risk of death from liver disease in our cohort was higher in the second decade of follow-up than in the first, suggesting that most of the HBV carriers were healthy at the entry of the cohort.
On the other hand, our study presents potential advantages in respect to the other published studies. Our comparison group was constituted by HBsAg-negative blood donors, and it leads to avoid, first, the bias known as ‘Healthy Donor Effect’ and, second, the inclusion of HBV carriers.13, 15, 19 These issues are not controlled in the studies in which general population is the comparison group. Third, we have carried out a nested case-control study to asses the role of other risk factors on the contribution to liver disease progression in Mediterranean HBsAg-positive subjects.
Regarding the mortality cohort study from nonliver diseases, a significantly higher risk of death from some specific causes has been found in men. The medical records review has revealed that the excess mortality for infectious diseases detected in the second period of the follow-up is attributable to HBV. The excess mortality for respiratory diseases could be attributable to a higher prevalence of smokers in the HBV carriers than in the HBsAg-negative subjects. Unfortunately, we cannot confirm this hypothesis because no such data was collected among HBsAg-negative men. However, if the hypothesis were confirmed, an excess of deaths from tobacco-related tumours and ischemic heart disease would be expected, which was not observed in the cohort. The observed increased risk of death from AIDS could be attributed to the fact that HBV and HIV share similar transmission routes, and thus, viral coinfection would not be rare. On the other hand, the excess of mortality in haematological neoplasms detected during the same period could be explained by the viral coinfection. Unfortunately, HIV seromarkers were not available in the cohort because of data confidentiality aspects. Therefore it has not been possible to link our cohort with the local AIDS registry. Our results are similar to those observed in the study in the UK, which detected an excess of AIDS and Non-Hodgkin's lymphoma (NHL) mortality.19 There are 2 case-control studies59, 60 that have detected an excess of NHL among HBV carriers. We could not estimate the specific NHL risk because specific data for haematological neoplasms subtype was not available in the RMC.
In conclusion, our study has confirmed that, in Spain, HBV chronic infection is related to HCC among men and to LC among both sexes. In the period 1972–2000, annual mortality rate for HCC in HBV-infected subjects was 34.1 per 100,000 person-years among men and 7.6 among women. For cirrhosis, the annual mortality rates were 86.8 and 22.9, respectively. Alcohol, HCV infection, the presence of seromarkers for HBV replication, diabetes and probably HIV and HDV infections increase the risk of death from liver disease in HBsAg-positive men. The increased risk of haematological neoplasms is likely associated to HIV coinfection. These results suggest that is necessary to ensure an adequate screening and follow-up in HBV carriers at higher risk of developing liver disease.
The authors acknowledge with gratitude the support of Dr. Josep Lluis Larriba of the Universitat Politècnica de Catalunya in the record linkage and of Dr. Michelle A. Mendez and Laura Pareja for their comments on this manuscript.
- 1Globocan 2000: cancer incidence, mortality and prevalence worldwide. IARC CancerBase No. 5, Version 1. Lyon: IARC, 2001., , , .
- 12A prospective cohort study of hepatitis B surface antigen carriers in a working population. Gann 1984; 75: 571–3., , , , .
- 16The risk of development of hepatocellular carcinoma in hepatitis B virus carrier in New York. A preliminary estimate using death-records matching. Hepatology 1982; 2: 15–20., .
- 21Validation of the death certificate using a telephone survey. Gac Sanit 1987; 2: 58–62., , , , , .
- 23International classification of diseases, injuries, and causes of death, 9th revision. Clinical modification. US Department of Health and Human Services, Washington, DC, 1979.
- 24Statistical methods in cancer research. Vol. 2: The design and analysis of cohort studies (IARC Scientific Publications No. 32). Lyon: IARC, 1987., .
- 25Epidemiologic analysis with a programmable calculator. Boston: Epidemiology Resources Inc., 1982., .
- 26Modern epidemiology, 2nd ed. Philadelphia: Lippincott, 1998., .