It is well known in infectious disease epidemiology that, in the absence of widespread vaccination and use of day care centers, first-born, and to a lesser extent, second-born children are usually exposed to common infections after they join primary school, whereas later-born children are frequently exposed much earlier, through their older siblings. 1–3 In recent years, birth order has been used as a proxy for the timing of exposure to infection in diseases with known or suspected infectious agents 4,5 and for diseases with suspected infectious origin without a focus on a specific agent or agents. 6,7
With respect to hepatitis B virus (HBV)-related hepatocellular carcinoma (HCC), it has been shown that early rather than later exposure to HBV is more frequently associated with chronic infection to this virus. 8 Over and beyond this fact, at least two studies have indicated that, among carriers, later-born children are at higher risk for HBV-related liver cancer than earlier-born ones. 9,10 On the contrary, later infection with hepatitis C virus (HCV) has been reported to increase the risk for HCV-related HCC. 11
Because the existing relevant information is limited, we have evaluated the role of birth order in the development of HBV-related HCC and HCV-related HCC in a large case-control study in Greece. In addition to the scientific implications, the issue is also of practical importance. Reduction of perinatal transmission and early vaccination for HBV could have a disproportionate effect on future incidence of HBV-related HCC, because these processes would reduce not simply the number of carriers but also the likelihood that the average carrier would eventually develop HCC.
Subjects and Methods
During a 4-year period between January 1995 and December 1998, 374 incident cases of HCC were admitted to three teaching hospitals in Athens (Hippokration, Western Attica, and Laiko General Hospital), of whom 41 (11%) were not enrolled in the study for various reasons. All 333 cases were diagnosed with HCC by a physician. Further confirmation of their HCC diagnosis was based on biopsy (N = 157), elevated alpha-fetoprotein level (N = 159), or echomography and/or other methods (N = 14); for three cases, details concerning diagnostic confirmation were missing. Details of the study have been given elsewhere. 12,13
Two groups of subjects were included as controls in this study. Subjects with metastatic liver cancer (MLC) were selected as patients manifesting symptomatology and clinical severity similar to those of the HCC cases, but with a primary cancer site unrelated to HBV or HCV chronic infection. In addition, patients admitted to the same hospitals for injuries or for eye, ear, nose, or throat conditions [other hospital controls (OHCs)] were included in the control series. We attempted to match each case by gender and age (±5 years) with one OHC and one MLC control. Among the controls, 31 MLC cases and 25 OHCs refused to participate in the study, and a properly matching control in one or the other control groups could not be identified for some HCC cases. Eventually, our control series included 632 subjects, comprising 272 MLC cases and 360 OHCs. The two control series were largely comparable for exposures other than those controlled for and were thus pooled for the purpose of this analysis.
Sera that had been obtained from each subject and stored at –25°C for 1–4 years were shipped in dry ice for serologic testing at the internationally certified Biomedicine Laboratories in Athens, Greece. Coded samples were tested for hepatitis B surface antigen (HBsAg), the marker for HBV infection, using the Auszyme monoclonal enzyme immunoassay (EIA) kit (Abbott, Chicago, IL). The presence of antibodies to HCV (anti-HCV) was determined with the Abbott HCV EIA 3.0. The anti-HCV assay detects antibodies to the HC-34 (core), HC-31 (NS3 and NS4), c100-3 (NS3 and NS4), and NS5 antigens. Both the HBsAg and anti-HCV assays are considered third-generation tests.
All HCC patients and controls were interviewed in person in the hospital using precoded questionnaires. Data concerning demographic, socioeconomic, and medical characteristics were recorded, and detailed histories of alcohol consumption and tobacco smoking were taken. Those who were currently smokers or had stopped smoking within the last 3 years were considered current smokers. Sibship size and birth order were based on the total number of liveborn children in the family. Because the study focused on postnatal environmental exposures, twins were considered as independent births.
We calculated odds ratios (ORs) for the occurrence of HCC associated with sibship size and birth order using stratified analysis and unconditional logistic regression modeling. Because individual matching was impossible in some HCC cases, we used unconditional logistic regression modeling, adjusting for the matching factors of age (in 10-year groups, categorically) and gender. We assessed the effect of sibship size after controlling for birth order (as an ordinal variable), and, likewise, we assessed the effect of birth order after controlling for sibship size (again, as an ordinal variable). Separate models were applied for HBsAg-positive cases compared with HBsAg-positive controls, anti-HCV-positive cases compared with anti-HCV-positive controls, and finally virus-negative cases in comparison with virus-negative control subjects. Eleven HCC cases and two controls were positive for both HBsAg and anti-HCV and were included in both respective models, with mutual adjustment for viral status as appropriate. Potential confounders that we controlled for were tobacco smoking (ordinal, as follows: never, <2 packs per day, and ≥2 packs per day), alcohol consumption (ordinal, as follows: never, <20, 20–39, and ≥40 glasses per week), and years of schooling (<12 vs ≥12 years). Complete data with respect to study variables and potential confounders were available for 331 of 333 (99%) cases and for 608 of 632 (97%) controls.
Cases and controls were similarly distributed by age and gender, because these were matching factors, and they also had similar distributions with respect to years of schooling (Table 1). On the contrary, chronic infection with HBV and/or HCV, tobacco smoking, and consumption of alcoholic beverages were all more frequent among HCC cases than among the control series. 12,13 The distributions of all of the above variables were similar between the two control groups, but there was a small overrepresentation of females and smokers in the MLC group. All analyses controlled for these variables through statistical modeling.
Table 2 shows the distribution by sibship size and birth order of HCC cases and controls, classified by evidence of chronic infection with HBV, HCV, or neither virus. With respect to sibship size, there is no evidence that it was associated with HCC in categories of HCV+ or virus-negative, although HBsAg-positive cases are generally of greater sibship size than HBsAg-positive controls. With respect to birth order, however, there is evidence for a positive association with HBsAg-positive HCC cases vs HBsAg-positive controls. These results are descriptively valid, but they are not directly interpretable, because sibship size and birth order are strongly positively associated (r among controls = +0.5) and therefore confound each other.
Table 3 shows model-derived OR estimates of sibship size and birth order for HCC positive for HBsAg, positive for anti-HCV, or negative for both viruses vs corresponding controls of identical viral status. All estimates are adjusted for age, gender, years of schooling, tobacco smoking, and alcohol consumption, as well as for HBsAg and anti-HCV, where appropriate. Sibship size was evaluated in at least four categories, but birth order was evaluated in three because of the a priori evidence that the first two children born are infected later than are later-born ones. After mutual adjustment for sibship size and birth order, there is little evidence of an association of sibship size with any etiological subgroup of HCC. In contrast, there is evidence (OR per increase in birth order = 2.0; 95% confidence interval = 1.2–3.6) that a higher birth order increases the risk for HBsAg-positive HCC.
The results of this study indicate that, given HBsAg carrier state, those who become carriers at a younger age because they were born later are more likely to develop HBsAg-positive HCC. This association is in addition to the tendency of later-born children to become HBsAg carriers once infected by HBV. 8 A possible explanation for this finding invokes the multistage theory of carcinogenesis, whereby early infection in life and consequent longer duration of exposure facilitates the accumulation of mutations in hepatocytes. The results of our study also indicate that there is no evidence for a similar birth order effect among anti-HCV-positive subjects. This finding is compatible with other reports 11 and our current understanding of the pathogenesis of HCC, in which chronic infection with HCV is considered a late-stage growth-enhancing process. 8 The results of our study also document that there is no strong sibship size effect on HBsAg-positive HCC, as the apparent positive association in Table 2 disappears after controlling for the confounding effects of socioeconomic status and birth order.
This study has several characteristics that support the validity of the principal result, that is, the positive association between birth order and risk for HCC among HBsAg positive individuals. The study is large, and the findings with respect to chronic infection with HBV, HCV, tobacco smoking, and alcohol intake are largely compatible with the collective evidence from previous studies. 12,13 No formal validation of exposure data was undertaken, but sibship size and birth order are generally accurately reported, as are data concerning demographic variables, tobacco smoking, and alcohol intake. Moreover, HBsAg and anti-HCV were determined with third-generation procedures. It should also be pointed out that in our analysis we have taken into account all recognized confounders and also that placental transmission has been found to be unrelated to children’s being infected with HBV in Greece. 14 Last, the documentation of birth order association through proper case-control investigation includes none of the built-in biases of earlier methods considered in the paper by Hsieh et al. 9
Two previous studies have evaluated birth order in relation to HBV-positive HCC, one in The Gambia 10 and one, independent of the present study, in Greece. 9 The study in The Gambia evaluated the association between birth order and liver cancer among all, rather than among HBsAg-positive, subjects and did not control for sibship size. Both drawbacks are minor, however, because most cases of HCC in The Gambia are HBV-related, and sibship size appears to play a minor role, as shown in the present study. The study by Hsieh et al 9 has a methodologic design similar to the present one, but it was smaller, and anti-HCV was not measured with a third-generation assay. Using a different approach, Evans et al 15 have also shown that longer duration of HBV carrier state increases the risk of HCC. These studies taken together strongly indicate that infection in early life substantially increases the risk of HBV-related HCC in later years. Our study also indicates that this association does not apply to infection with HCV and subsequent development of HCV-related HCC.
1. Russel WT. The Epidemiology of Diphtheria during the Last Forty Years: Medical Research Council Special Report Series No 247. London: His Majesty’s Stationery Office, 1943.
2. Taylor I, Knowelden J. Principles of Epidemiology, 2nd ed. Boston: Little, Brown, 1964; 304–307.
3. Fox JP, Hall CE, Elveback LR. Epidemiology: Man and Disease. New York: MacMillan Publishing Company, Inc., 1970; 199–201.
4. Blaser MJ, Chyou PH, Nomura A. Age at establishment of Helicobacter pylori
infection and gastric carcinoma, gastric ulcer, and duodenal ulcer risk. Cancer Res 1995; 55: 562–565.
5. Zilber N, Kahana E. Environmental risk factors for subacute sclerosing panencephalitis (SSPE). Acta Neurol Scand 1998; 98: 49–54.
6. Prener A, Hsieh CC, Engholm G, Trichopoulos D, Jensen OM. Birth order and risk of testicular cancer. Cancer Causes Control 1992; 3: 265–272.
7. Westergaard T, Melbye M, Pedersen JB, Frisch M, Olsen JH, Andersen PK. Birth order, sibship size and risk of Hodgkin’s disease in children and young adults: a population-based study of 31 million person-years. Int J Cancer 1997; 72: 977–981.
8. Okuda K, Tabor E, eds. Liver Cancer, 1st ed. New York: Churchill Livingstone, 1997.
9. Hsieh CC, Tzonou A, Zavitsanos X, Kaklamani E, Lan SJ, Trichopoulos D. Age at first establishment of chronic hepatitis B virus infection and hepatocellular carcinoma risk: a birth order study. Am J Epidemiol 1992; 136: 1115–1121.
10. Ryder RW, Whittle HC, Sanneh AB, Ajdukiewicz AB, Tulloch S, Yvonnet B. Persistent hepatitis B virus infection and hepatoma in The Gambia, west Africa: a case-control study of 140 adults and their 603 family contacts. Am J Epidemiol 1992; 136: 1122–1131.
11. Tradati F, Colombo M, Mannucci PM, Rumi MG, De Fazio C, Gamba G, Ciavarella N, Rocino A, Morfini M, Scaraggi A, Taioli E. A prospective multicenter study of hepatocellular carcinoma in Italian hemophiliacs with chronic hepatitis C. The Study Group of the Association of Italian Hemophilia Centers. Blood 1998; 91: 1173–1177.
12. Kuper HE, Tzonou A, Kaklamani E, Hadziyannis S, Tasopoulos N, Lagiou P, Trichopoulos D, Stuver S. Hepatitis B and C viruses in the etiology of hepatocellular carcinoma: a study in Greece using third-generation assays. Cancer Causes Control 2000; 11: 171–175.
13. Kuper H, Tzonou A, Kaklamani E, Hsieh C-C, Lagiou P, Adami H-O, Trichopoulos D, Stuver SO. Tobacco smoking, alcohol consumption and their interaction in the causation of hepatocellular carcinoma. Int J Cancer 2000; 85: 498–502.
14. Papaevangelou G, Hoofnagle J, Kremastinou J. Transplacental transmission of hepatitis-B virus by symptom-free chronic carrier mothers. Lancet 1974; 2: 746–748.
15. Evans AA, O’Connell AP, Pugh JC, Mason WS, Shen FM, Chen GC, Lin WY, Dia A, M’Boup S, Drame B, London WT. Geographic variation in viral load among hepatitis B carriers with differing risks of hepatocellular carcinoma. Cancer Epidemiol Biomarkers Prev 1998; 7: 559–565.